Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Longitudinal assessment of neural stem-cell aging
(USC Thesis Other)
Longitudinal assessment of neural stem-cell aging
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
LONGITUDINAL ASSESSMENT OF NEURAL STEM CELL AGING
by Albina Ibrayeva
A dissertation is presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHYLOSOPHY
BIOLOGY OF AGING
May 2022
Copyright 2022 Albina Ibrayeva
i
Acknowledgements
I would like to give a special thank you to my mentor, Dr. Michael Bonaguidi, for his
patience and excellent guidance throughout my graduate studies, endless support, and scientific
advice. Thank you so much for supporting me during this journey and shape me into a scientist
and researcher who I aim to be. I also want to express my gratitude for all current and past lab
members of Bonaguidi’s lab especially Maxwell Bay, Lei Peng, Elbert Pu, Dr. Ismael Fernandez-
Hernandez, Dr. Aswathy Ammothum Kandy, and undergraduate researchers: Congrui Lin,
Avinash Iyer, Galen Resler, Emily Sun, Lee McMahon, Jorge Franco Campos and Axel Hidalgo.
I would like to thank my committee members Professors Caleb Finch, Bérénice Benayoun,
Judith Campisi and Giorgia Quadrato for being a part of my committee, guidance, and scientific
advice over the years.
I would like to thank my collaborators Professors Benjamin Simons (University of
Cambridge) and David Cobrinik (USC/Children’s Hospital of Los Angeles) for your support and
advice over the years.
Finally, I want to thank my family and friends for constant love and support. Thank you
for always being there for me.
ii
Table of Contents
Acknowledgements .......................................................................................................................... i
List of Figures ................................................................................................................................ iv
List of Tables ................................................................................................................................. vi
Abstract ......................................................................................................................................... vii
Chapter I: Introduction .................................................................................................................... 1
Overview of the Neural Stem Cell Field in the context of aging .............................................. 1
Regulation and function of adult neurogenesis ......................................................................... 3
The role of adult neurogenesis in disease conditions and aging ................................................ 6
Emerging technologies for Adult Neurogenesis Research ........................................................ 9
New approaches towards translation ....................................................................................... 10
Chapter II: Early Stem Cell Aging in the Mature Brain ............................................................... 12
Abstract .................................................................................................................................... 12
Introduction ............................................................................................................................. 13
Results ..................................................................................................................................... 14
Asynchronous NSC decline during aging .......................................................................... 14
Increased quiescence drives NSC loss of homeostasis ...................................................... 17
Quiescent NSC molecular aging in the mature hippocampus ............................................ 20
Imatinib partially restores NSC function in the middle-aged brain ................................... 24
Discussion ................................................................................................................................ 26
Materials and Methods ............................................................................................................ 29
Animals and Tamoxifen Administration ............................................................................ 29
Immunostaining, Confocal Imaging, and Processing ......................................................... 30
Clonal Analysis .................................................................................................................. 31
Stereotaxic surgery (Imatinib infusion via osmotic pump) ................................................ 32
Fluorescence- activated Cell Sorting (FACS) of Individual cells from Adult Mouse
Dentate Gyrus ..................................................................................................................... 32
scRNA-seq library preparation and sequencing ................................................................. 33
Computational analysis of RNA-Seq data ......................................................................... 34
Computational modeling of clonal data ............................................................................. 37
Statistical analysis .............................................................................................................. 44
Chapter III: Different States and Discrete Populations of Adult Neural Stem Cells .................... 45
Abstract .................................................................................................................................... 45
Introduction ............................................................................................................................. 46
Results ..................................................................................................................................... 48
Labeling RGLs in different activation states ...................................................................... 48
Divergence in Cell Cycle Exit after Initial RGL Activation .............................................. 51
iii
Fate Choice Bias of Different RGLs .................................................................................. 52
Maintenance of Distinct Stem Cell Properties ................................................................... 53
Computational Assessment of Different States and Discrete Stem Cell Populations ........ 56
Plasticity of RGLs upon Injury .......................................................................................... 58
Convergence of discrete populations during physiological aging ..................................... 61
Discussion ................................................................................................................................ 62
Transient Stem Cell States and Discrete Stem Cell Populations ....................................... 63
Potential Neural Stem Cell Relationships in the Adult Hippocampus ............................... 64
Experimental Procedures ......................................................................................................... 67
Animals, Tamoxifen Administration and AraC Treatment ................................................ 67
Immunostaining, Confocal Imaging, and Image Processing .............................................. 69
Clonal Analysis .................................................................................................................. 71
Computational analysis ...................................................................................................... 73
Chapter IV: Developmental Origins of Adult Neural Stem Cells ................................................ 80
Abstract .................................................................................................................................... 80
Introduction ............................................................................................................................. 81
Results ..................................................................................................................................... 83
Clonal Analysis of Developmental Hippocampus using Gli1-CreER
T2
system ................ 83
Existence of NSC parallel lineages during development ................................................... 85
NSC clonal evolution during hippocampal tissue remodeling ........................................... 89
Computational modeling of NSC lineages indicates their lineage independency ............. 92
Discussion ................................................................................................................................ 94
Materials and Methods ............................................................................................................ 97
Animals and Tamoxifen Administration ............................................................................ 97
Immunostaining, Confocal Imaging, and Processing ......................................................... 98
Clonal Analysis .................................................................................................................. 99
Computational modeling .................................................................................................. 100
Quantification and statistical analysis .............................................................................. 100
Chapter V: Conclusion and Final Thoughts ................................................................................ 102
Supplemental Information .......................................................................................................... 106
References ................................................................................................................................... 129
iv
List of Figures
Figure 1.1. The Hallmarks of Aging. .............................................................................................. 1
Figure 1.2. A sagittal view of the adult mouse brain. ..................................................................... 4
Figure 1.3. Hippocampal neurogenesis developmental stages ....................................................... 5
Figure 2.1. Asynchronous neural stem cell (NSC) decline during aging. .................................... 15
Figure 2.2. Increased quiescence drives NSC loss of homeostasis. .............................................. 18
Figure 2.3.Quiescent NSC molecular aging in the mature hippocampus. .................................... 22
Figure 2.4. Imatinib partially restores NSC function in the middle-aged brain. .......................... 25
Figure 3.1. Clonally labeled RGLs in the adult dentate gyrus using two CreER
T2
lines exhibit
different cell cycle properties. ....................................................................................................... 50
Figure 3.2. Fate specification of Gli1
#
-RGLs and Ascl1
#
-RGLs ................................................. 53
Figure 3.3. Maintenance of stem cell properties by multipotent Gli1
#
-RGLs and neuronal fate
biased Ascl1
#
-RGLs. ..................................................................................................................... 55
Figure 3.4. Different states and distinct populations of stem cells revealed by a computational
approach. ....................................................................................................................................... 57
Figure 3.5. Plasticity of RGL fate after AraC-induced injury ...................................................... 59
Figure 3.6. Convergence of distinct properties during physiological aging. ................................ 62
Figure 4.1. Clonal Analysis of Developmental Hippocampus using GliCreER
T2
system. ........... 84
Figure 4.2. Establishment of parallel stem cell lineages in development ..................................... 87
Figure 4.3. Clonal evolution during hippocampal tissue remodeling ........................................... 91
Figure 4.4. Evolution of the NSC distribution in model of two populations of NSC. .................. 94
Figure S2.1. Population analysis and clonal lineage-tracing of individual NSCs in the adult
mouse dentate gyrus. ................................................................................................................... 106
v
Figure S2.2. Computational modeling of Nestin
#
-NSC and Ascl1
#
-NSC. ................................. 108
Figure S2.3. Experimental design of scRNA-seq experiment. ................................................... 109
Figure S2.4. Differential gene expression analysis. .................................................................... 110
Figure S3.1. Multiple strategies for targeting and clonal lineage-tracing of RGLs in the adult
mouse dentate gyrus .................................................................................................................... 117
Figure S3.2. Lineage trees and reporter comparisons from short-term analyses of Gli1
#
-RGL and
Ascl1
#
-RGL clones ..................................................................................................................... 118
Figure S3.3. Long-term analyses of Gli1
#
-RGL and Ascl1
#
-RGL clones at 30 and 60 dpi ....... 119
Figure S3.4. Computational assessment of RGL properties ....................................................... 120
Figure S3.5. Validation of computational models ...................................................................... 121
Figure S3.6. Changes of Gli1
#
-RGL and Ascl1
#
-RGL behavior upon AraC-induced injury ..... 122
Figure S3.7. Nestin
#
-RGL behavior after AraC treatment .......................................................... 123
Figure S4.1. Example of confocal images of the clones depicted at various timepoints ............ 125
Figure S4.2. Clonal lineage-tracing of individual NSCs in the mouse dentate gyrus during tissue
remodeling. ................................................................................................................................. 126
Figure S4.3. Clonal lineage-tracing of individual NSCs in the mouse dentate gyrus during
postnatal development ................................................................................................................ 127
vi
List of Tables
Table S2.1. Tamoxifen doses used to achieve clonal recombination among various promoter,
reporter and ages contexts over the analyzed time course. ......................................................... 111
Table S2.2. Number of clones for clonal analysis. Summary of number of all clones, NSC-
containing clones, animals across different ages and tracing time points for clonal analysis. ... 113
Table S2.3. Parameter list and best fit parameters for the theoretical mode. ............................. 115
Table S2.4. Primer sets that were used for genotyping. ............................................................. 116
Table S3.1. Tamoxifen doses used to achieve clonal recombination among various promoter,
reporter, and environmental contexts over the analyzed time course ......................................... 124
Table S4.1. Number of clones for clonal analysis. ..................................................................... 127
vii
Abstract
Brain plasticity underlies our ability to maintain cognitive ability and the capability to
efficiently interact with society. Such flexibility is well known to decline with age but determining
how to identify and prevent it at young ages remains unknown. It is now recognized that in adults
quiescent neural stem cells (NSCs) generate newborn neurons and astrocytes to modify existing
neural circuits. Despite this, neurogenesis and hippocampus function are markedly lower in older
animals. Unique among somatic tissues containing stem cells, the brain experiences a decline in
cell genesis early in young adult rodents and middle ages in humans. Moreover, numerous studies
indicate that a decline in adult stem cell function can drive aging-related diseases.
Aged animals have significantly less neural stem cell numbers, stem cell proliferation,
neuronal differentiation and newborn neuron survival compared to younger animals (Encinas et
al., 2011; Kuhn, Dickinson-Anson, & Gage, 1996; Ziebell, Dehler, Martin-Villalba, & Marciniak-
Czochra, 2018). Further, NSCs in old animals exhibit hallmarks of cellular aging including deficits
in proteostasis and receive high levels of inflammation (Kalamakis et al., 2019; Leeman et al.,
2018). Yet, the hippocampus experiences a loss of neurogenesis early in the mature brain of
rodents (Ben Abdallah, Slomianka, Vyssotski, & Lipp, 2010; Morgenstern, Lombardi, & Schinder,
2008) and by middle-age in humans (Knoth et al., 2010; Moreno-Jiménez et al., 2019; Spalding et
al., 2013). This decline is accompanied by epigenetic loss of DNA demethylation (Gontier et al.,
2018), suggesting NSCs could become dysregulated early during chronological aging.
Therefore, the central goals of my doctoral work were to identify cellular and molecular
mechanisms that underline cellular aging of adult NSC populations, as well as understand the
origins and heterogeneity of these NSC populations. I’ve applied new approaches in endogenous
single cell lineage tracing, computation modeling and single cell genomic profiling to address
viii
questions that have remained unanswered for more than twenty years: what primarily drives neural
cell genesis decline during physiological aging? What are the origins of different NSC
populations? How do they contribute to the whole physiological decline? What are the cellular and
molecular mechanisms that are responsible? Can targeted neural circuit remodeling restore stem
cell function to healthy, younger levels? A better understanding of neurogenesis during
development and aging could serve as a platform to develop novel therapeutic strategies against
the different age-related pathological and physiological neurological disorders.
Recent studies revealed the first demonstration of endogenous adult mammalian NSC
properties at the single cell level, which are now recognized as the new standard for the field of
stem cell research (Bonaguidi et al., 2011). During my doctoral studies I have developed various
lineage tracing strategies to track individual NSCs and their evolution within the mouse
hippocampus. I was able to establish single cell RNA sequencing (scRNA-seq) protocol for
capturing and deep sequencing rare subpopulations of adult neural stem cells.
In my doctoral work I present a new concept of stem cell aging where stem cell function declines
due to cellular and molecular changes that compromise their homeostasis. While aging is typically
examined among old organisms, biological aging occurs in a gradual and asynchronous manner
throughout the body (Schaum et al., 2020). By combining in vivo single cell clonal lineage tracing,
computational modeling approaches, scRNA-seq and systems level data science we
comprehensively investigated neural stem cell adaptation and restoration during development and
aging. I identified that NSCs undergo early aging by defining when, why and how NSCs lose
homeostasis. I showed that targeting mechanisms associated with the initial loss of NSC
homeostasis can overcome age-related NSC dysfunction later in life. I used a clinically relevant
drug Imatinib (Abl1/2 inhibitor) as a strategy to overcome NSC cellular aging. Indeed, intracranial
ix
infusion into the middle-aged brain was sufficient to overcome deep NSC quiescence and restore
NSC proliferation to younger levels. My study elucidated cellular and molecular origins of
neurogenesis decline in the middle-age adult and may serve as a new mammalian stem cell model
to study cellular aging. These findings provide a broadly useful resource to prioritize individual
genes or complementary gene families to create new directions towards age-related regenerative
medicine throughout the body.
Lastly, I developed a transgenic labeling approach in mice driven by Gli1
#
regulatory
elements combined with stochastic multicolor Confetti reporter (CFP, YFP, RFP) to trace the entire
process of DG neurogenesis from a single cell from development through adulthood. For
performing in vivo single cell lineage tracing, I inject time-pregnant female Gli1-CreER::Confetti
mice with 14 mg/kg tamoxifen to label pups at E17.5 and sacrificed them at multiple postnatal
timepoints. I found that multipotent NSC clones emerge by the end of the hippocampal postnatal
ontogenesis. In addition, neurogenesis expands within NSC-containing clones into the young
adult. I also was able to establish the sequential order of neuron, oligodendrocyte and astrocyte
production in the adult mouse hippocampus. In doing so, I revealed developmental properties of
neural precursors and their transition into adult neural stem cells.
1
Chapter I: Introduction
Overview of the Neural Stem Cell Field in the context of aging
Stem Cell Biology continues to become an attractive field for understanding basic tissue
biology and for its therapeutic potential in different diseases and disorders (Chagastelles and Nardi,
2011). Two types of stem cells exist in the adult organism: (1) adult somatic stem cells, which are
crucial for tissue maintenance and regeneration; and (2) germline stem cells, which are found
specifically in the reproductive system (Li and Xie, 2005). Germline stem cells are responsible for
production germ cells: sperm and egg (Spradling et al., 2011). Meanwhile, somatic stem cells are
present in many adult tissues and are responsible for tissue generation throughout life. In order to
do so, they must balance the production of newborn cells with maintenance of their own cell
genesis. However, when this equilibrium is progressively lost in many tissues (brain, blood,
muscle, skin) during aging, degeneration in tissue integrity and function occur with a reduced
capacity for regeneration upon injury. Aging is one of the most evident biological processes whose
molecular and cellular mechanisms are still weakly understood (Figure 1.1).
Figure 1.1. The Hallmarks
of Aging.
Genomic instability, telomere
attrition, epigenetic
alterations, loss of
proteostasis, deregulated
nutrient sensing,
mitochondrial dysfunction,
cellular senescence, stem cell
exhaustion, and altered
intercellular communication
(Adapted from López-Otín et
al., 2013).
2
Visible signs of aging are obvious: skin wrinkles, hair grays or is lost, body mass decreases,
memory and cognitive flexibility decay. While aging has been traditionally associated with being
old, approaches to diagnose and prevent decreasing function at young ages have become
progressively prevalent (Belsky et al., 2015; Gillman, 2005). The ability to ensure healthy function
as we age depends on individual cells that act in a coordinated manner to promote tissue and body
homeostasis. Biological aging occurs in a gradual and heterogeneous manner where specific cells
and eventually organ systems progressively deteriorate. As a result, progress in slowing declining
function has been hampered by inability to determine what and when represent the most vulnerable
targets to treat (Kennedy et al., 2014). Restoring or delaying functional loss earlier in the aging
process are likely to enhance resiliency into old age.
The adult brain experiences progressive loss of function during physiological aging. Aging
also affects the progression of neurological diseases and disorders. At the cellular level, aging of
the brain is accompanied by many changes including cell death, impaired cellular metabolism,
oxidative stress and DNA damages (Limke et al., 2002; Mattson et al., 1999). While these changes
alone are not sufficient to cause neurodegeneration, it is widely accepted that cellular aging
contributes to the predisposition of neurons to neurodegenerative diseases, such as Alzheimer and
Parkinson diseases (Keller and Mattson, 1998). Further, it is becoming more apparent that age-
related changes in brain structure and function do not occur uniformly across the whole brain or
across individuals (Glisky, 2007; Trollor and Valenzuela, 2001). The most notable changes that
occurs in brain during aging are the alteration in memory and cognition. For example a longitudinal
study, using two MRI scans separated by around 1-2 years (Scahill et al., 2003), has provided a
functional role of the hippocampus in various cognitive abilities. Hippocampal structure and
volume is well-documented to change during normal aging (Raz et al., 2005). It’s been also shown
3
that hippocampus being a vulnerable region to age, support well with the learning and memory
changes during physiological aging (Anderton, 2002; Barnes, 2003). The decline in function is
worsen when it’s accompanied with the neurodegenerative disorders, thus therapies that improves
physiological aging is a primary goal in aging research (Limke et al., 2002).
Regulation and function of adult neurogenesis
For many decades, the scientific community believed that the adult mammalian brain is
incapable to remodel existing circuits. However, present literature indicates that the adult brain
possesses the remarkable ability to adapt through alteration of neural pathways based on new
experiences, learning new information, creating memories, and in response to injury (Ming and
Song, 2011). Pioneering discoveries on neurogenesis, the growth and development of newborn
cells from neural stem cells (NSCs), have made a substantial contribution to our thinking of neural
circuit remodeling. Altman and colleges first showed the robust evidences for neurogenesis in
adult mammalian brain by using rat model (Altman and Das, 1965). After this initial discovery,
multiple groups across the world proved the existence of adult neurogenesis using different animal
models (songbird, reptiles, fish etc.) and advanced techniques (electron microscopy, confocal
microscopy, immunohistochemistry etc.) (Goldman and Nottebohm, 1983; Kaplan and Hinds,
1977; Paton and Nottebohm, 1984). In mammals, physiological neurogenesis predominately
occurs in two specific brain regions, subventricular zone (SVZ) near the lateral ventricles and
subgranular zone (SGZ) in the dentate gyrus (DG) of the hippocampus (Figure 1.2).
4
Figure 1.2. A sagittal view of the adult mouse brain.
Focusing on two major niches where adult NSCs reside: the subventricular zone (SVZ) and the
subgranular zone (SGZ). The SVZ is located along the lateral ventricle in the forebrain, while the
SGZ is located in the hippocampus along the dentate granule cell layer where it abuts the hilus.
CC, corpus callosum; DG, dentate gyrus; Hipp, hippocampus; LV, lateral ventricle; NSC, neural
stem cell; OB, olfactory bulb; RMS, rostral migratory stream; SC, stem cell; St, striatum. (Adapted
from Bond et al., 2015)
The SVZ is located in the walls of lateral ventricles, along the epidermal cell layer (Figure
1.2). NSC from SVZ give rise to transient amplifying progenitor cells that divide a few times
before becoming a neuroblast. Neuroblasts migrate through the Rostral Migratory Pathway to
supply new born neurons to the Olfactory Bulb (Doetsch et al., 1999). Meanwhile, radial glia-like
NSCs in the SGZ give rise to intermediate progenitor cells (IPCs) (Seri et al., 2001), which
undergoing limited rounds of proliferation and become neuroblasts that migrate a short distance
into the granule cell layer of the dentate gyrus and incorporate into the existing circuitry of the
hippocampus. Whereas the SVZ-OB is well established in rodents, accumulating evidence
indicates a lack of actively dividing NSC in human SVZ (Bergmann et al., 2012; Curtis et al.,
2007; Sanai et al., 2004)
Neurogenesis in the DG of the hippocampus has been established across nearly all
mammals. Evidence exists both for and against human adult hippocampus neurogenesis. Human
neurogenesis was first identified using BrdU injection (Eriksson et al., 1998), but due to the low
sample size and difficulties to reproduce results, several studies put some doubt regarding the
5
overall presence of neurogenesis in adults (Seress, 2007; Seress et al., 2001; Sorrells et al., 2018).
However, recent technological advances have provided evidence for the persistence of
neurogenesis later in life (Spalding et al., 2013). In addition, Boldrini and colleges used autopsy
samples of healthy individuals ranging from 14 to 79 years old, and found a similar number of new
born neurons and volume of dentate gyrus across ages (Boldrini et al., 2018). Complementary
studies using various approaches will be needed to convince a skeptical scientific community of
human neurogenesis in the human brain (Kempermann et al., 2018).
Neurogenesis is highly susceptible to alterations under physiological (learning and memory,
social interaction, exercise) and pathological (epilepsy, mental disorders, brain injury,
neurodegeneration) conditions (Bond, Ming, & Song, 2015; Christian, Song, & Ming, 2014). The
adult hippocampus has functional importance as the critical mammalian neural structure
modulating cognitive deconstruction and reconstruction of places, items and events. The dentate
gyrus functions as the input and a bottleneck to hippocampal circuitry allowing for a small fraction
of new cells to have significant influence on hippocampal function (Aimone et al., 2014; Deng,
Aimone, & Gage, 2010). Therefore, new cells make an important, and dynamically controlled,
contribution to learning, forgetting and emotional processing (Dranovsky & Leonardo, 2012;
Kuhn, Toda, & Gage, 2018). In
adult dentate gyrus neurogenesis,
NSCs give rise to new neurons and
astrocytes (Figure 1.3) via a
sequential process of activation,
proliferation, and generation of
intermediate progenitor cells (Bonaguidi et al., 2011; Jang et al., 2013).
Figure 1.3. Hippocampal neurogenesis developmental
stages: Radial glia-like cell = NSC
6
While hippocampal neurogenesis persists throughout adulthood, this process is highly
dysregulated during chronological aging (Navarro Negredo, Yeo, & Brunet, 2020). Aged animals
have significantly less neural stem cell numbers, stem cell proliferation, neuronal differentiation
and newborn neuron survival compared to younger animals (Kuhn et al., 1996) (Encinas et al.,
2011; Kuhn et al., 1996; Ziebell et al., 2018). Further, NSCs in old animals exhibit hallmarks of
cellular aging including deficits in proteostasis and receive high levels of inflammation (Kalamakis
et al., 2019; Leeman et al., 2018). Yet, the hippocampus experiences a loss of neurogenesis early
in the mature brain of rodents (Ben Abdallah et al., 2010; Morgenstern et al., 2008) and by middle-
age in humans (Knoth et al., 2010; Moreno-Jiménez et al., 2019; Spalding et al., 2013). This
decline is accompanied by epigenetic loss of DNA demethylation (Gontier et al., 2018), suggesting
NSCs could become dysregulated early during chronological aging. However, after 20 years of
investigation, the cellular and molecular mechanisms driving the initial loss of age-related
neurogenesis remain unidentified.
The role of adult neurogenesis in disease conditions and aging
Recent technological advances in the United States have been accompanied by the rapid
growth of population over 65 (Speaks, 2016). Currently, society is facing new challenges in the
public and private sectors in both acting to the needs and applying the resources of older people.
Significant improvements in longevity of the population have led to rising the more incidence of
chronic and age-associated diseases such as osteoporosis, heart failure, dementia, Alzheimer
diseases and others.
Deterioration in mental health due to chronic neurodegenerative diseases represents the
largest cause of disability around the world. Over 30% of population live with some types of brain
disorders. According to the World Health Organization (WHO), one of the leadership priorities is
7
“Addressing the challenge of non-communicable diseases” (WHO, 2015). The challenge of an
aging society has been commonly recognized and a wide variety of research and development
programs around the world have been introduced to tackle age-related diseases. But today, instead
of aiming single age-linked disorders, the mechanisms of the aging process itself are being
investigated with the goal of finding ways for intervention and prevention.
Hippocampal neurogenesis plays a role in rodent models of disease model such as epilepsy
and Alzheimer’s disease. It’s been shown that during the development of epilepsy, generation of
new born neurons is intensely interrupted as well as numerous cellular abnormalities can be
observed in CA1 and CA3 parts of the hippocampus (Hattiangady and Shetty, 2008). Neurogenesis
increases during the several days after an epileptic seizures, but decrease happens during chronic
phases (Hattiangady and Shetty, 2008; Sierra et al., 2015). Disrupted neurogenesis produces both
loss- and gain-of-functions defects (Christian et al., 2014). Loss-of-functions are resulted from
reduced neurogenesis during chronic epilepsy and contribute to the cognitive impairments and
memory loss (Eisch and Petrik, 2012). On the other hand, gain-of-function mediated by newborn
neuron production during and after the epileptogenic insults. New neurons trigger an aberrant
rewiring of the hippocampus and promote epileptogenesis (Cho et al., 2015). This abnormal cell
replacement differs from normal general neuronal circuitry maintenance caused by adding new
neurons and astrocytes with unique functional properties and structural plasticity of mature
neurons induced by new born neuron interaction (Christian et al., 2014).
Adult neurogenesis maintains a range of cognitive and memory functions, many of which
decline with aging (Danzer, 2016). There is a good amount of literature exploring the effect of
aging on neurogenesis. Aging is consistent with substantial reduction in neuronal and overall cell
proliferation, survival and maintenance. Previous studies have shown that BrdU (labels dividing
8
cells), in aged (21mo) rats was significantly reduced compared to the middle aged (6mo) (Kuhn et
al., 1996). The authors concluded that the overall number of newly dividing cells is dramatically
decreased by roughly nine-fold from middle age to aged rats, indicative of diminished
neurogenesis associated with aging. But it’s still not clear which population is a main contributor
of the age-related decline in hippocampal neurogenesis. One study suggested that Sox2+ (NSC
marker) population remains constant, but the proliferation rate is decreased, suggesting that NSC
activity might be a contributor to the reduced neurogenesis with age (Hattiangady and Shetty,
2008). In contrast, Hes5+ (bHLH gene NSC marker) cells in aged mice showed the overall drop
in total number of Hes5+ cells, indicating the significant decrease in NSC population with age
(Lugert et al., 2012). Even though those two studies showed conflicting evidence, clearly there is
a change in the entire NSC population with age.
Inspiration on the effects of aging on the brain can be drawn from other somatic stem cells
than in the brain. For example, how DNA damage, telomerase exhaustion, mitochondrial
interventions, limited proliferative capacity, extracellular signaling and epigenetic alteration
disturb the activity of stem cells and their progeny are not well established in adult hippocampus
neurogenesis. Such dysfunctions may be related to mechanisms that control organismal lifespan
and healthspan. However, the dynamic interplay between cell-intrinsic, environmental, and
systemic signals that may drive a decline in neurogenesis during aging is also not well
defined. Furthermore, the essential mechanisms of age-related neurological diseases generally
remain unclear in rodents and humans. A detailed model of cellular and molecular
mechanisms that ultimately lead to a cell death, senescence or loss of regenerative function will
be indispensable in facilitating the development of therapeutic interventions that can slow, and
possibly reverse, age-related neural disorders.
9
Emerging technologies for Adult Neurogenesis Research
For the past couple of decades, changing technologies have provided new insight into NSC
behavior and could be applied to aging such as in vitro cellular reprogramming, in vivo clonal
lineage tracing techniques, single cell RNA sequencing etc. In vitro assays provided controlled
conditions for isolation and expansion of neural precursor cells from mouse adult hippocampus
(Babu et al., 2011). Scientific discoveries of in vivo cellular reprogramming expanded the range
of neural precursor cell properties. Mammalian cells can be reprogrammed by using intrinsic and
extrinsic factors to gain new features beyond their physiological potential (Arlotta and Berninger,
2014). For example, resident astrocytes were reprogrammed by a single transcriptional factor -
Sox2 - into proliferative neuroblasts in the adult mouse brain (Niu et al., 2013). Another study
showed that striatal astrocytes can generate neurons in a stroke model by downregulating Notch
signaling pathway (Magnusson et al., 2014). Recently, advances using single cell approaches have
been able to tackle the question of heterogeneity within the NSCs population. For instance, in vivo
single cell clonal linage tracing techniques were able to reveal properties of NSCs and partially
address the question of heterogeneity (Bonaguidi et al., 2011; Calzolari et al., 2015). It’s being
anticipated that clonal analysis uses different genetic drivers to targets NSC subpopulations will
reveal some challenging questions regarding NSC behavior during development and aging.
Further, one of the main technical advancements in understanding molecular mechanisms
is single cell RNA sequencing (scRNA-seq). Yet, scRNA-seq has not yet been widely adopted for
in vivo adult somatic stem cell studies due to technical difficulties in obtaining individual stem
cells from complex tissues. In addition, the stochastic nature of gene expression in individual cells
lead to the overestimation of cellular heterogeneity. However, it became possible to assess with
SMART Seq technology, in both niches (SVZ and SGZ) adult NSCs showed molecularly
10
dynamics of NSC activation and neurogenesis, involving changes in gene expression, energy
sources and metabolism (Llorens-Bobadilla et al., 2015; Shin et al., 2015). Computational
approaches provide a relationship between individual transcriptomes for statistical quantification
(Raj et al., 2006). Unsupervised algorithms can increase the temporal resolution of transcriptome
dynamics using single-cell RNA-seq data collected at multiple time points (Juliá et al., 2015).
Further, the reconstruction of continuous developmental trajectories is now feasible to elucidate
the highly dynamic nature of stem cell molecular signatures in the brain (Shin et al., 2015).
However, such approaches have not yet been taken to uncover the molecular mechanisms
associated with NSC development and aging.
New approaches towards translation
The research community and pharmaceutical/biotech companies have begun taking the
first steps towards identifying biomarkers targeting broad anti-aging benefits. Biology of Aging
referred as a “geroscience” has brought scientific attention to how biological interventions can
improve lifespan and healthspan (Kennedy et al., 2014). Those effects could be mimicked by
drugs, with several antiaging compounds (rapamycin, resveratrol) showed to have encouraging
impact on several age-related phenotypes (Berman et al., 2017; Li et al., 2014). However, such
drugs have a great spectrum of side effects that negatively affects other part of the body. Therefore,
there is a need for understanding basic biology underlying alterations that accompany aging, as a
distinct form basic biology of a disease. Moreover, synergetic effort of basic science, technological
development and pharmaceutical approaches could accelerate preventative strategies for aging
across multiple tissues towards more effectively prolonging healthspan.
Existing literature support the idea that increased tissue regeneration potential by adult
stem cells may be able to delay signs of aging, whereas a decline in the stem cell function
11
contributes to the age-related phenotype (Sahin and DePinho, 2010). It’s now widely accepted that
adult neural stem cells play a role in modifying existing brain circuitry (Bond et al., 2015). This
ability of the brain to adapt and change is not only relevant during the developmental period when
brain structure and thus function is evolved, but also in cases of injury and disease. Neurogenesis
is highly susceptible to aging, neurodegeneration and injury causes the extensive decrease in the
physical and cognitive performance of the whole organism. Compare to the other parts of the body,
the brain experience the loss in cell genesis as early as young adult in rodents (Ben Abdallah et al.,
2010) and by middle-aged in humans (Spalding et al., 2013). Even though, our current ability to
restore neural remodeling and recover the functions remains limited. There is still hope that by
establishing a foundation for understanding how the brain naturally fights against aging and guide
the development of neurogenesis-mediated therapeutic strategies.
12
Chapter II: Early Stem Cell Aging in the Mature Brain
The following chapter is adapted from (Ibrayeva et al., 2021)
Abstract
Stem cell dysfunction drives many age-related disorders. Identifying mechanisms that
initially compromise stem cell behavior represent early targets to promote tissue function later in
life. Here, we pinpoint multiple factors that disrupt neural stem cell (NSC) behavior in the adult
hippocampus. Clonal tracing showed that NSCs exhibit asynchronous depletion by identifying
short-term (ST-NSC) and long-term NSCs (LT-NSCs). ST-NSC divide rapidly to generate neurons
and deplete in the young brain. Meanwhile, multipotent LT-NSCs are maintained for months, but
are pushed out of homeostasis by lengthening quiescence. Single cell transcriptome analysis of
deep NSC quiescence revealed several hallmarks of molecular aging in the mature brain and
identified tyrosine-protein kinase Abl1 as an NSC pro-aging factor. Treatment with the Abl-
inhibitor Imatinib increased NSC activation without impairing NSC maintenance in the middle-
aged brain. Our study indicates that hippocampal NSCs are particularly vulnerable and adaptable
to cellular aging.
13
Introduction
Aging is the progressive loss of physiological function due to accumulating cellular
damage (Kennedy et al., 2014; López-Otín et al., 2013). This process occurs at a gradual and
asynchronous rate where specific cells and then organ systems lose homeostasis (Almanzar et al.,
2020; Negredo et al., 2020; Schaum et al., 2020). Aging has been historically investigated in
chronological terms, with the majority of studies comparing young and old organisms. Yet,
approaches to prevent aging at younger ages have recently become more common (Belsky et al.,
2015; Gillman, 2005). However, progress in slowing or reversing a decline in tissue function has
been hampered by an inability to determine when and which cells begin to exhibit cellular aging
(López-Otín et al., 2013; Solanas et al., 2017).
Adult neurogenesis persists throughout life in the subgranular zone in the dentate gyrus of
the hippocampus and is highly compromised during chronological aging (Kirschen et al., 2019;
Kuhn et al., 2018;, 2019; Toda and Gage, 2018). Aged animals have significantly less neural stem
cell numbers, stem cell proliferation, neuronal differentiation and newborn neuron survival
compared to younger animals (Encinas et al., 2011; Heine et al., 2004; Kuhn et al., 1996; Ziebell
et al., 2018). Further, NSCs in old animals exhibit hallmarks of molecular aging including
alterations to proteostasis and inflammation (Kalamakis et al., 2019; Leeman et al., 2018; Negredo
et al., 2020). Yet, the hippocampus experiences a loss of neurogenesis early in the mature brain of
rodents (Ben Abdallah et al., 2010; Morgenstern et al., 2008) and by middle-age in humans (Knoth
et al., 2010; Moreno-Jiménez et al., 2019; Spalding et al., 2013). This decline is accompanied by
epigenetic loss of DNA demethylation (Gontier et al., 2018), suggesting NSCs could molecularly
age at early stages of chronological aging.
14
However, cellular origins driving the early neurogenesis decline remain unclear. In one
scenario, NSCs have been found to drop in number in the young hippocampus (Gontier et al.,
2018; Lugert et al., 2012). This is supported by the presence of a finite NSC pool containing limited
self-renewal capability (Encinas et al., 2011; Pilz et al., 2018). Consistently, forced accelerated
neurogenesis (Ehm et al., 2010; Jones et al., 2015; Renault et al., 2009) and pathological conditions
(Mu and Gage, 2011; Sierra et al., 2015) can result in premature NSC depletion. In another
scenario, studies (Bonaguidi et al., 2011; Dranovsky et al., 2011; Licht et al., 2016) indicate that
self-renewing NSCs are maintained for prolonged periods and neurogenesis could instead decline
during aging due to an increase in NSC quiescence (Hattiangady and Shetty, 2008; Ziebell et al.,
2018). While NSCs as a population clearly undergo early age-related changes, when and how
specific NSC subpopulations begin to exhibit dysfunction are unclear. In this study, we used single
cell approaches to investigate the cellular and molecular mechanisms that initially compromise
adult neurogenesis. We determined that NSC subpopulations undergo asynchronous decline and
exhibit early molecular aging. We further show that targeting these cellular aging mechanisms in
the middle-aged brain can partially overcome age-related NSC dysfunction.
Results
Asynchronous NSC decline during aging
We first investigated how adult NSC numbers change over time at the population level.
Nestin staining for radial glial-like neural stem cells indicated that NSC number decreases over
time (Figure 2.1A). Consistent with prior observations, NSC loss is most pronounced during the
transition from the young to mature and middle-aged brain (Gontier et al., 2018). We further
processed NSCs for the cell cycle marker MCM2 to quantify age-related changes in radial glia like
15
NSC proliferation. Similar to findings in older mice (Encinas et al., 2011; Hattiangady and Shetty,
2008; Ziebell et al., 2018), remaining NSCs become more quiescent early in the mature brain
(Figure 2.1C, S2.1A). Therefore, NSC number declines over time and remaining NSCs do not
divide as frequently at the population level (Figure 2.1A-C, S2.1A).
Figure 2.1.
Asynchronous neural
stem cell (NSC) decline
during aging.
(A) Confocal
immunofluorescence
images showing NSCs
(Nestin
+
) and cell
proliferation (Mcm2
+
) in
the 3 mo and 12mo
mouse hippocampus.
Arrow - Nestin
+
Mcm2
+
aNSC, Arrowhead -
Nestin
+
Mcm2
+
qNSC.
(B) Quantification of
total NSC number across
ages. N=4-6 mice;
Mean± SEM. ANOVA
with Bonferroni post-hoc
test (B-C). (C)
Quantification of
quiescent NSC (Nestin
+
MCM2
-
) percentage
among total NSCs across
ages. (D) Cartoon of
NSC subpopulations in
the adult hippocampus.
(left) Nestin
#
-labeled
multipotential NSCs.
(right) Developmental-
like NSCs labeled by Ascl1
#
. A=astroglial lineage, IPC=intermediate progenitor cell, N=neuronal
lineage, G0=quiescent state, G1->M=active state. (E) Experimental design of in vivo single cell
clonal lineage tracing for Nestin
#
-NSC and Ascl1
#
-NSC subpopulations (Table S2.1-2). (F)
Confocal images of distinct RGL and IPC morphologies from Nestin
#
and Ascl1
#
clones acquired
from 6mo mouse. Arrow - GFP
+
GFAP
+
NSCs (RGLs); arrowhead - GFP
+
GFAP
-
IPC/NB
(neuroblast). (G) NSC clone maintenance (the time until 50% of clones exhibit NSC depletion) in
2, 6, 12mo mice. Model fits of clonal data from N=29-104 clones (Table S2.2). Data is derived
from modeling (Figure S2.2). Error bars show variability in the inferred values due to model
3
6
9
12
0%
20%
40%
60%
80%
100%
Age (Months)
NSC Quiescence Ratio (%)
*
*
*
Hilus
12mo
GFP
GFP
1
2
1
2
A B C
H I J
D E
F G
3
6
9
12
0
5000
10000
15000
20000
25000
30000
Age (Months)
Total NSCs/ mm
3
**
*
n.s
DAPI MCM2 Nestin
3mo
12mo
GCL
GCL
Hilus
Hilus
2 3 4 5 6
0.0
0.2
0.4
0.6
0.8
1.0
Age (months)
Average NSC number
per all clones
Nestin
#
Ascl1
#
**
***
n.s
GFP GFAP DAPI
NSCs (RGLs) IPC/NB
Pilz et al., 2018 Bonaguidi et al., 2011
Nestin
#
(2mo)
Nestin
#
(6mo)
Nestin
#
(12mo)
Ascl1
#
(2mo)
Ascl1
#
(6mo)
Ascl1
#
(12mo)
0
5
10
15
20
50% NSC clone
maintenance (mo)
16
sensitivity; derived by varying all real-valued model parameters by ±5% for all possible
combinations. (H) Cartoon depicting clonal maintenance and depletion. (I) Confocal images of a
maintained clone containing NSC (1) and IPCs (2); and a depleted clone containing neurons (1)
and (2). (J) Quantification of NSC homeostasis duration. Mean± SEM. (N = 29-104 clones, Table
S2.2). ANOVA with Tukey’s multiple comparisons test. Scale bar, 100μm stitched images [boxed
25μm] (A), 20μm (F), 10μm (I). *p<0.05, **p<0.01, ***p<0.001, n.s - not significant. (See also
Figure S2.1 and S2.2)
The existence of NSC subpopulations within the radial glial-like NSC pool may mask the
ability to precisely uncover which mechanisms contribute to NSC dysfunction over time
(Bonaguidi et al., 2012, 2016). Therefore, extensive in vivo single cell clonal lineage tracing and
computational modeling of Nestin-CreER
T2
(Nestin
#
) and Ascl1-CreER
T2
(Ascl1
#
) mice were
performed under physiological conditions at 2-, 6- and 12 months old (mo) according to our prior
approaches (Bonaguidi et al., 2011) (Figure 2.1D-F, S2.1E-G, Table S2.1). Over 1500 clones were
assessed for NSCs exhibiting radial glia-like characteristics (Table S2.2), including Nestin and
GFAP protein expression. We found that clonally labelled Ascl1
#
-NSCs behave as a short-term
neuronal fate-biased subpopulation, consistent with a developmental-like program (Pilz et al.,
2018) (Figure 2.1D-G, S2.1C-D, S2D-E). Analysis of Ascl1
#
-NSC clonal number over time
demonstrates consistent short-term stem cell maintenance in the young (2 to 6mo), mature (6 to
8mo) and middle-aged (12 to 14mo) hippocampus (Figure 2.1G, S2.1B). Meanwhile, individual
NSCs marked as Nestin
#
-NSCs are longer-lived and slowly cycle to generate neurons, astrocytes,
and additional NSCs (Figure 2.1D-G, S2.1C-D). Importantly, these Nestin
#
-NSCs are homeostatic
in the young brain for a few months – where the average number of NSC per clone is close to 1 -
then transition out of homeostasis at around 4.5mo mice (Figure 2.1H-J). Computational modeling
revealed that Nestin
#
-NSC clones are better maintained than Ascl1
#
-NSCs clones in the young,
mature and middle-aged hippocampus (Figure 2.1G, S2.2, Table S2.3). Further, remaining
Nestin
#
-NSCs display progressively increased NSC maintenance at older ages compared to
17
younger ages (Figure 2.1G, S2.1C-D). These data show consistent short-lived and longer-lived
properties among NSC subpopulations across multiple ages. Thus, Nestin
#
-NSCs represent a long-
term NSC subpopulation (LT-NSCs) and can serve as a platform to pinpoint mechanisms that
mediate the initial loss of stem cell homeostasis in the mature brain.
Increased quiescence drives NSC loss of homeostasis
NSCs are homeostatic when their rate of symmetric self-renewal (expansion) balances their
rate of differentiation (Figure 2.2A). We reasoned that NSCs either increase their differentiation
or slow their expansion during their transition out of homeostasis. NSC clonal depletion was first
tested from 2- to 6mo mice to analyze the rate of NSC differentiation. Clone survival was
calculated by identifying the percentage of clones per time point that contain 1+ number of NSCs.
As expected (Pilz et al., 2018), Ascl1
#
-NSCs displayed maintenance for about a week, then
underwent a rapid initial depletion that slowed with time (Figure 2.2B and S2.1G). Meanwhile,
Nestin
#
-NSCs exhibited a constant and gradual loss of clones containing a single NSC, indicating
that homeostasis decline does not occur due to accelerating NSC differentiation (Figure 2.2B). We
next analyzed NSC-containing clones to identify age-related changes in the rate of NSC expansion.
We did not detect appreciable expansion over time in Ascl1
#
-NSCs (Figure 2.2C, (Pilz et al.,
2018)). On the other hand, Nestin
#
-NSCs did expand in young mice according to predicted values
that maintain homeostasis (Figure 2.2C). However, NSC expansion quickly began to diverge from
levels needed to maintain homeostasis in approximately 4mo mice. Hence, Nestin
#
-NSCs lose
homeostasis in the mature brain because of slowing NSC expansion while Ascl1
#
-NSCs
consistently do not exhibit homeostasis (Figure 2.2A-C).
18
Figure 2.2. Increased quiescence drives NSC loss of homeostasis.
(A) Cartoon illustrating NSC homeostasis and its imbalance by changes in NSC expansion or
depletion. (B) Quantification of NSC clone maintenance. Statistics for predicted values use a
nonlinear second order polynomial regression (curve fit), GraphPad. (N = 30-69 clones, Table
S2.2). (C) Quantification of NSC self-renewal in the mature hippocampus, Nestin
#
-NSCs slow
their expansion in approximately 4mo mice indicating homeostatic imbalance. PV=predicted value
to maintain NSC homeostasis (N = 30-69 clones, Table S2.2). (D) Quantification of average cell
2mo
6mo
0
20
40
60
80
100
120
Time to first NSC division
(T
entry
, day)
Nestin
#
***
2mo
6mo
0
20
40
60
80
100
120
Time to second division
(T
c
, day)
***
Nestin
#
GFP
GFP
GFP
K
H I J
GFP
A B C
D E F G
1
1
1
1
2
1
3
2 3 4 5 6
0%
20%
40%
60%
80%
100%
Age (months)
NSC clone
maintenance (%)
Nestin#
Ascl1#
NSC differentiation
2 3 4 5 6
1.0
1.2
1.4
1.6
1.8
Age (months)
Average NSC number per
NSC containing clone
Nestin#
Ascl1#
NSC self-renewal
PV for Nestin#
PV for Ascl1#
1-2dpi
7dpi
30dpi
60dpi
0
1
2
3
4
5
Average number of cells
per NSC containing clone
Nestin#
2mo
6mo
**
**
n.s
n.s
NSC-NSC
NSC-N
NSC-A
NSC-N-A
0%
10%
20%
30%
40%
50%
60%
Fate composition among
all maintained clones (%)
Cell fate choices for Nestin
#
2mo --> 2mpi
6mo --> 2mpi
12mo --> 2mpi
**
***
0
*
**
19
number per NSC-containing clones across 2- and 6mo animals at multiple days post tamoxifen
injection (N = 30-69 clones, Table S2.2). ANOVA with Bonferroni post-hoc test. (E) Cartoon of
time to the first and second NSC divisions. (G0 = quiescence; G1 -> M = activation). (F)
Quantification of Nestin
#
-NSC cell cycle entry based on lineage analysis from 2 and 6mo mice.
Mean± SEM. (N = 30-69 clones, Table S2.2). ANOVA with Bonferroni post-hoc test (F-G). (G)
Quantification of Nestin
#
-NSC cell cycle re-entry based on lineage analysis from 2- and 6mo mice.
Mean± SEM (N = 30-69 clones, Table S2.2). (H) Schematic illustration of NSC fate choices.
NSC=neural stem cell; A=astroglial lineage; IPC=intermediate progenitor cell; N=neuronal
lineage. (I) Representative confocal images of clonal NSC self-renewing fate choices. 1 – NSC; 2
– N; 3 – A. (J) Quantification of self-renewing cell fate choices outcome from Nestin
#
, at 2, 6 and
12mo animals at 60 days post tamoxifen injection (Table S2.1). 0 – no observed phenotype. Mean±
SEM. ANOVA with Bonferroni post-hoc test. (K) Cartoon summary of the cellular mechanisms
driving age-related NSC dysfunction. Scale bar 10μm (I). *p<0.05, **p<0.01, ***p<0.001, n.s -
not significant. (See also Figure S2.1 and S2.2)
NSC expansion occurs in a two-step process: transition from quiescence into cell cycle and
symmetric cell division. To determine which stem cell behavior drives the loss of NSC
homeostasis, we performed clonal analyses, cell cycle kinetics and cell fate analyses on Nestin
#
-
NSC clones induced at 2-, 6- and 12mo Nestin
#
mice and traced for up to 4 months (Tables S2.1-
2.2). The average number of cells per NSC-containing clone was quantified to assess clonal
expansion. NSC clone sizes at older ages were smaller compared to younger ages. Interestingly,
the initial decline in clonal expansion occurred at 6 months of age and was consistently lower at
12 months of age (Figure 2.2D, S2.1I-J). We then assessed NSC quiescence by calculating the
time to cell cycle entry and re-entry according to power-law decay fitting of clonal tracings (Figure
2.2E, (Pilz et al., 2018)). Both durations to the first and second division increased in Nestin
#
-NSCs
between 2- and 6mo mice (Figure 2.2F-G). These findings suggest that Nestin
#
-NSCs display a
substantial increase in quiescence in the young hippocampus and enter a deeper quiescent state in
the mature brain (van Velthoven and Rando, 2019). To determine age-related changes in
production, we followed the generation of neuronal, astroglial, and additional NSC lineages within
Nestin
#
-NSC clones. Fate compositions were calculated for each lineage (Figure 2.2H-J, S2.1H).
20
As previously reported (Bonaguidi et al., 2011), Nestin
#
-NSCs predominately made neuronal and
astroglial asymmetric fate choices in the young 2mo hippocampus. However, we found that
symmetric cell divisions predominantly occur in the 6mo mature brain. Intriguingly, the age-
related changes in Nestin-NSCs cell fate choices occurring at 6 months of age are recapitulated in
middle-aged 12mo mice (Figure 2.2J, S2.1H). Together, these findings suggest that Nestin
#
-NSCs
prolong their quiescence with each division, which drives them from homeostasis in the mature
hippocampus. Further, they then switch to a symmetric cell fate choice after NSC homeostasis has
been lost in the mature brain, which further suppresses neurogenesis (Figure 2.2K).
Quiescent NSC molecular aging in the mature hippocampus
NSCs are a rare cell population in the adult hippocampus that transition between quiescent
and active states (Bonaguidi et al., 2011). While molecular signatures distinguishing quiescent
from active NSCs have been well documented in the SVZ and hippocampus (Codega et al., 2015;
Dulken et al., 2017; Llorens-Bobadilla et al., 2015; Shin et al., 2015), age-related changes within
NSC quiescence have been more difficult to identify (Kalamakis et al., 2019; Leeman et al., 2018).
Nestin
#
-NSCs are homeostatic in the young hippocampus for a few months, then exit homeostasis
at around 4.5mo mice due to an increase in quiescence (Figures 2.1-2.2). To uncover molecular
mechanisms of deepening NSC quiescence, we performed single cell RNA sequencing (scRNA-
seq) (Figure 2.3A, S2.3A). NSCs and their immediate progeny were isolated using 2mo (n=3) and
4.5mo (n=3) Nestin::CFP mice (Encinas et al., 2006) – as a single biological replicate, and
immediately enriched using FACS (Figure S2.3B).
Single-cell cDNA libraries were built using a miniaturized SMART-Seq4 protocol,
barcoded, multiplexed and sequenced using paired-end reads (Figure 2.3A, S2.3). 16,302 genes
were detected across cells, where the average cell expressed 3,600 detected genes to permit a deep
21
exploration of NSC transcriptional dynamics within quiescence (Table S2.5). We distinguished
NSCs (Nestin, Vimentin, Apoe, Fabp7) from intermediate progenitor cells (IPCs: Eomes/Tbr2,
Sox11, Stmn1) and identified NSCs in quiescent (qNSC: Aldoc, Aqpr4, Id3, Hes1) or active states
(aNSC: PCNA, Mcm7, Cdk4, Cdk6) using well established markers (Shin et al., 2015) (Figure
2.3C, S2.4A-C). Importantly, both Ward minimum variance hierarchical clustering and principal
component analysis (PCA) showed shared cell-state-specific expression profiles between the two
time points, indicating that global transcriptomic differences are greater between quiescent and
active NSC states than within NSC states cells from 2 and 4.5mo mice (Figure 2.3B, S2.4D-F).
Next, we probed for changes within quiescence between 2 and 4.5mo NSCs using RNA
velocity. RNA velocity takes separate observations of unspliced and spliced reads to predict the
future gene expression state of individual cells on a timescale of hours and reveal lineage
progression (La Manno et al., 2018). RNA vectors in our analysis indicated that most NSCs remain
quiescent but when activated NSCs either differentiate or return to quiescence (Figure 2.3D-F).
We did not detect age-related differences in activated NSCs or IPCs isolated from 2 and 4.5mo
mice. Instead, NSCs in quiescence are more likely to move away from activation in 4.5mo
compared to 2mo mice (Figure 2.3F), providing computational corroboration that NSCs begin to
enter deeper quiescence.
22
Figure 2.3.Quiescent NSC molecular aging in the mature hippocampus.
(A) Schematic illustration of the single cell RNA-Seq. (B) Principal component analysis of NSC
transcriptomes from 2mo (N = 3), 4.5mo mice (N = 3, pooled, single biological replicate), a total
of 89 cells were used, colors are represented k-means clusters. (C) NSCs identified by gene
enrichment analysis using Nestin
+
, Vimentin
+
, Fabp7
+
, Aldoc
+
, Apoe
+
markers. (D) RNA velocity
defines single cell future states (arrows) identified in quiescent NSCs (qNSC), active NSCs
(aNSCs) and IPCs from 2mo. (E) RNA velocity in qNSCs, active aNSCs and IPCs from 4.5mo.
2 month old 4.5 month old State changes
A B C
D E F
G H I
qNSC
aNSC
IPCs
NSC gene enrichment
4.5mo 2mo
qNSC DE
qNSC
aNSC
IPCs
2 mo
4.5 mo
- 2 months old
- 4.5 months old
Principal component analysis
qNSC
aNSC
IPCs qNSC
aNSC
IPCs
PC1 (71.87%)
PC2 (6.99%)
Upregulated Downregulated
J K L
PC1 (71.87%) PC1 (71.87%)
23
(F) RNA velocity analysis reveals NSCs more resistant to activate in 4.5mo mice, whereas some
qNSCs from 2mo mice will enter cell cycle. (G) Quiescent NSC differential expression heat map
between 2 and 4.5mo mice. (H) Top upregulated gene ontology (GO) terms between qNSCs
derived from 2- and 4.5mo mice identify molecular aging. FDR-corrected p-values are shown (H-
I). (I) Top downregulated GO terms between qNSCs derived from 2 and 4.5mo mice also reveal
molecular aging. (J) Heat map of Neurogenesis, Gliogenesis and Cell cycle transcript changes
between qNSCs derived from 2 and 4.5mo mice. (K) Heat map of Semaphorin, Ras and Rho
signaling and signaling transcript changes between qNSCs derived from 2 and 4.5mo mice. (L)
Heat map of Histone Demethylation, Transcription, NIK/NF kappaB signaling, DNA
recombination, DNA repair and Double strand break repair transcript changes between qNSCs
derived from 2 and 4.5mo mice. (See also Figure S2.3 and S2.4)
We then sought to identify transcriptomic changes associated with this NSC deeper
quiescence. 493 upregulated and 576 downregulated genes were differentially expressed (DE) in
quiescent NSCs from 4.5mo compared to 2mo mice (Figure 2.3G, S2.4D-F, Table S2.8). We tused
TopGo (Alexa et al., 2006) to represent differentially expressed transcripts (DEs) by their
biological processes. Consistent with our lineage tracing experiments (Figure 2.2), gene ontology
revealed changes in neurogenesis (Socs2, Sox11, Igf1, Wnt3, Epha4), a gain in self-
renewal/gliogenesis (Ezh2, Disc1, Mag, Plp1) and a loss of cell cycle (Wee1, Nsl1, Mcm6, Heca).
In addition, we observed changes in cell signaling regulators (Abl1, Abl2, Crh, Lef1), including
those known to modulate adult NSC quiescence or progenitor proliferation such as semaphorin
signaling (Plxna4, Plxnb3, Nrp2, Farp2), Ras signaling (Arhger11, Arhgep32, Rassf1, Cdh13,
Icmt) and Rho signaling (Spata13, Myo9b, Tiam2, Mcf2l, Scai) (Chavali et al., 2018; Jongbloets
et al., 2017; Li et al., 2012) (Figure 3J-L, S4G-H, Table S2.8). Strikingly, the remainder of detected
terms are molecular hallmarks of an intertwined process that drives cellular aging (Kennedy et al.,
2014; López-Otín et al., 2013). These factors include epigenetic dysregulation from histone
demethylation (Kdm1b, Kdm2a, Kdm4a, Kdm5d, Uty), downregulation of transcription (Sox30,
Mycn, Creb1, Hif1an, Sertad2, Epcam), inflammation due to NIK/NF-kappaB signaling (Cd14,
Malt1, Eif2ak2, Tirap, Myd88), changes in metabolism and proteostasis (Agbl2, Agbl4, Pigc, Pigl,
24
Pigz, Irgm2, Hhat, Zdhhc8); and cellular stress from loss of DNA recombination (Atad5, Hus1,
Pms2, Lig3, Mcm9), DNA repair (Sesn2, Msh3, Msh6, Rfc3, Rfc5, Rad18) and increased double-
strand break repair (Chek1, Blm, Spidr, Fancb) (Figure 2.3H-L, S2.4G-H, Table S2.5-8).
Therefore, functional changes occurring in deep NSC quiescence are associated with NSC early
molecular aging within the mature hippocampus.
Imatinib partially restores NSC function in the middle-aged brain
NSCs undergo early cellular and molecular aging in the mature brain (Figures 2.1-2.3). We
hypothesized that targeting this process could overcome age-related NSC dysfunction later in life.
Gene networks are powerful approaches with the ability to prioritize genes based upon their degree
of connectedness to other genes and functions (Li and Horvath, 2007). We therefore probed for
differentially expressed genes shared between 3 or more GO terms for target identification. Our
analysis identified Abl1, Abl2, Igf1, Lef1, Per2 and Nup62 as hub genes most connected to changes
in age-related NSC function – including regulation of signal transduction, gene expression, cell
cycle, DNA repair and neurogenesis (Figure 2.4A). We focused on targeting Abl1, because it has
many context-dependent signaling functions, but an unknown role in NSC biology (Wang, 2014).
Abl1 expression decreased between 2- and 4.5mo mice at the transcriptome level under
physiological conditions (Figure 2.4B). However, immunostaining for c-Abl (Abl1) within radial-
glia like NSCs (c-Abl
+
Nestin
+
) revealed more prevalent c-Abl protein in NSCs with chronological
age advancing from 2 to 10mo (Figure 2.4C-D). We then sought to inhibit Abl with Imatinib as a
strategy to target NSC cellular aging. Imatinib or vehicle control were infused into the
hippocampal fimbria of 10mo mice for 6 days (Figure 2.4E, S2.4I). We found that Imatinib
treatment decreases the prevalence of Abl1 protein in NSCs (Figure 2.4F-G). Interestingly, Nestin
staining of radial-glial like NSCs with the cell cycle marker MCM2 revealed that Imatinib
25
increases the percent of proliferating NSCs in the middle-aged hippocampus back to younger
levels without altering NSC number (Figure 2.4H-J).
Figure 2.4. Imatinib
partially restores
NSC function in the
middle-aged brain.
(A) String network
graph depicting age-
related changes
downregulated in
4.5mo qNSCs. Shown
genes have 3+
connections to
biological processes.
(B) Violin plot of
Abl1, Abl2
expression in qNSCs
in 2 and 4.5mo mice.
edgeR (quasi-
likelihood F-test),
FDR-corrected p-
values are shown. (C)
Confocal images of c-
Abl and Nestin co-
expression across
multiple ages (2, 6, 10
mo). Arrowhead -
Nestin
+
c-Abl
+
NSCs.
(D) Quantification of
double positive NSCs
(Nestin
+
c-Abl
+
)
fraction among total
Nestin
+
NSCs in 2, 6
,10mo mice. Mean ±
SEM. (N = 2-3 animals per group). ANOVA with Bonferroni post-hoc test. (E) Schematic
illustration of the experimental design for intracranial drug infusion. (F) Confocal images of c-Abl
and Nestin co-expression in Vehicle-Control and Imatinib-treated brains. Arrowhead - Nestin
+
c-
Abl
+
NSC. (G) Quantification of double positive NSC (Nestin
+
c-Abl
+
) fraction among total
Nestin
+
NSCs in the 10mo Vehicle-Control and Imatinib-treated animals. Values represent ± SEM.
N = 3 animals per group. ANOVA with Bonferroni post-hoc test. (H) Confocal
immunofluorescence images for NSCs (Nestin
+
) and cell proliferation (Mcm2
+
) in 10mo Vehicle-
Control and Imatinib-treated mice for 6 days. Arrow - Nestin
+
NSC, Arrowhead - Nestin
+
MCM2
+
NSC. (I) Quantification of total NSC number (Nestin
+
) in 10mo Vehicle-Control and Imatinib-
treated mice for 6 days. Mean± SEM. N = 9 mice (Vehicle), N = 8 mice (Imatinib – treated). Man-
A B C D
2mo
6mo
10mo
0%
5%
10%
15%
c-Abl
+
Nestin
+
NSCs (%)
**
**
***
10mo Vehicle
Vehicle
Imatinib
0%
5%
10%
15%
c-Abl
+
Nestin
+
NSCs (%)
*
2mo
6mo
10mo
c-Abl Nestin DAPI
10mo Imatinib
10mo Vehicle
E F G
c-Abl Nestin DAPI
Vehicle
Imatinib
0%
10%
20%
30%
40%
Active NSCs (%)
*
6 days 6 days
Vehicle
Imatinib
0%
10%
20%
30%
Active NSCs (%)
*
Nestin MCM2 DAPI Nestin MCM2 DAPI
Nestin MCM2 DAPI Nestin MCM2 DAPI
DCX DAPI DCX DAPI
Vehicle
Imatinib
0
2000
4000
6000
8000
Total NSCs/ mm
3
n.s
Vehicle
Imatinib
0
2000
4000
6000
8000
Total NSCs/ mm
3
n.s
Vehicle
Imatinib
0
2000
4000
6000
8000
10000
12000
14000
DCX
+
cells/ mm
3
6 days
n.s
Vehicle
Imatinib
0
2000
4000
6000
8000
10000
12000
14000
DCX
+
cells/ mm
3
28 days
n.s
28 days 28 days
H I J
K L M
N O P
Imatinib Vehicle
Imatinib Vehicle
Imatinib Vehicle
6 days 28 days
26
Whitney U-test. (J) Quantification of the aNSC (Nestin
+
MCM2
+
) fraction among total Nestin
+
NSCs in 10mo Vehicle-Control and Imatinib-treated mice for 6 days. Mean± SEM. N = 9 mice
(Vehicle), N = 8 mice (Imatinib – treated). Man-Whitney U-test. (K) Confocal
immunofluorescence images for NSCs (Nestin
+
) and cell proliferation (Mcm2
+
) in 10mo Vehicle-
Control and Imatinib-treated mice for 6 days and sacrificed at day 28. Arrow - Nestin
+
NSC,
Arrowhead - Nestin
+
MCM2
+
NSC. (L) Quantification of total NSC number (Nestin
+
) in 10mo
Vehicle-Control and Imatinib-treated mice for 6 days and sacrificed at day 28. Mean± SEM. N =
7 mice per group. Man-Whitney U-test. (M) Quantification of the active NSC (Nestin
+
MCM2
+
)
fraction among total Nestin
+
NSCs in 10mo Vehicle-Control and Imatinib-treated mice for 6 days
and sacrificed at day 28. Mean ± SEM. N = 7 mice per group. Man-Whitney U-test. (N) Confocal
immunofluorescence images for newborn neurons (DCX
+
) in 10mo Vehicle-Control and Imatinib-
treated mice. Arrow – DCX
+
. (O) Quantification of newborn neuron (DCX
+
) number in 10mo
Vehicle-Control and Imatinib-treated mice for 6 days. Mean ± SEM. N = 9 mice (Vehicle), N = 8
mice (Imatinib – treated). Man-Whitney U-test. (P) Quantification of newborn neuron (DCX
+
)
number in 10mo Vehicle-Control and Imatinib-treated mice for 6 days and sacrificed at day 28.
Mean ± SEM. N = 7 mice per group. Man-Whitney U-test. Scale bar, 20μm (C, F, H, K, N)
*p<0.05, **p<0.01, ***p<0.001, n.s - not significant. (See also Figure S2.4)
We then sought to determine longer-term consequences of transient Imatinib treatment on
NSC function. Imatinib or vehicle control was infused into the hippocampal fimbria of 10mo mice
for 7 days and mice were sacrificed at day 28. We then assayed total NSC number and proliferation
(Nestin
+
MCM2
+
radial glia like cells). Surprisingly, the NSC pool was not prematurely depleted
by hyperactivation (Figure 2.4L), but instead became more quiescent (Figure 2.4K, M). Newborn
neuron number (DCX
+
cells) was not significantly changed in Imatinib-treated brains compared
to vehicle control after 6 and 28 days (Figure 2.4N-P). Together, these population-level data
indicate that Imatinib treatment causes NSC activation from quiescence in the middle-aged brain
and then NSCs rebound to become more quiescent without prematurely depleting.
Discussion
Neurogenesis is well known to decline with age. However, complexities within the NSC
pool has complicated the association of age-related neurogenesis decline to their cellular origins
(Bonaguidi et al., 2011; Dranovsky et al., 2011; Encinas et al., 2011; Pilz et al., 2018). Our single
27
cell lineage tracing suggests that asynchronous NSC dysfunction is a unifying principle of NSC
aging. We found that Ascl1
#
-labeled radial cells represent neuronal-biased ST-NSCs, whose
function are computationally consistent with “disposable” and “developmental-like” NSC
behavior (Encinas et al., 2011; Pilz et al., 2018). Meanwhile, co-existent Nestin
#
-labeled radial
cells are multipotential LT-NSCs that produce neurons, astrocytes, additional stem cells, and are
computationally consistent with “self-renewing” NSCs (Bonaguidi et al., 2011; Dranovsky et al.,
2011). However, LT-NSCs appear to increase their time in quiescence with each division until
homeostasis is compromised and their numbers begin to decline in the mature brain. In addition,
LT-NSCs change their fate from asymmetric to symmetric divisions in the mature brain and
maintain this lineage bias into middle age. These findings build upon recent intravital lineage
tracing in the young brain (Bottes et al., 2021) to extend the principle of short-term and long-term
NSCs into older ages. The two studies also implicate the presence of co-incident NSC
subpopulations and that animal age itself does not define those models of NSC behavior, as was
recently suggested (Bottes et al., 2021).
While stem cells undergo cellular aging in many old tissues (Goodell and Rando, 2015;
López-Otín et al., 2013), only recently have these mechanisms been explored in the brain.
Quiescent neural stem cells in the old SVZ exhibit alterations in proteostasis and inflammation
(Kalamakis et al., 2019; Leeman et al., 2018). However, transcriptome investigation between
young and old SVZ quiescent NSCs has thus far not revealed substantial age-related differences
(Kalamakis et al., 2019; Leeman et al., 2018). We therefore utilized deeper sequencing to provide
enhanced sensitivity (>1000 DEs) of transcriptomic changes at chronological ages not typically
analyzed. Remarkably, nearly all pillars of cellular aging (Kennedy et al., 2014) were found to
change within NSC quiescence. Consistent with NSCs from the aged SVZ (Kalamakis et al., 2019;
28
Leeman et al., 2018; Negredo et al., 2020), NSCs from the mature hippocampus already undergo
changes in proteostasis and receive inflammatory signals. Additionally, age-related DNA damage,
histone demethylase dysregulation, metabolic and transcription alterations, decreased cell cycle
entry and altered cell fate biases are all detected in quiescent NSCs within the mature brain. These
outcomes substantially differ from a recent study that performed transcriptome analysis on NSCs
from 6 month-old compared to 1-month-old mouse hippocampus and did not report pathways that
typify aging (Bottes et al., 2021). Instead, our study strongly implicates that NSCs exhibit early
aging in the hippocampus and presents a resource of associated molecular mechanisms at the
systems level.
Recent studies utilizing manipulation of blood factors, niche factors, and metabolism (diet
and exercise) have demonstrated the capability to partially restore hippocampal neurogenesis and
cognition (de Cabo et al., 2014; Horowitz et al., 2020; Katsimpardi and Lledo, 2018; Mahmoudi
et al., 2019). However, radial-glia-like NSCs are commonly perceived to be a poor target for anti-
aging interventions because they are stated to have limited self-renewal capability (Encinas et al.,
2011; Pilz et al., 2018) and NSC hyperproliferation results in their premature depletion (Ehm et
al., 2010; Jones et al., 2015; Renault et al., 2009; Sierra et al., 2015). Importantly, those latter
studies investigated NSCs upon sustained genetic perturbation and disease. We took a different
approach by investigating NSC responses to transient pharmaceutical intervention. Interestingly,
targeting this molecular aging component with Imatinib in the middle-aged brain was sufficient to
transiently boost NSC activation rates back to younger levels. But NSCs later rebounded to a more
quiescent state without prematurely depleting – suggesting an adaptive response. This unexpected
NSC plasticity provides indirect evidence for a role of Abl signaling in regulating NSC dynamics.
Such behavior could reflect NSC clonal selection where specific stem cells adapt to local niche
29
pressure as observed during aging in other tissues (T. Krieger & B. D. Simons, 2015; Liu et al.,
2019). Therefore, understanding the key mechanisms that drive NSC molecular aging will help
to create new directions towards advancing age-related regenerative treatments.
Materials and Methods
Animals and Tamoxifen Administration
All animal procedures were performed in accordance with institutional guidelines of
University of Southern California Keck School of Medicine and protocol (20287) approved by
Institutional Animal Care and Use Committee (IACUC). All mice used in the study were
backcrossed to the C57BL/6 background to ensure the reproducibility of clonal induction with
specific doses of tamoxifen. Animals were housed in a 12-hour light/12-hour dark cycle with free
access to food.
Nestin-CreER
T2
mice (Balordi and Fishell, 2007) and Ascl1-CreER
T2
mice (Strain:
Ascl1tm1.1(Cre/ERT2)Jejo/J) (Kim et al., 2011) were used to clonally label RGLs. The following
genetically modified mice were originally purchased from Jackson Labs: Rosa-YFP
f/f
(Strain:
B6.129X1-Gt(ROSA)
26Sortm1(EYFP)Cos
/J), mT/mG
f/f
(Strain: B6.129(Cg)-Gt(ROSA)
26Sortm4(ACTB-
tdTomato,-EGFP)Luo
/J), Confetti
f/f
(StrainGt(ROSA)26Sor
tm1(CAG-Brainbow2.1)Cle
/J).
Nestin-CreER
T2
and Ascl1-CreER
T2
mice were crossed to fluorescent reporter mice for
clonal analysis. Nestin-CreER
T2
::Confetti
f/+
mice were generated by breeding Nestin-CreERT2+/-
mice with Confetti
f/f
mice, or by crossing Nestin-CreER
T2+/-
::Confetti
f/f
mice with wild-type
C57BL/6 mice. Ascl- CreER
T2+/-
::Confetti
f/f
, Ascl1-CreER
T2+/-
::YFP
f/+
or Ascl1-CreER
T2+/-
::mTmG
f/+
mice were generated by crossing Ascl1-CreER
T2+/-
with Confetti
f/f
, YFP
f/f
or mTmG
f/f
,
respectively. For each CreER
T2
driver, we tested different tamoxifen doses and reporter lines, and
30
obtained combinations that exhibited high specificity, inducibility and reproducibility (Table S2.1-
2, S2.4).
At least 3 animals were checked for each reporter/driver combination to ensure there was
no recombination in the adult SGZ in the absence of tamoxifen. A stock of tamoxifen (66.6
mg/mL) was prepared in a 5:1 ratio of corn oil to ethanol at 37C with occasional vortexing. A
single tamoxifen or vehicle dose was intraperitoneally injected into 8- to 10-week-old, 26- to 30-
week-old, or 56-week-old mice at various concentrations for lineage tracing (Table S2.1). Injected
animals showed no signs of distress.
Primer sets from original publications were used to identify genetically modified mice
(Ahn and Joyner, 2005; Balordi and Fishell, 2007; Kim et al., 2011; Lemberger et al., 2007;
Muzumdar et al., 2007). Genomic tail DNA was isolated in a 25mM NaOH, 0.2 mM EDTA
solution and ran for 35 PCR cycles.
Immunostaining, Confocal Imaging, and Processing
Mice were anesthetized with isoflurane gas and underwent transcardial perfusion with
saline followed by 4% paraformaldehyde. Brains were post-fixed overnight in 4%
paraformaldehyde and then immersed in 30% sucrose for a subsequent 48 hours prior to sectioning.
Brains were sectioned into 45µm coronal sections through the entire dentate gyrus.
Immunohistology was performed with antibodies as previously described (Bonaguidi et al., 2011)
on sections in serial order using custom, in-house staining chambers. Brain sections were washed
in TBS with 0.3% Triton-X100 prior to staining and mounting. Goat anti-GFP (1:1000), rabbit
anti-RFP (1:1000), rabbit anti-GFAP (1:2000), chicken anti-Nestin (1:500), mouse anti-MCM2
(1:500), rabbit c-Abl (1:75) primary antibodies were used. Antigen retrieval for MCM2, Nestin
and c-Abl antibodies utilized DAKO citrate buffer (Dako, Cat#: S1699) at 95
o
C for 30 min and
31
then samples were left for 1 hour to cool at room temperature. Cells were then counted using every
8th section throughout the entire dentate (n = 6-7 sections per dentate) for a stereological analysis
and every section was processed for clonal analysis. GFP+ cells were identified with a Zeiss
AxioObserver.A1 microscope and were acquired as a z-stack on a Zeiss LSM700 confocal system
under 40X or 63X magnification (stitching were done at 0.5 overlap). Morphological analysis was
done using Imaris 8.0 Software.
Clonal Analysis
Clonal analysis and categorization of cell types by morphological and immunohistological
criteria were consistent with prior criteria (Bonaguidi et al., 2011). Analyzed dentate volume
included the stratum granulosum (granule cell layer), and SGZ. Serial sections were first screened
for candidate clones, which were defined as possessing at least (1) an NSC (RGL), (2) neuronal
cell(s) in close spatial proximity, or (3) astroglia in close spatial proximity to other astroglia or
neuronal cells. Approximately 6 - 24 clones per dentate per color (in the case of Confetti) allowed
for clonal analysis based on prior computer simulations (Bonaguidi et al., 2011) (Table S2.2).
Clones were randomly induced throughout the dentate.
Dividing clones were marked with two separate nuclei by DAPI under confocal microscopy.
The two nuclei were completely encompassed when using membrane-bound reporters (Rosa-CFP,
mT/mG, Confetti-CFP), or completely filled when using cytoplasmic-bound reporters (Rosa-YFP,
Rosa-RFP, Confetti-YFP, Confetti-RFP). R26R – Confetti is a Stochastic Multicolor Cre-reporter.
Upon Cre activation, the neomycin roadblock is excited, while the Brainbow 2.1 (Livet, J et al.
2007) recombines in a random fashion to four possible outcomes. GFP is a nuclear, CFP is
membrane associated, RFP and YFP are cytoplasmic. Clones with more than one NSC and
32
progeny were classified by the lineage produced: neuronal, astroglial, or both. NSC maintenance
was assessed as a percentage of clones that contained at least one NSC.
Stereotaxic surgery (Imatinib infusion via osmotic pump)
Stereotaxic surgery was performed on 10-months-old C57/BL6 mice as described (Pan,
2015; Pan et al., 2012; Wong et al., 2000). Mice were anesthetized using an isoflurane machine
(5% until recumbent, 2-3% maintenance). Prior to anesthesia, non-steroidal anti-inflammatory
analgesic - 1X Ketoprofen (5 mg/kg) was injected. Osmotic pump with the drug (1 mM Imatinib
treated) or 10% DMSO (Vehicle control) was unilaterally implanted at a rate of 0.5μl/hr for 6 days
(Figure 2.5 and 2.6) into the hippocampal fimbria with the following coordinates relative to
Bregma: - 0.8 mm posterior, - 0.75 mm medial-lateral, and 2.5 mm ventral and mice were
sacrificed on the day 7 or day 28 after drug infusion.
Fluorescence- activated Cell Sorting (FACS) of Individual cells from Adult Mouse Dentate
Gyrus
Homozygous Nestin:CFP
nuc
mice (Strain: B6.Cg-Tg(Nes-CFP)1Enik/J) (Encinas et al.,
2006) were used for all single-cell RNA-seq experiments (Shin et al., 2015). Mice were euthanized
by cervical dislocation, and brains were immediately immersed into cold Hibernate A solution
(BrainBits). Dentate gyri were dissected under a stereomicroscope as previously described
(Hagihara et al., 2009). All procedures were performed with approved protocols in accordance
with institutional animal guidelines. The single cell suspension was prepared by using Neural
Tissue Dissociation Kit (P), with the addition of one cleaning step with Percoll (1:10 dilution) to
remove myelin layer and cellular debris. Propidium Iodide was added to determine cell viability
and cells were sorted on a BD FACS Aria II with a 70μm nozzle at 13 psi (Figure S2.3). Single
33
cells were collected onto the ice-cold 8-sample parafilm-covered glass slides with 1.25uL of 1X
lysis mixture (Clontech Laboratories SMART-Seq v4 Ultra Low Input RNA Kit for sequencing)
and immediately transferred in individual 0.2-ml RNase-free 8-well strip (Figure S2.3). To avoid
any effects due to the FACS sorting protocol, cells were kept on ice the entire procedure, and the
actual collection was performed at 4C.
scRNA-seq library preparation and sequencing
cDNA was generated using the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech
Laboratories) for sequencing with modifications to bring down the cost. Briefly, all reagents from
the kit (for Reverse transcription and cDNA amplification) were miniaturized 10-fold by utilizing
FACS collection in a very low volume of the lysis buffer. The same quality of cDNA was achieved
as for the original kit. The amplification product was purified using Ampure XP beads according
to the manufacturer’s protocol. Quality control and validation were performed by using qPCR for
Actin expression and Agilent TapeStation 4200 (Agilent Technologies).
The amplification product was sheared using library preparation Nextera XT Kit (Illumina
Inc.) following the manufacturer’s protocol. In brief, the purified cDNA was tagmented and
fragmented using a miniaturized protocol. cDNA fragments were barcoded using i5 and i7 indices
(Illumina Inc.) and the product was purified using Ampure RNA clean beads. The purified ligation
product underwent 13 cycles of PCR. A Mantis Liquid Handler (Formulatrix, Inc.) was used to
handle low volume pipetting. The libraries were multiplexed and sequenced using Illumina
NextSeq 550 (Illumina Inc.) 75 paired end run. To minimize the batch effect on the sequencing all
cells were loaded into one chip.
34
Computational analysis of RNA-Seq data
Preprocessing, Alignment and Count Estimation of RNA-Seq data
FASTQ files were demultiplexed into two cell-specific FASTQ files by i5 and i7 adapter
sequences for forward and reverse ends of the paired-end reads. Raw reads were trimmed of TSO
sequence and adaptor contaminants using Trimmomatic (Bolger et al., 2014). Sequences were then
aligned using STAR (Dobin et al., 2013) to the mm10 mouse transcriptome, constructed using
UCSC annotation framework, generating cell-specific BAM files. Only uniquely mapped reads,
concordant between both directions of the paired end reads were preserved. RSEM (Li and Dewey,
2011) was then employed to produce spliceoform-level count estimates, which were summated to
give gene-level counts per cell. Cells with fewer than 150,000 uniquely mapped reads were
excluded. Remaining cells then underwent transcripts per million (TPM) normalization (Li and
Dewey, 2011) to control for variable gene length within samples and sequencing depth between
samples. Cells with fewer than 1,000 detected genes were also excluded, where a gene was
considered detected in a cell if > 2 TPM. After cell exclusions, 48 cells from 2-month-old mice
and 41 cells from 4.5-month-old mice remained, totaling 89 cells. 3,603 genes were detected on
average (SD = 833).
Clustering, Dimensionality Reduction and RNA Velocity
K-means clustering, employing the Hartigan-Wong algorithm, was run on all 89 cells in
full 24,411-dimensional expression space. K was set equal to 5, as (Shin et al., 2015) described 5
clusters present in adult neural stem cells. Dimensionality reduction of 89 cells was performed by
PCA, using the FactoMineR (Lê et al., 2008) package with default parameters.
RNA Velocity (La Manno et al., 2018) was used to estimate the future expression state of
single cells. Count estimates of unspliced and spliced cells were computed with the following
35
command velocityto run_smartseq2 -d 1 $bam_files $ucsd_annotation_mm10.gtf. Unspliced and
spliced count estimates were imported to the gene.relative.velocity.estimate() function to produce
future expression state estimates for every cell. Future expression states were represented in
previously computed PCA embedding as arrows. An independent samples t-test was conducted to
compare the magnitude of future state prediction vectors from 2 month old and 4.5 month old
individual cells (Figure 2.3F). The cells from 4.5 month old animals had significantly longer
vectors in principal component space (M = 4,568.99, SD = 2,095.80) than cells from 2 month old
animals (M = 3,292.23, SD = 1,589.42); t(49.94) = 2.63, p = 0.011.
To represent the extent to which aging dampens the tendency for cells to proceed down the
neurogenic trajectory, we visualized the future state vectors of each age group at each cluster.
Vectors were centered on the mean coordinate within their respective cluster in PC space. The X
corresponds to PC1 coordinate space, where Y corresponds to PC2 coordinate space. First mean,
coordinates for each cluster were derived. Cells within any one cluster were computed thusly –
Future state predictions were then aligned atop each other at the mean coordinate by -
36
This computation was run independently on cells from the 2-month-old and 4.5-month-old groups
to produce single cell estimates.
Differential expression and Gene Ontology enrichment
Differential expression on NestinCFP
+
scRNA-Seq generated for this paper was computed
using the edgeR package in R (Robinson et al., 2010). Specifically, we employed the general liner
model quasi likelihood fit test edgeR (edgeRQLFDetRate), as it was optimal across all major
evaluation criteria in a recent survey of contemporary differential expression methods developed
for single cell RNA Sequencing (Soneson and Robinson, 2018). In all use cases, age group was
the only factor for the generalized linear model, where there were two age groups (2-month-old
and 4.5-month-old). Gene filtering is required for accurate implementation of edgeR. Only genes
with a mean count greater than 5 in at least one of the age groups were included any analysis. P-
values were group-adjusted using the Benjamini-Hochberg (Hochberg and Benjamini, 1990)
method to produce a false detection rate (FDR). Genes with an FDR <=0.05 were preserved for
downstream analyses.
TopGO (Alexa et al., 2006) was employed to derive enriched Gene Ontology terms
associated with aging in neural stem cell niche. Two ontology enrichment lists were generated for
each niche - one from upregulated genes and another from downregulated genes. Enrichment was
calculated using the classic Fisher’s test (Fisher, 1956), where exact p-values were corrected for
multiple comparisons using Benjamini-Hochberg (Hochberg and Benjamini, 1990). The gene
universe was set to 16,302 which is equivalent to the number of genes detected in at least one cell
across all sequenced cells.
String network visualization
37
To represent connectivity between genes and GO terms associated with NSC aging, we
wrote custom functions to visualize a string network. The underlying object for this network is
represented as the bipartite graph G = (V,E) where, V(G) is a set of vertices, which can be
partitioned into two independent sets, GO terms A, and genes B. To reduce visual complexity, we
generated the subgraph H where V(H)⊆ V(G) and E(H)⊆ E(). The minimum allowable edges is
a, thus conditionally δ(H) = a. Once constructed, graphs were metricized and visualized as a string
network using force-directed Kamadakawai spatial organization in the GGally package (Schloerke
et al., 2014).
Computational modeling of clonal data
To consolidate the discussion in the main text, in the following we further explain the fitting
and modelling approaches used.
Figure 2.2C fits (linear rate changes) were calculated from Figure 2.2B by using to the
number of NSCs required to expand to offset clonal loss as a function of time (age). As in Figure
2.2B, we performed a nonlinear regression - second order polynomial - to fit the Nestin
#
data.
Ascl1
#
data utilizes scatter nearest neighbor since other fits do not perform well.
Distribution of activation times
For Figure 2.2F, the activation time of a radial glial-like NSC (RGL) is the time taken to
first enter into cell cycle upon labelling (Tentry). To deduce the distribution of activation times from
the clonal data, we calculated at each time point the fraction of RGLs that had not yet divided.
NSCs divide with a transition rate (corresponding to an average cycle time) and go through rounds
of duplication (symmetric self-renewing divisions), followed by rounds of asymmetric division,
generating a certain amount of neuronal progeny upon each division, followed by round of terminal
differentiation.
38
As the labelling protocols targeted a small proportion of IPCs in addition to RGLs, there
was a degree of error involved in our assignment: we could not decide unambiguously whether a
clone consisting entirely of IPCs and differentiated cell types was originally derived from an RGL
or an IPC. In the case of Ascl1
#
-NSC 2-month-old clonal data, the number of such clones was so
small that their exclusion would not significantly affect our results. For Nestin
#
-NSC 2-month-old,
10/47 clones consist only of 1-2 IPCs at 2 dpi; we assumed that these were IPC-derived and
excluded them from the analysis. By 7 dpi, due to the short cell cycle time of IPCs (Berg et al.,
2018; Hayes and Nowakowski, 2002; Hodge et al., 2008), we would expect IPC-derived clones to
have grown in size; we therefore took clones containing 6 or more IPCs to be IPC-derived (2/40),
and clones consisting only of 2 IPCs to be RGL-derived (4/40). At later time points, we assumed
that all clones were RGL-derived. In case of the 6- and 12-month-old data Nestin
#
-NSC and
Ascl1
#
-NSC subpopulations all non-RGL containing clones were excluded from analysis. As a
second caveat, a clone consisting of a single RGL was scored as an undivided RGL but may have
given rise to progeny that were subsequently lost through cell death. However, effects of any
erroneous assignments were likely negligible compared to experimental noise over this timescale.
By performing a weighted least-squares fit, we deduced that Nestin
#
-NSCs enters cell cycle
at a rate = 0.044 ± 0.005 per day, equivalent to a mean cell cycle time of = 23 ± 3 days. Similar
fits to the early-time Ascl1-CreER
T2
data suggested mean activation times of 0.35 ± 0.04 days,
respectively.
During each differentiating division, the amount of neuronal progeny generated
(interpreted as containing all cell types down the neuronal lineage) is exponentially distributed
with a mean proliferative output. This procedure takes into account amplification by intermediate
progenitors. Neuronal progeny dies at a rate. Initially, a clone starts as a single NSC (with
39
probability) with uniform probability to find it at any state within the programme or a neuronal
progenitor/neuron (with probability).
Cell cycle re-entry times
For Figure 2.2G we can estimate cell cycle re-entry times from the fraction of cells that
have divided exactly once. For a population of cells with activation rate and cell cycle re-entry
rate , the fraction of cells divided once at time , denoted , satisfies
!"
!
($)
!$
=∙
&
()−∙
'
(),
where
&
()=
( )$
is the fraction of cells that remain undivided at time .
Multiplying by the integrating factor
*$
yields
*$
∙
!"
!
($)
!$
=∙
(*( ))$
−∙
*$
∙
'
(),
with the solution
'
()=
)
*( )
∙
+
(#$%)'
, -
+
#'
, where is a constant. Since
'
(0)=0, we must have
=−1 and therefore
'
()=
)
*( )
∙
+
(#$%)'
( '
+
#'
=
)
*( )
∙/
( )$
−
( *$
0.
Given our estimates of the activation times, weighted least-squared fits suggested cell time
to the first Nestin
#
- NSC division for 2 months old is 23 ± 3 days, for 6 months old is 38 ± 3 days;
cycle re-entry times for 2 months old is 40 ± 3 and for 6 months old is 100 ± 4 days (Figure 2.2F-
G).
Model of a developmental-like stem cell fate program (Short Term - Ascl1
#
-NSC). Here we
introduce a simplified Markovian model that captures the progression of radial-glia like neural
stem cells (NSCs) through a developmental-like program that consists of a proliferative phase,
followed by a neurogenic phase and terminal differentiation (Pilz et al., 2018). In our model, NSCs
divide with a transition rate (corresponding to an average cycle time =
( '
) and go through
&
rounds of duplication (symmetric self-renewing divisions), followed by
'
rounds of
λ
µ
t
R
1
(t)
t
40
asymmetric division, generating a certain amount of neuronal progeny upon each division,
followed by
.
=1 round of terminal differentiation. During each differentiating division, the
amount of neuronal progeny generated (interpreted as containing all cell types down the neuronal
lineage) is exponentially distributed with a mean proliferative output
/
. This procedure takes
into account amplification by intermediate progenitors. Neuronal progeny dies at a rate
/
.
Initially, a clone starts as a single NSC (with probability ) with uniform probability to find it at
any state within the programme or a neuronal progenitor/neuron (with probability 1−).
Formally, the state of a clone is given by the vector (,)=
(
&,&
,...,
&,"
(
( '
,
&,'
,...,
&,"
!
( '
,
.,&
,), where
1,2
is the number of NSCs that are in phase ,
where =0 corresponds to the proliferative phase, =1 corresponds to the phase of asymmetric
neurogenic divisions and =2 corresponds to the phase of terminal differentiation, and is the
number of divisions that a cell has gone through in the respective phase; denotes the neuronal
content. The dynamics of the probability =(,,) to find a clone with configuration (,)
at time is described by the master equation
(1)
where we have defined the following operators:
1
() describes the progression through the phase
=0,1,2 of the developmental-like programme,
1
() = ∑
23&
"( .
[(
1,2
+1)ℝ
1,2
,
(ℝ
1,2, '
(
)
.( 1
ℙ
/
(
/
)
1
−
1,2
]
+(
1,"( '
+1)ℝ
1,"( '
,
(ℝ
1, ',&
(
)
.( 1
ℙ
/
(
/
)
1
−
1,"( '
, (2)
41
where the
1,2
±
are ladder operators that increase/decrease the number of RGLs at stage (,)
according to
1,2
±
(...,
1,2
,...)=(...,
1,2
±1,...) and
ℙ
/
()= ∑
53&
6
ℰ
5
()(ℕ
(
)
5
(3)
is the operator that generates neuronal progeny with an exponentially distributed number of
newborn cells with average with ℰ being a discrete exponential distribution; here, ℕ
±
are the
ladder operators for the variable . Furthermore, we define the operator describing death of
neuronal progeny as
=(+1)ℕ
,
− . (4)
Here, we impose the boundary condition (,,)=0 if
1,2
<0 for at least one (,).
Motivated by the fact that Ascl1
#
-targeted cells seem to be primed for entry into cycle (see main
text), we allow the first division to occur at a faster rate
'
=, where >1 indicates the fold-
change as compared to the re-entry rate . Formally, this entails a straightforward modification of
Eq. (1), which we omit here since it obscures the basic features of the stochastic dynamics. The
model is solved numerically using a standard stochastic simulation algorithm of the Gillespie-type
(Gillespie, J. Chem. Phys. 1977).
To describe the clonal dynamics of the Ascl1
#
-targeted population, we identify model parameters
using the following strategy. First, all parameters are determined simultaneously by comparing the
42
model to the 2-months age dataset, fitting the time-dependent average clone content , the time-
dependent ‘RGL survival’ , defined as fraction of clones containing at least one RGL (Figure
S2.2) and the average number
∗
of RGLs per RGL-containing clone,
()=O
1,2
P
1,2
()Q
∗
()
,
()=
'( ∑
)
9(,;,$)
9
∗
($)
,
∗
()= ∑
1,2
⟨=
+,-
($)⟩
'( ∑
)
9(,;,$)
,
where
∗
()=1−(0,0,) is the total surviving fraction of clones. For successive later ages, all
parameters are held constant except for the neuronal death rate
/
, which is determined
individually for each age by refitting the model to the respective dataset. To implement this fit
strategy, we compute the sum of squared differences between experimental data points and
simulations for each of these quantitites individually and then obtain a combined cost function
by multiplying these individual residuals,
() =(∑
2
(
?2@
(
2
|)−
+AB
(
2
))
.
)(∑
2
(
?2@
(
2
|)−
+AB
(
2
))
.
)(∑
2
(
?2@
∗
(
2
|)−
+AB
∗
(
2
))
.
)
,
where is the respective set of parameters and
2
are the experimentally available time points. The
cost function is minimized using a stochastic optimization algorithm (a Covariance Matrix
Adaptation Evolution Strategy using the cma-es Python package (Hansen & Auger, 2014). All fit
parameters are found in Supplemental Table S2.3.
Model of cell fate dynamics within a small stem cell niche (Long Term - Nestin
#
-NSC)
43
In this model, NSCs divide within a small niche region that can harbor a maximum of
NSCs. NSCs stochastically divide at a rate . If the niche is not maximally occupied, NSCs
duplicate with probability and differentiate with probability 1− upon which NSCs are lost
while generating a certain amount of neuronal and glial progeny. As in the model of the
developmental-like programme, the amount of progeny generated (interpreted as containing all
cell types down the neuronal or glial lineage, respectively) is exponentially distributed with a mean
proliferative output
/
or
C
, respectively. If the niche is maximally occupied, the duplication
channel is blocked, and cells differentiate with probability 1. Neuronal and glial progeny dies at
rates
/
and
C
, respectively. The state of a clone is given by the vector (,,) where is the
number of NSCs, denotes the neuronal content and denotes the glial content. Initially, the
clone starts as a single NSC. The dynamics of the probability =(,,,) to find a clone with
configuration (,,) at time is described by the master equation
(5)
where the operators
A
and
A
being defined analogously to Eqs. (3) and (4) with
C
given by
/
with
±
being replaced by the ladder operator
±
for the glial progeny number . Here we impose
the boundary conditions (−1,)=(+1,)=0, which also ensure that the niche occupation
cannot exceed its maximum capacity .
All parameters are determined in the same way described in the previous section for the
developmental-like program, except that only the cycle rate is refitted for later ages since the
first division occurs at the identical rate in the niche-like model.
44
Statistical analysis
Statistics were performed as indicated in each Figure legends. For in vivo experiments, the
number of independent experimental replicates are indicated in figure legends, with n representing
experimental replicates (clones). For the single cell RNA-seq experiment, 2-3 mice per group were
used to obtain enough cells as one biological replicate. Mann-Whitney U-test was conducted to
compare the Imatinib effect against Vehicle controls. All multiple comparisons were performed
with GraphPad Prism’s one-way ANOVA function with Bonferroni’s multiple comparisons test.
All clones observed at each time point were treated as statistically equivalent. No randomization
or blinding was used in the animal studies. Mouse dentate gyri with exceedingly low (<6) or high
(>24) clones per color (Confetti) per hemisphere were omitted from the study. Sample sizes were
estimated in accordance to prior clonal studies (Bonaguidi et al., 2011; Song et al., 2012). Error
bars in the study represent the standard errors in mean frequencies, calculated as Z(1−)/
where p is the frequency of a given characteristic and n the number of clones considered.
Mathematical modeling was performed by weighted least squares using custom-written MATLAB
scripts. Representative images are depicted in Figure S2.2.
Data and code availability
The accession number for the raw RNA sequencing data reported in this paper is GEO:
GSE168031. The codes to implement (i) scRNA-seq analysis
(https://github.com/bonaguidilab/NSC_aging.git) and (ii) computational modeling
https://github.com/BenSimonsLab/Ibrayeva_Cell-Stem-Cell_2021 are available on GitHub.
Supplemental tables S2.5-S2.8 can be found on Mendeley at
https://data.mendeley.com/datasets/yskfbt8ssk/draft?a=c74843a5-e5d6-47f8-b2d4-bd07db9f1d85
45
Chapter III: Different States and Discrete Populations of Adult Neural
Stem Cells
Abstract
Individual stem cells exhibit hallmark properties of repeated self-renewal and
differentiation into specialized progeny. Complexities observed in dynamic somatic stem cell
behavior have led to uncertainty regarding their identities, characteristics, and relationships. Clonal
lineage-tracing has identified a population of quiescent radial glia-like neural stem cells (RGLs)
marked by Nestin-CreER
T2
(Nestin
#
) in the adult mouse hippocampus. In addition, recent studies
that utilize live imaging showed differences in a stem cell behavior. Here we show that individual
RGLs marked by Gli1-CreER
T2
(Gli1
#
) and by Ascl1-CreER
T2
(Ascl1
#
) exhibit adult neural stem
cell characteristics but differ in cell cycle and self-renewal properties. Long-term clonal lineage-
tracing and computational analyses show that Gli1
#
marks multipotent Nestin
#
-RGLs at the pre-
activation state, whereas Ascl1
#
-RGLs represent a distinct, neuronal fate-biased population.
Interestingly, Gli1
#
and Nestin
#
-marked RGLs acquire Ascl1
#
-RGL-like properties upon injury.
Our study resolves stem cell heterogeneity into both different states of the same population and
discrete populations that co-exist in the same somatic tissue.
46
Introduction
Adult somatic stem cells contribute to ongoing tissue maintenance and repair after injury
(Pilz et al., 2018; Weissman, Anderson, & Gage, 2001). Individual stem cells can give rise to one
(unipotential, such as muscle stem cells) or many lineages (multipotential, such as hematopoietic
stem cells) and their division mode may be constitutively biased or stochastic (Gebara et al., 2016b;
Klein & Simons, 2011; Pilz et al., 2018). Identifying adult stem cells in their native environment,
elucidating characteristic fate behaviors, and revealing lineage relationships are fundamental goals
toward understanding tissue homeostasis and plasticity, which has significant implications for
regenerative medicine.
An emerging principle of stem cell biology is the manifestation of heterogeneity within a
given tissue. Apparent diversity may arise from differences in stem cell characteristics such as
proliferation dynamics, lineage bias, self-renewal capacity, or function during tissue maintenance,
aging and repair (Li & Clevers, 2010; Martín-Suárez, Valero, Muro-García, & Encinas, 2019;
Muller-Sieburg, Sieburg, Bernitz, & Cattarossi, 2012; Solanas & Benitah, 2013). In addition,
observations may vary depending upon the technological approach used. These complexities have
resulted in substantial controversy over the identity and properties of stem cell populations in many
somatic compartments, including hematopoietic, intestinal, epithelial and neural systems
(Bonaguidi et al., 2011; Buczacki et al., 2013; Ema, Morita, & Suda, 2014; Encinas et al., 2011;
Mascre et al., 2012). At the apex of the stem cell compartment, “dormant” cells are thought to be
primed for survival while more active, but long-lived and slow-cycling, cells are biased towards
differentiation and loss, as observed during hematopoiesis (van der Wath, Wilson, Laurenti,
Trumpp, & Lio, 2009) and in muscle (Rocheteau, Gayraud-Morel, Siegl-Cachedenier, Blasco, &
Tajbakhsh, 2012). Yet, recent lineage-tracing studies based on genetic labeling and live-imaging
47
of the intestinal crypt (Ritsma et al., 2014), hair follicle (Rompolas, Mesa, & Greco, 2013),
germline (Hara et al., 2014) and brain (Pilz et al., 2018) have provided evidence that stem cells
may vacillate between these compartments, adjusting their bias and changing levels of gene
expression in response to signals from the niche. As a result, different, but transitory, states can be
observed in a single precursor population and confound interpretation of distinct stem cell
identities. Approaches designed to account for temporal heterogeneity may distinguish discrete
stem cell populations from different states of the same population and provide a general framework
to help resolve stem cell ambiguity.
Single-cell analysis with computational reconstruction has become the method of choice
for probing cellular heterogeneity (Blanpain & Simons, 2013; Etzrodt, Endele, & Schroeder, 2014;
Kretzschmar & Watt, 2012). In the adult nervous system, clonal lineage-tracing of quiescent radial
glia-like precursors (RGLs) has demonstrated the presence of self-renewing, multipotent neural
stem cells (NSCs) within the subgranular zone (SGZ) in the dentate gyrus (Bonaguidi et al., 2011;
Ibrayeva et al., 2019) , and fast-dividing NSCs with limited renewal in the subventricular zone
along the lateral ventricles (Calzolari et al., 2015). Yet, evidence from population studies also
points to the presence of multiple neural precursors in the adult dentate gyrus with different fate
biases (D. A. Berg et al., 2019; Bonaguidi et al., 2016; Bonaguidi et al., 2011; Encinas et al., 2011;
Steiner et al., 2004). In addition, structural and phenotypic heterogeneity exists in response to
physiological changes and injury (DeCarolis et al., 2013; Gebara et al., 2016). Whether these
observed differences reflect discrete neural stem cell populations remains unknown.
To deconstruct the complexity of stem cell behavior, we developed genetic marking
strategies for clonal lineage-tracing of individual RGLs in the adult mouse dentate gyrus using
multiple Cre-drivers. We performed time-course experiments and computational analyses to
48
account for temporal heterogeneity and quantify fundamental precursor properties of RGLs labeled
by different approaches. We further explored the plasticity of RGL properties upon injury. Our
study revealed co-existence of neural stem cells with different characteristics in the adult
mammalian brain and provides novel insight into endogenous somatic stem cell diversity.
Results
Labeling RGLs in different activation states
Previous studies have used different mouse lines, including Nestin-CreER
T2
, Gli1-
CreER
T2
, and Ascl1-CreER
T2
, to fate-map RGLs at the population level in the adult dentate gyrus
(Ahn & Joyner, 2005; Bonaguidi, Song, Ming, & Song, 2012; Dranovsky et al., 2011; Encinas et
al., 2011; Kim, Ables, Dickel, Eisch, & Johnson, 2011; Pilz et al., 2018) (Figure 3.1A). To
investigate functional properties of RGLs labeled by different approach, we developed single-cell
lineage-tracing approaches using Gli1-CreER
T2
and Ascl1-CreER
T2
following our previous
strategy for Nestin-CreER
T2
-based clonal analysis (Bonaguidi et al., 2011). For each CreER
T2
driver, we tested different tamoxifen doses and reporter lines, including Z/EG(Novak, Guo, Yang,
Nagy, & Lobe, 2000), mT/mG(Muzumdar, Tasic, Miyamichi, Li, & Luo, 2007), and Rosa-
YFP(Srinivas et al., 2001), and obtained combinations that exhibited high specificity, inducibility
and reproducibility (Extended Data Figure 3.1A-B and Extended Data Table 3.1). At one day post-
tamoxifen induction (dpi), over 90% of labeled precursor cells within the adult dentate gyrus of
each line were Nestin
+
GFAP
+
RGLs with soma located within the SGZ and basal branches
extending through the granule cell layer into the molecular layer (n = 3; Figure 3.1B and Extended
Data Fig. 3.1c). Approximately 6-18 precursors were labeled per dentate gyrus at 1 dpi and the
number of cell clusters remained generally constant over the period of analysis, whereas the clonal
49
size increased over time (Figure S3.1D-E). Previous computational simulations have suggested
over 95% probability of clonality with this initial labeling density in the adult dentate gyrus
(Bonaguidi et al., 2011).
Astroglial lineage
(TA/Astro)
RGL (R)
Gli1::CreER
Nestin::CreER
Ascl1::CreER
0
20
40
60
80
100
RGLs among
GFP
+
precursors (%)
Gli1
#
Ascl1
#
Gli1
#
-RGL
activation (%)
Days post induction (dpi)
Ascl1
#
-RGL
activiation (%)
Days post induction (dpi)
A
B
C
D
Quiescent
Quiescent Activated
Activated
GFP
DAPI
GFP
DAPI
GFP
DAPI
GFP
DAPI
Gli1-CreER::mT/mG Ascl1-CreER::mT/mG
Neuronal lineage
(IPC/IN/MN)
(A) (N)
0246 8
0
20
40
60
*
*
RGL cell cycle
re-entry (%)
Time (dpi)
Gli1
#
Ascl1
#
0
20
40
60
80
100
*
MCM2
+
RGLs (%)
Gli1
#
Ascl1
#
G
I
GFP MCM2
GFAP
GFP MCM2
GFAP
R
(R) N
RN
GFP GFAP
DAPI
Gli1-CreER::Z/EG Ascl1-CreER::EYFP
Ascl1-CreER::EYFP
3 dpi
0246 8
0
20
40
60
80
100
02 46 8
0
20
40
60
80
100
E
F
H
K J
50
Figure 3.1. Clonally labeled RGLs in the adult dentate gyrus using two CreER
T2
lines
exhibit different cell cycle properties.
(A) Schematic illustration of neural precursor lineage relationships within the adult hippocampus.
Colored bars indicate approaches used in this study to lineage trace Nestin
#
-RGLs, Gli1
#
-RGLs,
and Ascl1
#
-RGLs at the clonal level. A: astroglial lineage; TA: transition astrocyte; Astro:
Astrocyte; N: neuronal lineage; IPC: intermediate progenitor cell; IN: immature neuron; MN:
mature neuron. (B) Quantification of the percentage of GFP
+
RGL-containing clones in the dentate
gyrus at 1 dpi. Values represent mean ± SEM. (n = 4 or 8 dentate gyri). (C-D) Sample confocal
images of GFP-labeled quiescent and activated RGL clones indicated by the presence of more than
one DAPI
+
nuclei in Gli1
#
-RGL (C) and Ascl1
#
-RGL (D). Scale bars, 10 µm (5 µm for inserts).
(E-F) Time course of Gli1
#
-RGL and Ascl1
#
-RGL activation following clonal labeling. RGL
activation was scored based on the presence of immediate adjacent progeny. Values represent
mean ± SEM (n = 5 - 9 dentate gyri). Red lines represent best fit. (G) Confocal images of Ascl1
#
-
RGL clones with multiple divisions within 7 days. Shown are examples of an Ascl1
#
-RGL with a
neuronal R-N division (1) after a symmetric R-R division (2; A) and an Ascl1
#
-RGL with multiple
neuronal divisions (B). Orthogonal views are also shown. Scale bars, 10 µm (5 µm for inserts).
(H) Quantification of cell cycle re-entry based on lineage analysis at 1, 3 and 7 dpi. Values
represent mean ± SEM (n = 5-12 dentate gyri for each time point; *p < 0.01; Student’s t-test). (I-
K) Assessment of cell cycle re-entry of Ascl1
#
-RGL and Gli1
#
-RGL clones at 3 dpi with MCM2
immunohistology. Shown are sample confocal images of a quiescent MCM2
-
Gli1
#
-RGL (I) and
an active MCM2
+
Ascl1
#
-RGL (J). Also shown is the quantification of MCM2
+
RGL clones at 3
dpi (K). Values represent mean ± SEM (n = 3-4 dentate gyri; *p < 0.01; Student’s t-test). See also
Figure S3.1 and S3.2.
We first examined the activation of labeled RGLs based on short-term lineage analysis. By
7 dpi nearly all Gli1
#
-RGL and Ascl1
#
-RGL clones already contained adjacent progeny, indicative
of RGL activation (Figure 3.1C-D). This result is in sharp contrast to previous findings of Nestin
#
-
RGLs, few of which divide within 7 dpi (Song et al., 2012). Time-course analysis showed that
most Gli1
#
-RGL clones contained a single RGL without any progeny at 0.5 dpi, the earliest time
point to detect sufficient GFP expression, and nearly all produced progeny by 3 dpi (Figure 3.1C).
In contrast, Ascl1
#
-RGL clones already contained progeny at 1 dpi (Figure 3.1D), consistent with
a recent population-level fate-mapping study(Andersen et al., 2014). To estimate the cell cycle
status at the time of labeling, we performed immunohistological analysis of RGLs with Cre
antibodies. Consistent with the short-term lineage result, 18 ± 2% and 70 ± 1% of Cre
+
RGLs were
MCM2
+
in the dentate gyrus of adult Gli1-CreER
T2
and Ascl1-CreER
T2
mice, respectively (Figure
51
S3.2). These results showed that, in contrast to Nestin
#
-RGLs labeled in a quiescent state
(Bonaguidi et al., 2011), Gli1
#
-RGLs were labeled in a pre-activation state, whereas Ascl1
#
-RGLs
were labeled in an active state in the adult mouse dentate gyrus. Therefore, our genetic labeling
strategies capture different cell cycle states of adult neural precursors in vivo.
Divergence in Cell Cycle Exit after Initial RGL Activation
Upon division, RGLs can re-enter cell cycle or return to quiescence (Bonaguidi et al., 2011;
Encinas et al., 2011). Birth-dating the initial division permits high resolution in vivo analysis of
cell cycle exit and re-entry. Quantitative short-term lineage analysis revealed marked differences
in the number of cell divisions by Gli1
#
-RGLs and Ascl1
#
-RGLs within 7 days post-induction
(Figures 3.2A-B and S3.2B-C). While cell cycle re-entry of Gli1
#
-RGLs was consistently low from
1 dpi (7 + 7%) to 7 dpi (13 + 5%), this fraction of Ascl1
#
-RGLs increased from 21 ± 4% at 1 dpi
to 55 ± 8% at 7 dpi (Figure 3.2C). Additionally, three rounds of RGL divisions could be observed
in some Ascl1
#
-RGL clones during this period, but not in any Gli1
#
-RGL clones (Figure S3.2B-
C). To validate results for cell cycle re-entry, we examined MCM2 expression in activated RGL
clones at 3 dpi. Indeed, a significantly higher percentage of Ascl1
#
-RGLs that already had progeny
remained MCM2
+
compared to Gli1
#
-RGLs (Figure 3.1D-F). Notably, for both Gli1
#
-RGLs and
Ascl1
#
-RGLs, percentages of MCM2
+
RGLs at 3 dpi were consistent with the fraction of RGLs
that had divided by 7 dpi based on lineage-tracing (Figure 3.1C-F). These results confirmed that
RGLs prospectively identified using MCM2 would divide as assessed by retrospective lineage-
analysis. Together, these two independent approaches show that Gli1
#
-RGLs largely return to
quiescence after activation, whereas Ascl1
#
-RGLs exhibit an increased probability of cell cycle re-
entry, indicative of a more proliferative precursor population beyond the initial cell division.
52
Fate Choice Bias of Different RGLs
Stem cells are characterized according to their capacity of lineage generation (Weissman
et al., 2001). Nestin
#
-RGLs in the adult dentate gyrus have previously been shown to undergo three
types of self-renewing divisions: symmetric division generating two RGLs, asymmetric
neurogenic division producing a GFAP
-
IPC of the neuronal lineage, and asymmetric astrogenic
division resulting in a GFAP
+
bushy astroglia. One major limitation of the previous Nestin-
CreER
T2
-based clonal analysis is a lack of precise birth-dating of quiescent neural stem cells. Since
Gli1
#
-RGLs and Ascl1
#
-RGLs both divide shortly after labeling (Figure 3.1E-F), our current
approach allows for high-resolution direct quantification of fate decisions of adult neural precursor
cells in vivo. In addition, fate choices over multiple divisions within a clone can be deciphered
based on the type of progeny and distance from parent RGL (See Materials and Methods). We
therefore constructed lineage trees for RGL-containing clones from short-term lineage-tracing,
classified according to the progeny generated and the number of divisions the RGL underwent
(Figure S3.3A). Similar to Nestin
#
-RGLs (Bonaguidi et al., 2011), Gli1
#
-RGLs exhibited three
modes of self-renewal division (Figure S3.3A): symmetric (Figure 3.2A), asymmetric neurogenic
(Figure 3.2B), and asymmetric astrogenic (Figure 3.2C). Quantification of 116 Gli1
#
-RGL clones
showed 29 + 4% symmetric, 45 + 6% neurogenic and 26 + 4% astrogenic divisions (Figure 3.2E).
In contrast, quantification of 163 Ascl1
#
-RGL clones indicated mostly neurogenic divisions (76 +
3%) with much less symmetric (18 + 3%) and astrogenic (6 + 2%) divisions (Figure 3.2D-E). We
obtained similar results with two different reporters for both Gli1
#
-RGLs and Ascl1
#
-RGLs (Figure
S3.2D-E). Together, these results indicate that Gli1
#
-RGLs generate multiple neural lineages,
whereas Ascl1
#
-RGLs are largely neurogenic, revealing additional differences between these two
RGL populations.
53
Figure 3.2. Fate specification of Gli1
#
-RGLs and Ascl1
#
-RGLs
(A-D) Sample confocal images of representative division modes by Gli1
#
-RGLs, including
symmetric RGL-RGL division (R-R; A), asymmetric neurogenic RGL-IPC division (R-N; B), and
asymmetric astrogenic RGL-astrocyte division (R-A; C), and the typical neurogenic RGL-IPC
division mode in an Ascl1
#
-RGL clone (D). Single confocal section images in inserts highlight
newly generated progeny. Orthogonal views are also shown to reveal co-localization of two
immunostaining signals. Scale bars, 5 µm. (E) Quantification of fate choice made by Gli1
#
-RGLs
and Ascl1
#
-RGLs during 7 days after labeling (combined data from 1, 3, 7 dpi). Values represent
mean ± SEM (n = 18 or 24 dentate gyri; *p < 0.05, **p < 0.01; Student’s t-test). Also see Figure
S3.2.
Maintenance of Distinct Stem Cell Properties
In addition to lineage generation, bona fide stem cells are defined by their capacity for self-
renewal without differentiating for an extended period of time (Gage, 2000). Whether RGLs are
maintained after division is controversial, as it has been suggested that RGL cell cycle entry is
coupled to their depletion (Encinas et al., 2011). To directly examine the maintenance of RGLs
after division, we quantified Gli1
#
-RGL and Ascl1
#
-RGL clones at 30 and 60 dpi (Figure 3.3A-B),
a sufficient duration to probe the “division-coupled differentiation” model (Encinas et al., 2011).
While some depletion gradually occurred over time, more than 60% of Gli1
#
-RGL and Ascl1
#
-
RGL-RGL
GFP DAPI
GFAP
RGL-IPC
GFP DAPI
GFAP
RGL-Astro
GFP DAPI
GFAP
RGL-IPC
GFP DAPI
GFAP
Gli1-CreER::mT/mG Gli1-CreER::mT/mG
Gli1-CreER::mT/mG Ascl1-CreER::EYFP
Gli1
#
(18)
Ascl1
#
(24)
A B
C D
E
Fate choice of RGL divisions (%)
R-R R-N R-A
0
25
50
75
100
*
**
**
54
RGL clones retained at least one RGL after the initial RGL division at 30 days and 30% of clones
maintained RGLs at 60 days (Figure 3.3C). Similar to Nestin
#
-RGL clones (Bonaguidi et al.,
2011), some Gli1
#
-RGL clones contained RGL(s) together with cells of both neuronal and
astrocytic lineages, demonstrating RGL maintenance and multipotentiality at the single-cell level
(Figure 3.3A). Meanwhile, most RGL-maintaining Ascl1
#
clones contained cells of the neuronal
lineage at various developmental stages, indicating multiple rounds of neurogenic self-renewal at
the single-cell level (Figure 3.3B). Therefore, RGL clonal tracing results from three independent
lines are inconsistent with the model of invariant coupling of NSC division to its differentiation.
To further examine properties of labeled RGLs over time, we constructed lineage trees
from RGL-containing clones at 30 dpi according to cell type composition (Bonaguidi et al., 2011).
Consistent with enhanced cell cycle re-entry at earlier time points, Ascl1
#
-RGLs divided more
frequently over the 30-day period (Figure S3.3B). Compared to Ascl1
#
-RGLs, Gli1
#
-RGL clones
displayed greater diversity of progeny with significantly higher percentages of multipotent
differentiation (21 ± 10% vs. 1 ± 1%), astrocyte generation (26 ± 8% vs. 6 ± 4%) and RGL
amplification (22 ± 8% vs. 6 ± 3%; Figure 3.3D). On the other hand, Ascl1
#
-RGL clones were
significantly more neurogenic than Gli1
#
-RGL clones (86 ± 5% vs. 31 ± 8%; Figure 3.3D).
Importantly, similar frequencies of self-renewal modes were observed at 3 and 30 dpi for Gli1
#
-
RGLs, while Ascl1
#
-RGLs maintained a bias towards neuronal generation (Figure 3.3E),
indicating that RGLs labeled by these two approaches represent long-term distinct stem cell
populations under physiological conditions. Together with additional long-term tracing
confirmation at 60 dpi (Figure S3.3C), these results support the model that Gli1
#
-RGLs typify
multipotent NSCs, whereas Ascl1
#
-RGLs represent a distinct population of neuronal fate-biased
NSCs.
55
Figure 3.3. Maintenance of stem cell properties by multipotent Gli1
#
-RGLs and neuronal
fate biased Ascl1
#
-RGLs.
(A) Sample confocal projection and single-section images of a multipotential Gli1
#
-RGL clone at
30 dpi. It contained two RGLs (via R-R division), with one generating IPCs of the neuronal lineage
(1; R-N division) and another producing an astroglia (2; R-A division), thus exhibiting 3 different
modes of self-renewal. Scale bars, 5 µm. (B) Sample confocal projection and single-section images
of an Ascl1
#
-RGL clone that had undergone repeated neuronal divisions. Cell progeny from
different developmental states include mature neurons (4), immature neurons (3), Tbr2
+
IPCs (2),
and the most recent IPC division from the RGL (1). Scale bars, 5 µm. (C) Quantification of RGL
maintenance by the fraction of clones containing at least one RGL. Values represent mean ± SEM
(n = 4-10 dentate gyri). (D) Quantification of the frequency of clone composition types among
RGL-retaining clones at 30 dpi for Ascl1
#
-RGLs and Gli1
#
-RGLs. Values represent mean ± SEM
(n = 5 or 8 dentate gyri; **p < 0.01; *p < 0.05; Student’s t-test). (E) Quantitative comparison of
self-renewal fate choices made by Ascl1
#
-RGLs and Gli1
#
-RGLs at 3 dpi and over 30 days (30
dpi). Values represent mean ± SEM (n = 5 or 8 dentate gyri; *p < 0.05; n.s: p > 0.1; Student’s t-
test). Also see Figure S3.3.
R
R IPC IPCs INs MN
1
2
3
4
1 2 3 4
R+R R+N R+A R+N+A
0
20
40
60
80
100
R-R R-N R-A
0
20
40
60
80
100
Clone composition among
all maintained clones (%)
Fate choices of
RGL divisions (%)
Gli1
#
:30 dpi (5)
Ascl1
#
:30 dpi (8)
Gli1
#
:30 dpi
Ascl1
#
:1-3 dpi
Gli1
#
:1-3 dpi
Ascl1
#
:30 dpi
A
B
C D E
RGL-IPCs
1
1
2
DAPI GFAP
GFP
RGL-A 2
DAPI GFAP
GFP
Gli1-CreER::Z/EG
Ascl1-CreER::EYFP
GFP GFAP GFP GFP
Tbr2
Immature
neuron
Mature
neuron
IPC RGL
*
**
* *
n.s.
*
*
*
*
n.s.
n.s.
*
n.s.
R
(R) (R)
RNRA
RGL maintanence
(% clones)
Days post injection (dpi)
Gli1
#
Ascl1
#
010 20 30 40 50 60
0
20
40
60
80
100
56
Computational Assessment of Different States and Discrete Stem Cell Populations
To further assess our model, we employed a computational approach that accounts for
differences in activation kinetics (Clayton et al., 2007; Klein & Simons, 2011) in order to
quantitatively describe the behavior of different RGLs over an extended period of time (Figure
3.4A and S3.4-3.5; See Materials and Methods). Raw clonal lineage-tracing data from different
time points were analyzed to infer the initial activation time upon labeling (Tentry; Figure S3.4A-
C), the probability of cell fate choices (P; Figure S3.4D-F) and kinetics of subsequent cell cycle
re-entry (Tc; Figure S3.4G-I and S3.5; See Computational Modeling). Our analyses indicated that,
upon labeling, Gli1
#
-RGLs and Ascl1
#
-RGLs rapidly entered cell cycle with an average Tentry of
0.78 ± 0.02 and 0.35 ± 0.04 days, respectively, while that for Nestin
#
-RGLs was 23 ± 3 days
(Figure 3.4B and S3.5). Despite the initial activation time offset, Nestin
#
-RGLs and Gli1
#
-RGLs
exhibited similar Tc, which was significantly longer than that of Ascl1
#
-RGLs (Figure 3.4C).
Furthermore, Nestin
#
-RGLs and Gli1
#
-RGLs displayed very similar probabilities of the various
self-renewal modes, which were also significantly different from those of Ascl1
#
-RGLs (Figure
3.4D). Altogether, our results provide quantitative descriptions of different adult NSC life cycles
in vivo (Figure 3.4F).
We next challenged the computational assessment by comparing the average cell content
of RGL-containing clones labeled by three different approaches as a function of time. To
compensate for the initial activation time offset, the average content per RGL-containing clone for
Nestin
#
-RGLs at 30, 60 and 120 dpi were compared to Gli1
#
-RGLs and Ascl1
#
-RGLs at 7, 30, and
60 dpi, respectively. Strikingly, the composition of Nestin
#
-RGL and Gli1
#
-RGL clones mirrored
one another over time (Figure 3.4E). In contrast, Ascl1
#
-RGL clones contained more cells of the
neuronal lineage and fewer cells of the astroglial lineage (Figure 3.4E). These results further
57
support our model that Gli1
#
marks the same stochastic, multipotent quiescent RGL stem cells as
Nestin
#
but in a pre-activation state, whereas Ascl1
#
labels a discrete neuronal fate biased RGL
population (Figure 3.4F).
Figure 3.4. Different states and distinct populations of stem cells revealed by a
computational approach.
(A-D) Neural stem cell activation kinetics and basic characteristics derived from computational
analysis of Nestin
#
-RGL, Gli1
#
-RGL and Ascl1
#
-RGL clonal lineage-tracing data. Shown is a
schematic depicting the life cycle of adult hippocampal RGL neural stem cells with following key
parameters: time to first division upon initial labeling (Tentry), cell cycle re-entry time (Tc), and fate
choices probabilities including astrocytic (PRA), symmetric (PRR), and neuronal (PRN) self-renewal
modes (A). Also shown are summaries of mean Tentry (B), TC (C) and fate choices (D) as inferred
from computational modeling and fit to experimental data between 1-60 dpi. Values represent
mean ± SEM (n = 82-161 clones for B-C; n = 59-156 clones for D; *p < 0.05; **p < 0.1; Student’s
t-test). (E) Statistical ensembles of RGL clones representing the average content of different
G
0
G
1
M
G
1
M
0.1
1
10
100
1000
**
**
**
PRR PRN PRA
0
20
40
60
80
100
*
**
**
**
**
Fate choice (%)
1
10
100
1000
**
**
Nestin
#
Gli1
#
Ascl1
#
Nestin
#
Gli1
#
Ascl1
#
Nestin
#
Gli1
#
Ascl1
#
Cycle
re-entry
First
division
Fate choice
Time to first RGL division
(Tentry, day, log
10
)
Cell cycle re-entry time
(TC, day, log
10
)
A B C D
E
Average content
per RGL-containing clone
Average content
per RGL-containing clone
Average content
per RGL-containing clone
RGL Neuronal Astroglial
Nestin
#
Gli1
#
Ascl1
#
Time (dpi) Time (dpi) Time (dpi)
Tentry
Tc
PRA
PRR
PRN
R
R
A
R
IPC
IN
020 40 60 80 100120
0
2
4
6
010 20 30 40 50 60
0
2
4
6
010 20 30 40 50 60
0
2
4
6
E
F
58
progeny types per clone as a function of time. The neuronal lineage includes IPCs, immature
neurons (IN) and mature neurons (MN), and the astroglial lineage includes transition astroglia
(TA) and astrocytes (A). Values represent mean ± SEM (n = 102-225 clones). (F) Cartoon
summary depicting different states and discrete populations of RGL neural stem cell populations
in the adult mouse hippocampus during physiological condition. Also see Figure S3.4 and S3.5.
Plasticity of RGLs upon Injury
As both lineage-tracing and computational analyses indicated that Gli1
#
-RGLs/Nestin
#
-
RGLs and Ascl1
#
-RGLs exist as distinct stem cell populations under physiological conditions, we
next examined whether their fate behaviors are fixed or could be influenced by extrinsic
environmental cues in the adult neurogenic niche. Stem and progenitor cells in various somatic
compartments can be differentially utilized under physiological or pathological conditions (Li &
Clevers, 2010). In the adult brain, quiescent precursor cells are activated to restore depleted IPCs
and neuroblasts after Cytosine Arabinofuranoside (AraC) treatment, a chemotherapeutic agent that
kills dividing cells (Ahn & Joyner, 2005; Doetsch, Caille, Lim, Garcia-Verdugo, & Alvarez-
Buylla, 1999). It remains unknown, however, whether and how quiescent RGL behavior changes
during injury-induced regeneration. We labeled Gli1
#
-RGLs at the clonal level (Table S3.1),
waited for the initial division to complete to avoid killing them, and then infused AraC for 4 days
(Figure S3.5A). EdU was injected during the AraC administration period to monitor efficacy of
proliferating cell depletion (Figure S3.5B-C). After the 7-day chase, the AraC group had no EdU
+
cells within the SGZ (Figure S3.5B). GFAP immunohistology further revealed reactive gliosis
after AraC treatment within the dentate gyrus, demonstrating a local injury response (Figure
S3.5C).
59
Figure 3.5. Plasticity of RGL fate after AraC-induced injury
(A-B) Sample confocal projection and single-section images of Gli1
#
-RGL clone (A) and Ascl1
#
-
RGL clone (B) 7 days after the end of AraC treatment. Shown are examples of the Gli1
#
-RGL that
generated multiple cells of the neuronal lineage (A) and the Ascl1
#
-RGL that underwent symmetric
cell division (B). Lineage trees indicating self-renewal modes are shown next to the projection
images. Scale bars, 10 µm (5 µm for inserts). (C-D) Quantification of percentages of RGL clones
that divided (C, left) and that re-entered cell cycle (C, right), and division modes (D) under
different conditions. Values represent mean ± SEM (n = 5-6; *p < 0.01, **p < 0.001, n.s : p > 0.1;
Student’s t-test). (E) Cartoon summary depicting different states and discrete populations of RGL
neural stem cell populations in the adult mouse hippocampus upon injury. Nestin
#
and Gli1
#
mark
the same population of self-renewing, multipotent and quiescent RGL neural stem cells (NSCs) in
distinct states. Nestin
#
labels RGLs during a more quiescent state, while Gli1
#
marks RGLs in a
pre-activation state, hours before division. Ascl1
#
labels a distinct population of self-renewing,
neuronal fate-biased and more proliferative RGL neural stem cells in the same region of the
hippocampus under physiological conditions. Note the duration until first RGL division after
labeling (Tentry) for lineage-tracing and the duration of cell cycle re-entry (TC). Upon AraC injury,
Gli1
#
-RGLs and Nestin
#
-RGLs behave similarly to Ascl1
#
-RGLs under physiological conditions,
A B
C D
Percentage of RGL
activation
Percentage of RGL
cycle re-entry
R
(R) N
RN
**
1
2
RGL IPC
1
2
RGL RGL
Gli1-CreER::mT/mG Ascl1-CreER::mT/mG
R
(R)
R
(R)
RR R
n.s.
**
n.s.
Gli1
#
;-AraC
Gli1
#
;+AraC
*
0
20
40
60
80
100
20
40
60
80
100
n.s.
*
Ascl1
#
;-AraC
Ascl1
#
;+AraC
Fraction of divisions
R-R R-N R-A
*
1
2
1 2
Gli1
#
;-AraC
Gli1
#
;+AraC
Ascl1
#
;-AraC
Ascl1
#
;+AraC
n.s.
n.s.
n.s.
*
*
0.0
0.2
0.4
0.6
0.8
1.0
n.s.
n.s.
E
60
whereas Ascl1
#
-RGLs also change their properties to primarily symmetric division. Strength of
arrows denotes relative probability of fate choices. See also Figure S3.6 and S3.7.
Upon confirming the AraC model, we next analyzed RGL-containing clones 7 days after
injury. Remarkably, nearly all Gli1
#
-RGLs became activated in response to AraC injury and
exhibited repeated self-renewal and IPC generation, resembling Ascl1
#
-RGLs under physiological
conditions (Figure 3.5A-C). To define Gli1
#
-RGL behavior in response to injury more precisely,
we again constructed lineage trees (Figure S3.6D). The frequency of Gli1
#
-RGL cell cycle re-entry
after injury was significantly increased to a level similar to Ascl1
#
-RGLs under basal conditions
(Figure 3.5C). Quantification of self-renewal modes indicated that neurogenic divisions of Gli1
#
-
RGLs increased in response to injury, with a corresponding decrease in astrogenesis (Figure 3.5D
and S3.6E). Therefore, after AraC injury, Gli1
#
-RGLs phenocopy basal characteristics of Ascl1
#
-
RGLs. In addition, Nestin
#
-RGLs after injury exhibited similar properties to Gli1
#
-RGLs after
injury and Ascl1
#
-RGLs under basal conditions (Figure S3.7). These results provide additional
support for our model that Nestin
#
-RGLs and Gli1
#
-RGLs represent the same stem cell population,
and further demonstrate that the stochastic behavior of RGLs can be altered in response to changes
in the environment.
Finally, we examined Ascl1
#
-RGLs using the same injury protocol. The number of labeled
RGL-containing clones was significantly reduced immediately following AraC treatment (Figure
S3.6G), suggesting that many proliferative Ascl1
#
-RGLs were depleted upon this injury. We
followed the fate of remaining Ascl1
#
-RGLs at 7 dpi and found that these RGLs exhibited
increased symmetric cell division at the expense of neurogenic self-renewal (Figures 3.5B and
3.5D). Therefore, the properties of Ascl1
#
-RGLs upon injury differ from those under basal
conditions and are also distinct from the injury response of Nestin
#
-RGLs and Gli1
#
-RGLs.
61
Together, these results revealed remarkable plasticity of different RGL populations and identified
their differential contribution to cell genesis under basal and injury conditions.
Convergence of discrete populations during physiological aging
Since we observed that properties of the different RGL populations can be changed under
injury conditions suggestive of the remarkable plasticity, next we wanted to look at the NSC
behavior during physiological aging. It has been known for a decade that RGL number and
function decline with age (Encinas et al., 2011; Kuhn et al., 1996; Kuhn et al., 2018; Pilz et al.,
2018). However, it remains unknown, how RGL populations behavior changes with time. To
address this question, we clonally labeled Nestin
#
-RGLs, Gli1
#
-RGLs and Ascl1
#
-RGLs at 12
months of age and followed their behavior after 7, 30 and 60 days.
We found that with time under physiological conditions Nestin
#
-RGLs/Gli1
#
-RGLs
population increase self-renew division at the expense of neurogenic division (Figure 3.6A-B).
More interestingly, quantification of self-renewal modes indicated that symmetric divisions of
Ascl1
#
-RGLs increased at both 7- and 30-days post tamoxifen injections at 12 months of age, with
a corresponding decrease in neurogenic self-renewal and increase in astrogenic division (Figure
3.6C). RGL population behavior are less heterogenous as Ascl1
#
-RGLs more closely resemble
Nestin
#
-RGLs/Gli1
#
-RGLs at 12 months of age (Figure 3.6D). Therefore, we have found that both
RGL populations exhibit distinct changes when compared to 2-month-old mice under
physiological conditions (Figure 3.6D-E). These results support the existence of different RGL
populations and their differential but converging behavior in response to physiological aging.
62
Figure 3.6. Convergence of distinct properties during physiological aging.
(A) Quantification of RGL self-renewing cell fate divisions at 30- and 60-days post-tamoxifen
injection into 2 and 12 months old Nestin
#
mice. RGL = radial glia-like neural stem cells;
A=Astroglial lineage; N=neuronal lineage. Values represent mean ± SEM (*p<0.05, **p<0.01,
***p<0.001, n.s. - not significant; ANOVA with Bonferroni post-hoc test). (B-C) Quantification
of RGL self-renewing cell fate divisions at 7- and 30-days post-tamoxifen injection into 2 and 12
months old Gli1
#
(B) and Ascl1
#
(C) mice. RGL = radial glia-like neural stem cells; A=Astroglial
lineage; N=neuronal lineage. Values represent mean ± SEM (*p<0.05, **p<0.01, ***p<0.001, n.s.
- not significant; ANOVA with Bonferroni post-hoc test). (D) Summary quantification of
percentages of RGL self-renewing cell fate divisions at 30-days post-tamoxifen injection into 2
and 12 months old Nestin
#
, Gli1
#
and Ascl1
#
mice. Values represent mean ± SEM (*p<0.05,
**p<0.01, ***p<0.001, n.s. - not significant; ANOVA with Bonferroni post-hoc test). (E-G)
Sample confocal images of representative symmetric RGL-RGL division by Nestin
#
-RGLs, Gli1
#
-
RGLs and Ascl1
#
-RGLs that RGLs predominantly divide at 12 month of age. (H) Cartoon
summary depicting different states and discrete populations of RGL neural stem cell populations
in the adult mouse hippocampus at 2-month-old. (I) Cartoon summary depicting RGL neural stem
cell populations in the adult mouse hippocampus at 12-month-old.
Discussion
Stem cells often display complex patterns of behavior over time due to their dynamic
nature. Using adult neural stem cells in the hippocampus as an in vivo model system, we combined
clonal lineage-tracing of multiple transgenic lines with computational approaches to resolve stem
A B C
D E F G
R-R R-N R-A
0
20
40
60
80
100
Fate choice of
RGL divisions (%)
Nestin
#
12mo:30dpi (3)
Nestin
#
12mo:30dpi (3)
Nestin
#
2mo:30dpi (6)
Nestin
#
2mo:60dpi (7)
***
***
***
***
n.s n.s
R-R R-N R-A
0
20
40
60
80
100
Fate choice of
RGL divisions (%)
Gli1
#
12mo:7dpi (3)
Gli1
#
12mo:30dpi (4)
Gli1
#
2mo:7dpi (5)
Gli1
#
2mo:30dpi (5)
***
*** ***
***
n.s n.s
R-R R-N R-A
0
20
40
60
80
100
Fate choice of
RGL divisions (%)
Ascl1
#
2mo:30dpi (8)
Ascl1
#
12mo:30dpi (6)
Ascl1
#
2mo:7dpi (7)
Ascl1
#
12mo:7dpi (3)
***
***
***
***
*** ***
H I 12-month-old Multipotential NSC
R+R R+N R+A
0
20
40
60
80
Fate choice of
RGL divisions (%)
Gli1
#
12mo:30dpi (4)
Nestin
#
12mo:30dpi (3)
Ascl1
#
12mo:30dpi (6)
n.s
n.s
n.s
2-month-old Fate biased, neurogenic NSC 2-month-old Stochastic multipotential NSC
63
cell heterogeneity and plasticity. Our study reveals the co-existence of both stochastic multipotent
NSCs, marked in quiescent (Nestin
#
) and pre-activation states (Gli1
#
), and a distinct population of
Ascl1
#
-marked neuronal fate-biased NSCs which appear to act in parallel in young adulthood
under physiological conditions (Pilz et al., 2018) (Figure 3.5E, 3.6H). In addition, we provide
evidence that apparent neural stem cell temporal heterogeneity may reflect the observation of
different states of the same endogenous stem cell population. Finally, we reveal the acute plastic
nature of adult NSCs in response to injury and uncover the age-related changes in behavior during
middle age. These findings not only provide novel insight into the biology of NSCs, their intrinsic
properties and plasticity in the adult mammalian brain, but also suggest an integrative approach
with broad applicability to the general stem cell field.
Transient Stem Cell States and Discrete Stem Cell Populations
Somatic stem cells in many tissues exhibit cycles of activation, fate specification and return
to quiescence, which are difficult to quantitatively assess in vivo (T. Krieger & B. D. Simons,
2015). Commonly used approaches for lineage-tracing and fate-mapping target precursors at
stages of these cycles and provide a snapshot of stem cell behavior. Here, we illustrate that
quantitative determination of stem cell kinetics and clonal evolution over a prolonged period of
time can overcome this limitation by resolving stem cell activation, fate choice, self-renewal and
return to quiescence (Blanpain & Simons, 2013). Specifically, our conclusion that Nestin
#
-RGLs
and Gli1
#
-RGLs in the adult mouse dentate gyrus represent the same stem cell population labeled
in quiescent and pre-activation states, respectively, was drawn from similar cell cycle re-entry
times, cell fate probabilities, kinetics of lineage production and response to injury using two
labeling paradigms (Figures 3.4-3.6 and S3.7). As the time to first cell cycle entry upon labeling
of Gli1
#
-RGLs is significantly shorter than their cell cycle time, our results support the notion that
64
the behavior of stem cells shortly after labeling may not reflect their fundamental properties.
Instead, consistent characteristics over multiple rounds of division more accurately reflect intrinsic
properties of a particular stem cell population.
Our results indicate that long-term analysis of clonal behavior is paramount for properly
interpreting lineage-tracing results. For example, examination of initial proliferation alone could
lead to the misinterpretation that Gli1
#
-RGLs and Ascl1
#
-RGLs represent the same NSC
population, or along with Nestin
#
-RGLs represent three distinct NSCs. Meanwhile, considerations
of the relationship between lineage tracing tools and gene expression they represent influence the
synthesis of results from complementary approaches. For instance, while Mash1 protein enriches
in Ascl1
#
-RGLs compared to Gli1
#
-RGLs, as revealed by Cre staining (Figure S3.1G-H), 75% of
Ascl1
#
-RGLs are Mash1-positive and 15% of Gli1
#
-RGLs are positive. Ascl1 expression oscillates
in embryonic NSCs (Imayoshi et al., 2013), suggesting Mash1 could also be present at various
levels in adult Ascl1
#
- and Gli1
#
-RGLs. Therefore, Ascl1 loss of function or downregulated could
regulate neurogenic behavior in both NSC populations (Andersen et al., 2014; Pilz et al., 2018;
Urban et al., 2016). These results suggest an imperfect correlation between regulatory regions used
during our lineage tracing and their corresponding gene expression levels following clonal
induction. Instead, our study demonstrates the utility of clonal analysis tools and computational
reconstruction as a general strategy to define stem cell identity and relationships by accounting for
potential differences in states of labeled precursors.
Potential Neural Stem Cell Relationships in the Adult Hippocampus
A common theme of constitutive cell genesis is the hierarchical organization of stem and
progenitor cell populations in vivo (Buczacki et al., 2013; Codega et al., 2014; DeCarolis et al.,
2013; Dranovsky et al., 2011; Ema et al., 2014; Encinas et al., 2011; Mascre et al., 2012; Merkle,
65
Mirzadeh, & Alvarez-Buylla, 2007; Mich et al., 2014; Pilz et al., 2018). However, while a single
precursor cell type may function under both physiological and pathological conditions, many
tissues recruit a distinct second precursor type upon injury (Blanpain & Fuchs, 2014). This
“reserve” cell population contributes to repair but may lack stem cell properties during normal
tissue homeostasis (Bottes et al., 2021; Collins et al., 2005; Leung, Coulombe, & Reed, 2007;
Mascre et al., 2012; Stange et al., 2013). In contrast, our study reveals ongoing contributions to
cell genesis from two parallel adult NSC populations – under both physiological and injury
conditions during young adulthood - that exhibit similar morphology, share certain common
lineages, and do not compartmentalize along spatial boundaries. Our study raises the question of
whether similar scenarios occur in other somatic tissues with ongoing cell genesis.
The distinct contributions of two stem cell populations under physiological conditions
functionally enables cellular diversity alongside the robust production of a specific cell type.
Ascl1
#
-RGLs, which repeatedly undergo asymmetric neurogenic self-renewing divisions, may act
as the primary driver of constitutive adult hippocampal neurogenesis. On the other hand, stochastic
neural stem cells (Nestin
#
/Gli1
#
-RGLs) may contribute less frequently to basal tissue maintenance
but act as a reserve for injury and diversify cell genesis by creating new astroglia and additional
RGLs. We rarely observed temporary spikes in neurogenesis within the Gli1
#
-RGL population or
increases in astrogenesis within the Ascl1
#
-RGL population over the period examined (Figure
3.5E), suggesting minimal conversion between these two populations under physiological
conditions during young adulthood. However, with time we observed a convergent stem cell
behavior among two discrete population. In young adults, we observed a distinct fate choice
composition of Gli1
#
/Nestin
#
-RGLs with a switch toward symmetric division in the mature brain
at the expense of neurogenic divisions. On the other hand, neuronal fate-biased Ascl1
#
-RGLs
66
change their cell fate composition towards symmetric divisions at the expense of both neurogenic
and astrogenic divisions. After injury, on the other hand, neural stem cells that are normally
stochastic in fate choice (Gli1
#
/Nestin
#
-RGLs) become activated and appear to acutely adopt the
behavior of neuronal fate-biased stem cells (Ascl1
#
-RGLs). Interestingly, the increased generation
of the neuronal lineage comes at the expense of gliogenesis. We hypothesize that neural stem cells
can adapt their fate choices to meet the current regenerative demand of the local niche – in this
case, stochastic NSCs lose multipotentiality but gain repetitive cycling ability to regenerate the
neuronal lineage. Meanwhile, neurogenic biased stem cells expand through symmetric divisions
to compensate for lost Ascl1
#
-RGLs. Whether these adaptive responses occur more broadly, for
example following physiological alterations, other forms of brain injury and neurodegeneration,
remains to be investigated.
Understanding stem cell function during homeostasis and upon injury elucidates the origins
of tissue plasticity and may ultimately guide future regenerative strategies. Here, we provide a
framework for resolving complex neural stem cell behaviors to establish a consensus model of
neural stem cell identity. Our study indicates that seemingly divergent neural stem cell phenotypes
at the population level (Encinas et al., 2011) can be attributed to specific subpopulations at the
single-cell level. Specifically, neuronal bias and cell cycle re-entry preferentially occur in Ascl1
#
-
RGLs, whereas RGL differentiation into astrocytes predominantly occurs in stochastic NSCs
(Nestin
#
-RGLs). We do not observe invariant “division-coupled differentiation” of NSCs (Encinas
et al., 2011), but instead speculate that this phenotype reflects an end state of stochastic NSC
behavior. Furthermore, NSC expansion and symmetric division has been observed in some studies
(D. A. Berg et al., 2019; Bonaguidi et al., 2011; Dranovsky et al., 2011; Gebara et al., 2016a;
Ibrayeva et al., 2019; Jang et al., 2013; Pilz et al., 2018; Song et al., 2012), but not others (Balordi
67
& Fishell, 2007; Calzolari et al., 2015; DeCarolis et al., 2013; Encinas et al., 2011). This
discrepancy may be attributed to whether stochastic NSCs are labeled during homeostasis
(Bonaguidi et al., 2011; Jang et al., 2013), under physiological conditions (Dranovsky et al., 2011;
Song et al., 2012), or following injury (this study). Ultimately, our results show that context-
specific quantitative analysis of intrinsic stem cell properties and their response to environmental
changes enable the ability to define the identity and lineage relationships of different stem cell
states.
Experimental Procedures
Animals, Tamoxifen Administration and AraC Treatment
Animals were housed in a 14 hr light/10 hr dark cycle with free access to food. All
procedures were performed in accordance with institutional animal guidelines of Johns Hopkins
University School of Medicine. Nestin-CreER
T2
mice (Balordi & Fishell, 2007), Gli1-CreER
T2
mice (Ahn & Joyner, 2005), and Ascl1-CreER
T2
mice
(Kim et al., 2011) were used to clonally label
RGLs. The following genetically modified mice were originally purchased from Jackson Labs:
Gli1-CreER
T2
(Strain: Gli1
tm3(cre/ERT2)Alj
/J; stock:007913), Z/EG
f/+
(Strain: Tg(CAG-
Bgeo/GFP)21Lbe/J; stock: 3920)(Novak et al., 2000), Rosa-YFP
f/f
(Strain: B6.129X1-
Gt(ROSA)
26Sortm1(EYFP)Cos/
J; stock: 006148), mT/mG
f/f
(Strain: B6.129(Cg)-Gt(ROSA)26Sor
tm4(ACTB-
tdTomato,-EGFP)Luo
/J; Stock: 007676), Confetti
f/f
(StrainGt(ROSA)26Sor
tm1(CAG-Brainbow2.1)Cle
/J). Nestin-
CreER
T2
, Gli1-CreER
T2
, or Ascl1-CreER
T2
were crossed to fluorescent reporter mice for clonal
analysis. Nestin-CreER
T2+/-
::Z/EG
f/+
animals were generated by breeding nestin-CreER
T2+/-
and
Z/EG
f/+
mice or nestin-CreER
T2+/-
::Z/EG
f/+
with wild-type C57BL/6 mice. Gli1-CreER
T2+/-
::
Z/EG
f/+
or Gli1-CreER
T2+/-
::mT/mG
f/+
animals were generated by crossing Gli1-CreER
T2+/-
mice
68
and Z/EG
f/+
or mTmG
f/f
, respectively, or by breeding Gli1-CreER
T2+/-
::Z/EG
f/+
or Gli1-CreER
T2+/-
::mT/mG
f/f
with wild-type C57BL/6 mice. Ascl1-CreER
T2+/-
::YFP
f/+
or Ascl1-CreER
T2+/-
::mTmG
f/+
mice were generated by crossing Ascl1-CreER
T2+/-
with YFP
f/f
or mTmG
f/f
,
respectively (Figure
S1A-B). For each CreER
T2
driver, we tested different tamoxifen doses and reporter lines, including
Z/EG (Novak et al., 2000), mT/mG (Muzumdar et al., 2007), and Rosa-YFP (Srinivas et al., 2001),
and obtained combinations that exhibited high specificity, inducbility and reproducibility. At least
3 animals were checked for each reporter/driver combination to ensure there was no recombination
in the adult SGZ in the absence of tamoxifen. Stock tamoxifen (62 mg/ml; Sigma; T5648) was
made with a 5:1 ratio of corn oil/ethanol and heated to 37°C with periodic mixing. For lineage
tracing, 8 -12 week-old mice or 52-53 week-old were i.p. injected with a single dose of tamoxifen
or vehicle at various concentrations (see Table S1). No signs of distress were observed in injected
animals. We did not detect any obvious difference in densities of MCM2
+
proliferating cells or
DCX
+
immature neurons in the adult dentate gyrus of Gli1-CreER
T2
(95 + 8% for MCM2
+
cells
and 117 + 2% for DCX
+
cells; n = 3) or Ascl1-CreER
T2
knock-in mice (80 + 6% for MCM2
+
cells
and 90 + 4% for DCX
+
cells; n = 3) compared to Nestin-CreER
T2
transgenic mice, suggesting a
lack of gross haploinsufficiency effect on adult hippocampal neurogenesis.
Primer sets from original publications were used to identify genetically modified mice
(Ahn & Joyner, 2005; Balordi & Fishell, 2007; Kim et al., 2011; Lemberger et al., 2007;
Muzumdar et al., 2007; Novak et al., 2000). Genomic tail DNA was isolated in 25 mM NaOH, 0.2
mM EDTA and run for 35 PCR cycles. Z/EG mice were phenotyped in X-gal reactions [50 mM
K3Fe(CN)6, 50 mM K4Fe(CN)63H2O, 1 M MgCl2, 10 mg/ml X-Gal in PBS] for 4 hr to overnight.
For AraC treatment-induced injury, adult Gli1-CreER
T2+/-
::mT/mG
f/+
mice were injected
with a clonal dose of tamoxifen (Table S3.1) and housed for 3 to 7 days following tamoxifen
69
injection to avoid killing RGLs in a proliferative state (Figure S3.7A). AraC stock solution (1X,
2% weight per volume, 82 mM) was made by adding 5 ml of 0.9% 0.22 μm filtered sterile saline
to the 100 mg bottle of AraC (Sigma C1768). Pumps (Alzet Model 1007D Duret brain infusion kit
3) were assembled using sterile procedures. A supplied syringe tip was used to fill the pump with
AraC and one spacer was placed on the needle (0.5 mm thickness). Pumps were then incubated
overnight at 37°C in 0.9% saline. Animals were anesthetized and mini-osmotic pumps were
inserted into the mouse right hemisphere at the coordinates of 0.75 mm medial lateral and -0.5 mm
posterior using a stereotaxic injection machine and secured to the skull with superglue (Loctite
454). The incision was sutured. The osmotic pump was allowed to infuse for 4 days in mice under
normal housing conditions. Three days after pump installation, two shots of EdU (41.1 mg/kg body
weight) 2 hr apart were delivered to assess cell proliferation. EdU labelling was performed with a
Click-iT EdU Alexa Fluor imaging kit (Invitrogen). To stop drug flow, mice were anaesthetized,
and the osmotic pump was removed. Mice were once again housed for another 7 days before
analyses. Sham control animals were induced with the same level of tamoxifen and given the same
anesthesia but were otherwise uninjured when performing clonal analysis. Injury was assessed
both by EdU elimination in the SGZ and by GFAP immunostaining for reactive gliosis (Figure
S3.7B-C). Regions distal to the injection site containing reactive gliosis were processed for clonal
analysis.
Immunostaining, Confocal Imaging, and Image Processing
Mice were anesthetized and transcardially infused with saline and then 4%
paraformaldehyde. Brain sections were sectioned coronally (45 μm thickness) and maintained in
serial order throughout the entire dentate gyrus using custom, in-house staining chambers.
Immunohistology was performed using antibodies as previously described (Bonaguidi et al., 2011;
70
Ge et al., 2006). Cre staining was performed using TSA amplification (Cat# NEL704A001KT,
Perkin Elmer). Sections were washed in PBS with 0.3% Triton X-100. Endogenous peroxidase
activity was quenched with 0.3% Hydrogen Peroxide in PBS for 10 min. Sections were blocked
for 1 hr in TNB buffer (0.1 M TRIS-HCl pH7.5, 0.15 M NaCl, 0.5% blocking reagent) then
incubated overnight in 1:12000 rabbit anti-Cre antibody (Lemberger et al., 2007). Sections were
washed and incubated with Biotin-SP-Donkey anti-rabbit secondary antibody (1:200; Cat #711-
065-152; Jackson Immuno) in TNB buffer for 2 hr. Sections were washed and then incubated in
SA-HRP (1:100) in TNB buffer for 30 min. Sections were washed and then incubated 6 min in
fluorophore tyramide (1:50) in 1X amplification diluent. After a final wash, sections were
counterstained, mounted and imaged. Following antibodies were used: Nestin (1:500, chicken;
Cat#NES; Aves), GFAP (1:2000, rabbit; Cat#Z0334, DAKO), DCX (1:500, goat; Cat#SC-9066;
Santa Cruz). Antigen retrieval for MCM2 and Nestin antibodies utilized DAKO citrate buffer
(Dako S1699) at 95
o
C for 20 min and then left to cool at room temperature. Cells were then counted
using every 8
th
section throughout the entire dentate (n = 6-7 sections per dentate).
GFP
+
cells were identified with an Axiovert 200M microscope (Zeiss) and then acquired
as z-stacks on Zeiss 710 single-photon confocal microscope using 40X or 63X objectives. To
facilitate 3D reconstructions of GFP
+
cells spanning multiple sections, optical stacks were taken
of the entire clone then serially aligned with Reconstruct 1.1.0 (John C. Fiala, Human Brain
Project, the National Institutes of Health) as previously described (Bonaguidi et al., 2011). To
achieve alignment between images of adjacent physical sections, the last of the 2D optical images
of one section was translated and rotated in the X-Y plane to match those of the subsequent section.
Recorded keystrokes were propagated to all preceding optical sections within the Z-stack. This
process was repeated across all Z-stacked confocal images to reconstruct entire clone across
71
multiple sections. Full resolution aligned images were exported into Imaris 8.5 (Bitplane) with
voxel sizes adjusted according to dimensions specified in the LSM file for visualization of
reconstructed clone in 3D.
Clonal Analysis
Clonal analysis and categorization of cell types by morphological and immunohistological
criteria were performed as previously described (Bonaguidi et al., 2011). Precursors were defined
as both RGLs and IPCs in assessing initial induction enrichment (Figure 3.1B). Clonal distance
measurements were made using Imaris (Bitplane), and an in-house MATLAB script to determine
distances in 3D. A clone was defined as consisting of cells located within a radius of 150 μm from
the clone center. Consistent with criteria used previously (Bonaguidi et al., 2011), approximately
6-18 clones per dentate allowed for clonal analysis based on computer simulation. Clones were
induced randomly throughout the dentate, showing no preferential distribution among anterior-
posterior, medial-lateral and suprapyramidal-infrapyramidal locations. To confirm a consistent
level of induction, the number of clones per dentate was assessed at varying time points following
induction, which was shown to be comparable over time (Figure S3.1D). At short time points (0.5
- 1 dpi) clones were single cells or dividing clusters of cells indicative of initial labeling of one
cell. Active, dividing clones were defined as clones that underwent division as assessed by having
two clear nuclei by DAPI using confocal microscopy (Figure 3.1C-D). When assessing cell
division, two nuclei were either completely encompassed (in the case of the membrane bound
reporter mT/mG) or filled (in the case of Rosa-YFP or Z/EG reporter) by the GFP signal in 3D.
MCM2 staining was additionally used as a mitotic marker to confirm proper scoring during
lineage-tracing (Figure 3.1F). Short time points (Ascl1
#
at 1 dpi; Gli1
#
at 3 dpi) were chosen to
confirm consistency among different reporter mouse lines (Z/EG, YFP, mT/mG; Figure S3.2D-
72
E). Clones were categorized according the clone composition among RGL-containing clones
(Bonaguidi et al., 2011; Song et al., 2012). Clones containing more than 1 RGL and progeny were
classified by the lineage produced: neuronal, astroglial, or both. For all time points, RGL
maintenance was assessed as percent of clones that contained at least one RGL (Figure 3.3C).
Lineage trees were generated from clones containing RGLs and progeny (Figure S3.2A,
3.3A, 3.6A-D and S3.6D). Progeny were identified using morphological and immunohistological
criteria as previously described (Bonaguidi et al., 2011). Unlike the proceeding clone
compositions, each fate choice (R-R, R-N, R-A) was considered as a separate division at shorter
time points (1, 3, 7 dpi). Repeated RGL-IPC divisions were scored if the clone contained an IPC
cluster that migrated away and the RGL was in the midst of RGL-IPC division. Order of division
was assessed by inverse spatial proximity of progeny to RGL such that progeny at a greater
distance were considered an earlier division. For a longer chase at 30 dpi, lineage trees additionally
accounted for the appearance of immature neurons, mature neurons and transition astrocytes. The
presence of an IPC/neuroblast (small compact soma with tangential process), immature neuron
(DCX
+
radial process), and mature neuron (large soma, radial morphology with elaborate
dendrites/spines were scored as a separate division due to prolonged developmental kinetics during
adult hippocampal neurogenesis (Sun et al., 2013; Zhao, Teng, Summers, Ming, & Gage, 2006).
The Gli1-CreER
T2
line allowed for birth-dating the first division before AraC treatment and
assessment of RGL activation, cell cycle re-entry and lineage trees after injury (Figure 3.5 and
S3.7). Division quantification was derived from computational analysis and also included division
modes prior to AraC treatment (Figure 3.5D).
73
Computational analysis
Distribution of activation times. The activation time of an RGL is the time until it first
enters cell cycle upon labelling (Tentry). To deduce the distribution of activation times from the
clonal data, we calculated at each time point the fraction of RGLs that had not yet divided.
As the labelling protocols targeted a small proportion of IPCs in addition to RGLs, there
was a degree of error involved in our assignment: we could not decide unambiguously whether a
clone consisting entirely of IPCs and differentiated cell types was originally derived from an RGL
or an IPC. In the case of Gli1
#
-RGL and Ascl1
#
-RGL clonal data, the number of such clones was
so small that their exclusion would not significantly affect our results. For Nestin
#
-RGL, 10/34
clones consist only of 1-2 IPCs at 2 dpi; we assumed that these were IPC-derived and excluded
them from the analysis. By 7 dpi, due to the short cell cycle time of IPCs (D. K. Berg et al., 2015;
Hayes & Nowakowski, 2002; Hodge et al., 2008), we would expect IPC-derived clones to have
grown in size; we therefore took clones containing 6 or more IPCs to be IPC-derived (2/40), and
clones consisting only of 2 IPCs to be RGL-derived (4/40). At later time points, we assumed that
all clones were RGL-derived. As a second caveat, a clone consisting of a single RGL was scored
as an undivided RGL but may have given rise to progeny that were subsequently lost through cell
death. However, effects of any erroneous assignments were likely negligible compared to
experimental noise over this timescale.
For a population of cells that divide stochastically at a constant rate , the fraction of
undivided cells at time , denoted , decays exponentially over time according to .
For all three labelling protocols, exponential decay curves provided excellent fits to the
experimental data (Figure S4A-C). By performing a weighted least-squares fit, we deduced that
Nestin
#
-RGLs enter cell cycle at a rate = 0.044 ± 0.005 per day, equivalent to a mean cell cycle
λ
t R
0
(t) R
0
(t)=e
−λt
λ
74
time of = 23 ± 3 days. Similar fits to the early-time Gli1-CreER
T2
and Ascl1-CreER
T2
data
suggested mean activation times of 0.78 ± 0.02 days and 0.35 ± 0.04 days, respectively (Figure
4B).
Distribution of cell cycle re-entry times. Similar to RGL activation times, we can estimate
cell cycle re-entry times from the fraction of cells that have divided exactly once. For a population
of cells with activation rate and cell cycle re-entry rate , one can show that the fraction of cells
divided once at time , denoted , takes the form
.
For Ascl1
#
-RGLs and Gli1
#
-RGLs, divisions were scored based on the ability to clearly
identify newly generated cell types. At early time points (1, 3, 7 dpi), some RGL retaining clones
contained additional RGLs, IPCs and astroglia, whereas at 30 dpi immature neurons (INs), mature
neurons (MNs) and transition astrocytes (TAs) also appeared. Generation of each cell type was
scored as a unique division due to the differentiation kinetics from IPCs to MNs along the neuronal
lineage (Zhao, Deng, & Gage, 2008). Likewise, the same approach was applied to non-neuronal
lineages due to the discrete fate specification of new RGLs and the astroglial lineage (Bonaguidi
et al., 2011). Conservatively, each cell type along the neuronal and astroglia lineage was scored as
1 division irrespective of cell number within each category. In cases where multiple cell types were
present, the order of cell generation was defined by (i) an anti-correlate of distance from the RGL
and (ii) the anti-correlate of maturation through the lineage.
Given our estimates of the activation times, weighted least-squares fits suggested cell cycle
re-entry times of = 33 ± 7 days for Nestin
#
-RGLs, 53 ± 12 days for Gli1
#
-RGLs and 5.3 ± 0.8
days for Ascl1
#
-RGLs (Figure 4C and S4G-I).
1/λ
λ µ
t R
1
(t)
R
1
(t)=
λ
µ−λ
e
−λt
−e
−µt
( )
1/µ
75
Fate choice probabilities of Nestin
#
-RGL. Since the estimated re-entry time into cell cycle of
Nestin
#
-RGL is close to their activation time, we assumed in the following that Nestin
#
-RGLs may
be modelled as stochastically entering into cell cycle (with the timing between consecutive cell
divisions statistically uncorrelated – Markovian) at a constant rate up to 60 dpi.
We next turned to the question of how their fate choices may be inferred from the lineage
data. At every division, an RGL produces two daughter cells, choosing two out of the different cell
types it could, in principle, generate. Previous work has shown that RGLs can directly give rise to
RGLs, IPCs, transition astroglia (TA), and astroglia (A) (Bonaguidi et al., 2011). We may ask
whether these fate decisions are made stochastically, or whether there is evidence for a
predetermined program.
For an ensemble of clones with varying cellular composition, we can define the probability
that any given clone will have a particular composition at a time
after
induction. This probability evolves over time according to the transitions between different cell
types. For a population of equipotent cells dividing stochastically according to the same constant
division rate and fixed probabilities of fate choices, its evolution may be captured mathematically
through a Master equation. As IPCs divide rapidly and many die (D. K. Berg et al., 2015; Sierra
et al., 2010; Song et al., 2013), it is not currently feasible to infer the dynamics within the neuronal
lineage from the clonal data. However, we were only concerned with the fate choices of RGLs,
which may be inferred from the RGL and TA/A cell counts alone. We needed only to distinguish
among RGL self-renewal (R), differentiation down the neuronal lineage (I), and down the
astroglial lineage (A). The corresponding fate choice probabilities for RGL divisions are denoted
by for divisions that give rise to two RGL, for divisions that gives rise to one RGL and one
P
{comp}
(t) {comp} t
r
RR
r
RI
76
cell of the neuronal lineage, etc. The Master equation for the probability that a clone contains
RGLs and TA/A at time can be written as
where is the constant rate of cell division.
Numerically integrating the Master equation and computing the resulting clone size
distributions, we obtained very good agreement with experimental data (Figure S4D), suggesting
that Nestin
#
-RGLs do indeed behave as a population of stochastically dividing stem cells, with no
evidence for a deterministic program. Least-squares fitting to the RGL and TA/A content of clones
results in a division rate = 0.04 per day, equivalent to a cell cycle time of = 25 days. The
corresponding fate choice probabilities are given in Figure 4D. As the clone size distributions are
well approximated by the fit, the model prediction for the average RGL and TA/A content of clones
is also consistent with the data (Figure S5A).
The clonal data for Nestin
#
-RGLs are thus consistent with a model in which RGLs
constitute an equipotent population of cells that divide stochastically according to constant fate
choice probabilities, at least over the 60 day time course. While it is, of course, possible that more
complicated models might fit the data equally well, we can gain confidence in our conclusion from
four independent self-consistency checks of our model:
P
m,n
(t)
m n t
dP
m,n
(t)
dt
= λ r
RR
(m−1)P
m−1,n
(t)−mP
m,n
(t)
( )
+r
RA
mP
m,n−1
(t)−mP
m,n
(t)
( ) {
+r
II
(m+1)P
m+1,n
(t)−mP
m,n
(t)
( )
+r
AA
(m+1)P
m+1,n−2
(t)−mP
m,n
(t)
( )
+r
IA
(m+1)P
m+1,n−1
(t)−mP
m,n
(t)
( )}
= λ r
RR
(m−1)P
m−1,n
(t)+r
RA
mP
m,n−1
(t)+(m+1) r
II
P
m+1,n
(t)+r
AA
P
m+1,n−2
(t)+r
IA
P
m+1,n
(t)
( ) {
−mP
m,n
(t) r
RR
+r
RA
+r
II
+r
AA
+r
IA
( )}
,
λ
λ 1/λ
77
• First, we note that the cell cycle time obtained by fitting the solution of the Master equation
to the experimentally observed clone size distributions is consistent with our earlier estimate from
the fraction of undivided RGLs.
• Second, the probability of symmetric RGL self-renewal is balanced by the total probability
of RGL loss due to differentiation ( ). We would therefore predict that the average
number of RGLs per clone remains close to 1, which is indeed the case (Figure S5A).
• Third, we expect on theoretical grounds that the fraction of persisting clones, i.e. clones
that retain at least one RGL at time , should follow a power-law decay of the form
(Clayton et al., 2007). The data are consistent with this prediction (Figure S5B).
• Fourth, we can compare the fate choice probabilities obtained from the fit with those
inferred directly from the clonal data. To this end, we assign to every RGL-derived clone in the 2,
7 and 30 dpi Nestin
#
-RGL
data the first division likely undergone by the initially labeled RGL (n
= 33). Our assignments are subject to some ambiguity; in particular, RGL self-renewal followed
by differentiation of one or both daughter cells could be erroneously scored as a differentiating
division. Comparing two sets of parameters, the fate choice probabilities found in this way indeed
suggest a lower value for than predicted by the fit, but overall the results are roughly consistent
(Figure S5C).
Over the first 60 days following induction, the clonal data for Nestin
#
-RGL is thus
consistent with a model in which RGLs constitute an equipotent population of cells that divide
stochastically according to constant fate choice probabilities.
Fate choice probabilities of Gli1
#
-RGL and Ascl1
#
-RGL. As the activation (Tentry) and re-entry
times (Tc) of Ascl1
#
-RGL and Gli1
#
-RGL
differ markedly, we cannot apply the same Master
r
RR
=r
II
+r
AA
+r
IA
t 1/(1+r
RR
λt)
r
RR
78
equation approach we employed for the Nestin
#
-RGL data, which assumes a constant average time
between divisions. Instead, we estimated their fate choice probabilities directly from the clonal
data in the same way as we did for Nestin
#
-RGLs. We inferred the first division of labelled RGLs
at 1 dpi for Ascl1
#
-RGLs and at 1-3 dpi for Gli1
#
-RGLs, choosing different time windows due to
the slower activation time of Gli1
#
-RGLs (n = 76 for Ascl1
#
-RGLs and 88 for Gli1
#
-RGLs). As
discussed in the main text, fate choice probabilities differed significantly between Gli1
#
-RGLs and
Ascl1
#
-RGLs, with the latter displaying an obvious bias towards the neuronal lineage (see also
Figure 4D). Interestingly, the fate choice probabilities of Gli1
#
-RGLs were roughly consistent with
those of Nestin
#
-RGLs. It is therefore conceivable that the two labelling protocols target the same
population, with Gli1
#
-RGL labelling RGLs that are about to enter cell cycle.
To resolve whether the Gli1
#
-RGL clonal data is consistent with the Nestin
#
-RGL
dynamics, we simulated the time evolution of 10,000 RGLs that enter cycle according to the Gli1
#
-
RGL activation time, but subsequently evolve according to the Nestin
#
-RGL division rate and fate
choice probabilities. At every time point, a number of clones equivalent to that observed
experimentally is selected at random, and the resulting clone size distributions are computed.
Repeating this process 5,000 times, we calculated the average expected clone size distributions as
well as 95% plausible intervals. We found that the Gli1
#
-RGL clonal data is indeed consistent with
the Nestin
#
-RGL dynamics (see Figure S4D-E). The average RGL and TA/A content of clones are
also well predicted by the model over a 30 day period (Figure S5D). Taking into account the offset
in activation times, this mirrors our observations at later times in the Nestin
#
-RGL data.
In contrast to the Nestin
#
-RGL and Gli1
#
-RGL data, no combination of division rate and
fate choice probabilities results in a satisfactory fit of the same model to the Ascl1
#
-RGL data,
suggesting that Ascl1
#
-RGLs do not behave as a population of stem cells cycling at a constant rate
79
and with fixed fate choice probabilities. For example, assuming that Ascl1
#
-RGLs keep dividing
at the estimated cell cycle re-entry rate and with the fate choice probabilities determined above,
the model predicts that RGLs would be lost faster and more TA/A would have been generated by
30 dpi than experimentally observed (Figure S4F and S5E).
In summary, the data obtained from the three labeling protocols thus suggest that activating
RGLs conform to one of two populations over the one-month period we examined: a fast-cycling
cohort in which RGLs are strongly biased towards neuronal differentiation, and a slower-cycling
group with more balanced fate choice probabilities.
Statistics
Statistics were performed as indicated in each Figure legends. For in vivo experiments, the
number of independent experimental replicates are indicated in figure legends, with n representing
n experimental replicates (clones) using at least n animals. Mann-Whitney unpaired two tailed t-
test was conducted to compare the Imatinib effect against Vehicle controls. All multiple
comparisons were performed with GraphPad Prism’s one-way ANOVA function with
Bonferroni’s multiple comparisons test. All clones observed at each time point were treated as
statistically equivalent. For computational analysis, all clones observed at each time point were
treated as statistically equivalent. No randomization or blinding was used in the animal studies.
Sample sizes were estimated based upon previous clonal studies (Bonaguidi et al., 2011; Song et
al., 2012). Standard errors in mean frequencies were estimated as Z(1−)/, where is the
frequency of a given characteristic at a time point and the number of clones considered. Fits
were performed by weighted least squares using custom-written scripts in MATLAB (MathWorks,
Inc.).
80
Chapter IV: Developmental Origins of Adult Neural Stem Cells
Abstract
Adult neural stem cells exist as a sparse population within multicellular tissue. Yet, the
exact developmental origins of neural stem cell (NSC) populations and how they contribute to
adult neurogenesis have not been elucidated. To gain insight into the pattern of NSC division and
neuron/glia production we developed a new genetic marking technique that includes the stochastic
multicolor Confetti reporter for increased tracing throughput and computational modeling
approaches to elucidate the NSC clonal dynamics during development and aging. Here we
identified that Gli1+ precursors labeled at late embryonic stage E17.5 in the mouse dentate gyrus
originate from both dorsal and ventral brain structures. Clonal tracing showed that neural
progenitor cells give rise to two parallel lineages, one unipotent population that divide rapidly and
continuously generates only granule neurons, and a multipotent population that is slow cycling
and generate neurons, astrocytes and/or oligodendrocytes through development and adulthood.
Computational modeling identified a “two populations” model where NSCs lost at different rates
can replicate the observed evolution of the NSC clone size distribution. Together our findings
suggest the existence of two parallel NSC lineages where one contributes to neurogenesis only and
follows a “continuous” model, while another generates neurons and glia in a sequential order.
81
Introduction
In adult mammalian brain, neurogenesis happens in two specific regions, subventricular
zone (SVZ) near the lateral ventricles and subgranular zone (SGZ) in the dentate gyrus (DG) of
the hippocampus (Bond et al., 2015; Kuhn et al., 2018). In the adult hippocampus, neural stem
cells (NSCs) give rise to new dentate granule neurons and astrocytes through a process of cell
cycle entry and generation of intermediate progenitor cells (Bonaguidi et al., 2011; Bond et al.,
2015). Development represents a period of exceptional plasticity and resiliency which
subsequently declines with tissue maturation. Since the initial discovery of adult neural stem cells
(NSCs) in the hippocampus dentate gyrus, a central question for over 10 years remains - from
where do they arise? However, little is known about the developmental process that leads to the
establishment of adult neural stem cells.
There are three models explaining how adult NSCs could be generated during
development. It should be noted that these models are not mutually exclusive and can coexist (A
& A, 2009; D. A. Berg et al., 2019; Hochgerner, Zeisel, Lonnerberg, & Linnarsson, 2018). The
first “Sequential” model proposes that NSCs produce different neuronal subtypes during
embryonic stages, followed by astroglia generation during the postnatal period, and then the
residual NSCs are converted into adult progenitors to generate specific neurons in the adult brain
(A & A, 2009). To date, this model has no experimental support. Meanwhile, a second “Set-aside”
model has been proposed for adult subventricular zone (SVZ) neural progenitors, in which a
precursor pool initially generates cortical, striatal, or septal neurons and then diverges during mid-
embryonic development, with one pool continuing to generate neurons and glia and the other
remaining dormant until adulthood and activates to exclusively produce olfactory bulb
interneurons (LC et al., 2015; S et al., 2015). A third “Continuous” model has been recently
82
proposed in which a common dentate neural progenitor population exclusively contributes to
dentate neurogenesis from development into adulthood (D. A. Berg et al., 2019). However, other
cell types make up the dentate gyrus, and the sequential order of neuron, astrocytes and
oligodendrocyte production remains to be identified.
A population study that utilized Gli1-CreER
T2
genetic fate-mapping (Li & Clevers, 2010)
showed that neural progenitors located in the ventral SVZ at embryonic stage 17.5 (E17.5) give
rise to a subset of NSCs in the adult dentate gyrus. In addition, there has been known heterogeneity
that exist in the adult hippocampal NSC population (Bonaguidi et al., 2012; Ibrayeva et al., 2019).
We took advantages of the previous population studies and developed single cells clonal lineage-
tracing technique in mice driven by Gli1-CreER
T2
regulatory element combined with stochastic
multicolor Confetti reporter to trace the entire process of DG formation from a single cell during
development through adulthood.
Clonal analysis, through labeling of individual progenitor cells and following their progeny
in the developing hippocampus in vivo, could provide definitive answers to the ontogeny of dentate
neurons and glia. Indeed, previous studies using thymidine analogues, reporter mice, and various
immunohistological methods have contributed to the current framework of hippocampal
neurogenesis (Bonaguidi et al., 2011; G, L, G, & SJ, 2013; Gebara et al., 2016; Li & Clevers,
2010). However, the lack of cellular resolution of progeny cell fate, vital for dissecting progenitor
division patterns, lineage specification, lineage relationships have so far precluded a definitive
understanding of this complex and dynamic process. In this study, we exploited the unprecedented
resolution of Gli1-CreER
T2
::Confetti labeling and performed a computational clonal analysis of
NSC division and lineage relationship in the mammalian dentate gyrus.
83
Results
Clonal Analysis of Developmental Hippocampus using Gli1-CreER
T2
system
We developed a transgenic labeling approach in mice driven by Gli1
#
regulatory element
combined with stochastic multicolor Confetti reporter (CFP, YFP, RFP) to trace the entire process
of DG neurogenesis from a single cell during development through adulthood (Figure 4.1A). To
perform in vivo single cell lineage tracing, we injected time-pregnant female Gli1-CreER::Confetti
mice with a single dose 14.4 mg/kg of tamoxifen (TMX) to label pups at E17.5 and analyze brains
at postnatal day (P)7, P15, P30, P90, P180 and P365 (Figure 4.1B). We found no labeling in the
absence of TMX treatment (n = 3 brains). We titrated the TMX dose to achieve sparse labeling,
single i.p. injection of 14.4 mg/kg TMX yielded an abundant number of clones that were
simultaneously discernable. Across all ages Confetti reporter yielded ~34% CFP, ~28% YFP and
~39% RFP clones, averaging 5-6 clones per mouse brain, ~1.5 clone per color per hemisphere
(Figure 4.1C-D). Clones contained NSCs, neurons, astrocytes, oligodendrocytes, and newborn
cells across all timepoints (Figure 4.1E-G, S1A-C). To recover all labeled cells, we performed
serial sectioning and image analysis of individual clones (Figure 4.1F-G, S1A-C).
84
Figure 4.1. Clonal
Analysis of
Developmental
Hippocampus using
Gli1CreER
T2
system.
(A) Schematic
representation of genetic
approach used for in vivo
single cell clonal lineage
tracing of Gli1
#
-NSCs in the
developing mouse dentate
gyrus. (B) Experimental
timeline: time pregnant
female mice were given a
single dosage of tamoxifen
(TAM) at E17.5 for clonal
lineage tracing analysis at
different timepoints. See
also Table S1. (C)
Quantification of clone
percentage per color per
hemispheres across ages.
Mean± SEM. (N = 41-96
clones, Table S1). (D)
Quantification of average
number of clones per color
per hemispheres across
ages. Mean± SEM. (N = 41-
96 clones, Table S1). (E)
Quantification of overall
cell type distribution.
NSC=neural stem cell;
TA/A=astroglial lineage;
IPC/NB/N=neuronal
lineage;
OPC/OL=oligodendrocyte
lineage. Mean± SEM. (N = 41-96 clones, Table S1). ANOVA with Tukey’s multiple comparisons
test. (F) Cartoon illustrating NSC cell fate choices during postnatal development. TA/A=astroglial
lineage; IPC/NB/N=neuronal lineage; OPC/OL=oligodendrocyte lineage. (G) Confocal images of
NSC cell fate choices during postnatal development. TA/A=astroglial lineage;
IPC/NB/N=neuronal lineage; OPC/OL=oligodendrocyte lineage.
Scale bar, 10μm (G). *p<0.05, **p<0.01, ***p<0.001, n.s - not significant.
85
Existence of NSC parallel lineages during development
The developmental origins of adult NSC in the dentate gyrus remain unclear. Some studies
suggest that NSCs originate from ventral SVZ while others strongly believe that it is restricted to
a dorsal region (G, H, SR, & SJ, 2009; G et al., 2013; LC et al., 2015). The mouse hippocampus
is located that the mediotemporal edge of the neocortex and connects the thalamus along its
septotemporal axis. Dorsal to the thalamus is the dorsal hippocampus, while ventral to the thalamus
is the ventral hippocampus. We performed the sagittal sectioning at different mediolateral levels
to comprehensively examine the Gli1-CreER
T2
::Confetti+ cells through the both dorsal and ventral
developing hippocampus (Figure 4.2A). Unexpectedly, Gli1-CreER
T2
::Confetti+ cells were
present at both ventral and dorsal-forming DG at postnatal day 0 (P0/P1) (Figure 4.2B-C). Some
Gli1-CreER
T2
::Confetti+ cells were distributed from the ventral to the dorsal arm at the transitional
(middle) region (Figure 4.2B). These data results demonstrated that progenies labeled at E17.5
with Gli1-CreER
T2
line are not restricted in one region, but rather evenly distribute in both dorsal
and ventral hippocampus by birth, give rise to the NSCs that settle in the SGZ.
Next, we performed clonal lineage tracing of E17.5 Gli1+ precursors during postnatal
dentate development. During DG tissue ontogenesis and remodeling we observe clones containing
neural stem cells (NSCs) and non-NSC containing clones. Interestingly, there is an early rapid
decrease in the fraction of NSC-containing clones from P7 to P15, 60% to a 40% respectively
(Figure 4.1F), before the fraction of clones remains approximately constant at around ~40%, until
after P90 where the number of surviving clones falls again to around ~15% mark (Figure 4.1D).
This is consistent with a decrease in developmental progenitors as ontogenesis terminates at around
P15. The location of the plateau suggests that 40 ± 5 % of developmental progenitor cells labelled
at E17.5 under the Gli1 promoter go on to produce adult NSCs.
86
The average number of cells per NSC-containing clone was quantified to assess clonal
expansion. The average number of NSCs within clones containing NSCs (Figure 4.2E) is amplified
to ~4 NSCs per clone by the end of hippocampal ontogenesis (P15), before gradually
monotonically declining to ~1.2 NSCs at P365 (Figure 4.3E), indicating substantial NSC
activation, and subsequent loss occurring after P30. The peak in average NSCs number per clone
suggests that by P15 all the NSCs that were produced during development have been made,
although these NSCs can symmetrically divide later. Within NSC-containing clones we divided
clones that contain only neurons (Unipotent) and ones that contain neurons and glia [astrocyte and
oligodendrocytes] (Multipotent) (Figure 4.2G). We saw that during the development the fraction
of clones that produce neurons and neuron and glia is around ~50%, varying at P15 where the most
substantial fraction of clones produces neurons, but not glia (Figure 4.2G-H). The average number
of NSCs per NSC-containing clones shows a similar trend of initial increase ~2 vs. ~4 NSC per
clone, followed by decay for both unipotent and multipotent lineages (Figure 4.2I). Such a decay
can be attributed to activation of NSCs in adult and potential aging effect (Figure 4.3E).
We next explored the cumulative distribution of neuron number (IPC/NB/N) in clones
(Figure 4.2J, S4.2A, D) that can provide the insight to the variability of neuronal production for
developmental NSCs. The neuronal output of an induced neural progenitors is determined by
summing together the number of intermediate progenitor cells (IPCs), neuroblasts (NBs) and
neurons (Ns). Since individual IPCs produce a variable number of neurons, neuronal output is
underestimated at the earliest time-points. Neuronal number in clones (Figure 4.2D, S2D) shows
an initial increase from 8 ± 1.2 at P7, 9.2 ± 2.0 at P15, to 13.0 ± 1.7 at P30. This suggests that at
87
completion of ontogenesis each developmental NSC marked at E17.5 produces roughly 13-15
neurons.
Figure 4.2. Establishment of
parallel stem cell lineages in
development (See also Figure
S1) (A) Experimental paradigm
for P0/P1 experiments. (B)
Quantification of Gli1
+
cells
per region of the DG (dorsal,
ventral, middle) at P0/P1
timepoint. (Table S1). (C)
Representative confocal images
of Gli1
+
cells in the DG at
P0/P1 timepoint. (D)
Quantification of the fraction of
clones containing NSC. Mean±
SEM (N = 41-96 clones, Table
S1). (E) Quantification of
average cell number of NSCs
per NSC-containing clones at
multiple days post tamoxifen
injection. Mean± SEM (N = 41-
96 clones, Table S1). (F)
Quantification of percentage of
clones that contain NSC at P7,
P15, P30 timepoints. Mean±
SEM (N = 41-96 clones, Table
S1). (G) Quantification of
Unipotential vs Multipotential
clone composition among
maintained clones at 7, 15, 30
days post tamoxifen injection.
Mean± SEM. (N = 41-96
clones, Table S1). ANOVA
with Tukey’s multiple
comparisons test. (H)
Quantification of fate
composition among all
maintained clones at 7, 15, 30
days post tamoxifen injection.
Mean± SEM. (N = 41-96
clones, Table S1). ANOVA
with Tukey’s multiple
comparisons test. (I)
88
Quantification of average number of NSCs per all, Unipotent and Multipotent clones at 7, 15, 30
days post tamoxifen injection. Mean± SEM. (N = 41-96 clones, Table S1). ANOVA with Tukey’s
multiple comparisons test. (J) Quantification of average number of intermediate progenitor
cells/neuroblasts (IPCs/NB) and neurons (N) per neuron-containing clones at 7, 15, 30 days post
tamoxifen injection. Mean± SEM. (N = 41-96 clones, Table S1). ANOVA with Tukey’s multiple
comparisons test. (K) Quantification of average number of transition astrocytes (TA) and mature
astrocytes (A) per astrocyte-containing clones at 7, 15, 30 days post tamoxifen injection. Mean±
SEM. (N = 41-96 clones, Table S1). ANOVA with Tukey’s multiple comparisons test. (L)
Quantification of average number of oligodendrocyte progenitor cells (OPC) and mature
oligodendrocytes (OL) per oligo-containing clones at 7, 15, 30 days post tamoxifen injection.
Mean± SEM. (N = 41-96 clones, Table S1). ANOVA with Tukey’s multiple comparisons test. (M)
Quantification of the relative fraction of N + A clone (A), N + O clone (O), and N + A + O clone
(A+O) at 7, 15, 30 days post tamoxifen injection. (N = 41-96 clones, Table S1). (N) Quantification
of the overall fraction of three different types of gliogenesis clones (A, ~68%; O, ~20%; A + O,
~13%) at 7, 15, 30 days post tamoxifen injection. (N = 41-96 clones, Table S1). Scale bar, 100μm
(G). *p<0.05, **p<0.01, ***p<0.001, n.s - not significant.
In case of glia generation only ~25% of clones contain glia, while the contribution from
oligodendrocyte lineages is ~15% (Figure S4.2B-C). The average number of astroglia per clone in
clones containing astroglia is picking at P15 ~20 cells per Astrocyte-containing clones and ~5 cells
per NSC-containing clones (Figure 4.2K, S4.2E). By contrast, the number of OPCs and
oligodendrocytes per clone is roughly picking at P30 ~10 cells Oligo-containing clones and at P7
~2 cells per NSC-containing clones, suggesting an oligodendrocyte wave proceeds the generation
of astrocytes (Figure 4.2K-L, S2E-F). Therefore, the data suggests that during postnatal
hippocampal ontogenesis and remodeling on average 13-15 neurons are being produces by neural
progenitor cells followed by oligodendrocyte and then astroglia generation.
We next analyzed glial cell composition of individual clones. We consistently observed the
following three types of clones: neuron plus astrocyte, neuron plus oligodendrocyte, and neuron
plus astrocyte and oligodendrocyte. Moreover, the relative fraction of N + A, N + O, and N + A +
O clones across P15 and P30 was largely consistent where NSCs-containing clones consist of
~70% of astrocytes, ~ 12% only oligodendrocytes, 17% producing both. In contrast, at P7
timepoint we observed ~40% of gliogenic NSC-containing clones contain only oligodendrocytes,
89
~50% producing only astrocytes, and ~10% producing both (Figures 4.2M-N). These results
suggest that, upon transition into the gliogenic phase, NSCs exhibit an initial propensity in
producing oligodendrocytes (P7), and later in astrocytes and/or oligodendrocytes (P15-P30).
Together these results show that Gli1-expressing precursors that are labeled at late
embryonic stage E17.5 are consist in two parallel lineages where one continuously contributes to
neurogenesis only, and another one generates neurons and glia through late embryonic and early
postnatal dentate development.
NSC clonal evolution during hippocampal tissue remodeling
Next, we wanted to comprehensively investigate lineage relationships between
developmental and adult neurogenesis. Since we observed the existence of unipotent and
multipotent clones in early postanal development, we wanted to see how clones are maintained
through adulthood. First, we looked at the potency of the NSC-containing clones, and it appeared
that there is ~50:50 distribution between unipotent and multipotent clones (Figure S4.3A-C).
Further, the potency of NSCs does not appear to be significantly different to the potency of
developing NSCs. If a substantial fraction of NSCs produced at the termination of neurogenesis
went on to produce astrocytes then we would expect a gradual increase in the fraction of Neuron
+ Glia clones, matching a decrease in neuron-only clones. However, this is not observed the
fraction of neuron + glia clones decrease between P30 and P180 (Figure 4.3 C-E) for all clones
which can be attributed to cell death of glia.
After ontogenesis is complete, the neuron number for N-containing clones increases at P15,
before increasing significantly again between P30 and P90 (Figure 4.2J, 4.3F, S4.3E). There are
two effects to consider which can increase the mean neuron number for surviving clones after the
developmental phase. As NSCs are stochastically activated and lost, clones with fewer NSCs are
90
more likely to be lost. If there is initially a correlation NSC number and neuron number for NSC-
containing clones, then loss of smaller clones (fewer NSCs) is artificially increased by the mean
neuron number. On the other hand, the average astrocyte number increases in all clones picking at
P180 (Figure 4.3G, S4.3C, F). This is consistent with production of astrocytes by NSCs. The
significant drop in average oligodendrocyte number from P30 to P90 (Figure 4.3H) suggests a late
termination of oligogenesis with progenitor continue to produce astrocyte well after P30.
Focusing on clones producing both neurons and glia provides insights into patterns of
astrogenesis and oligogenesis. At early times, surviving clones correspond to clones late in
ontogenesis. Until P30, a small fraction of NSC-containing clones is seen with oligodendrocytes
and neurons, but not astrocytes. As these clones are not seen after P30, it is likely that these clones
were late in development at P15 but did not produce any NSCs. From P7 to P90, oligodendrocytes
are still seen in NSC-containing clones, but only in clones containing astrocytes. Additionally,
there is a fraction of extinct clones that produce oligodendrocytes and neurons, but not astrocytes.
This is 75% of extinct neuron + glia clones at P7, but rapidly decreases at P15 before dropping to
0% by P180 (Figure 4.2M, 4.3I). This suggests that astrogenesis may occur either simultaneously
with or after oligogenesis. After P180, there are no clones with oligodendrocytes without
astrocytes, suggesting that the small fraction of clones induced at E17.5 which complete
development without astrocytes, become neuron-only clones due to cell-death of oligodendrocytes
(Figure 4.3I).
91
Figure 4.3 Clonal evolution
during hippocampal tissue
remodeling. (See also Figure
S2)
(A) Schematic illustration of
hippocampal NSC cell fate
decisions in the adult brain.
(B) Confocal images of NSC
clones at P30 (Unipotential
NSC-N clone) and P90
(Multipotential NSC-N-A
clone). NSC=neural stem cell;
TA/A=astroglial lineage;
IPC/NB/N=neuronal lineage;
OPC/OL=oligodendrocyte
lineage. (C) Quantification of
Unipotential vs
Multipotential clone
composition among
maintained clones at 30-, 90-,
180- and 365-days post
tamoxifen injection. Mean±
SEM. (N = 41-96 clones,
Table S1). ANOVA with
Tukey’s multiple
comparisons test. (D)
Quantification of fate
composition among all
maintained clones at 30-, 90-,
180- and 365-days post
tamoxifen injection. Mean±
SEM. (N = 41-96 clones,
Table S1). ANOVA with
Tukey’s multiple
comparisons test. (E)
Quantification of average
number of NSCs per all,
Unipotent and Multipotent
clones at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM. (N = 41-96 clones,
Table S1). ANOVA with Tukey’s multiple comparisons test. (F) Quantification of average number
of intermediate progenitor cells/neuroblasts (IPCs/NB) and neurons (N) per neuron-containing
clones at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM. (N = 41-96 clones,
Table S1). ANOVA with Tukey’s multiple comparisons test. (G) Quantification of average
number of transition astrocytes (TA) and mature astrocytes (A) per astrocyte-containing clones at
30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM. (N = 41-96 clones, Table S1).
ANOVA with Tukey’s multiple comparisons test. (H) Quantification of average number of
92
oligodendrocyte progenitor cells (OPC) and mature oligodendrocytes (OL) per oligo-containing
clones at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM. (N = 41-96 clones,
Table S1). ANOVA with Tukey’s multiple comparisons test. (I) Quantification of the relative
fraction of N + A clone (A), N + O clone (O), and N + A + O clone (A+O) at 30-, 90-, 180- and
365-days post tamoxifen injection. (N = 41-96 clones, Table S1). (J) Quantification of the overall
fraction of three different types of gliogenesis clones (A, ~79%; O, ~5%; A + O, ~16%) at 30-,
90-, 180- and 365-days post tamoxifen injection. (N = 41-96 clones, Table S1).
Scale bar, 50μm (G). *p<0.05, **p<0.01, ***p<0.001, n.s - not significant.
Computational modeling of NSC lineages indicates their lineage independency
Simple mathematical modelling can be used to understand the changes in the distribution
of NSCs per clone over time, as well as related quantities such as fraction of no-NSC clones,
average number of NSC per NSC-containing clone. Here we probed a single population model of
NSCs to one with two independent populations.
In a first model, consider a single population of neural stem cells remaining at the end of
the tissue ontogenesis (after P15 timepoint). These NSCs stochastically duplicate at rate λp or
differentiate with rate λ(1-p). At the population level, over time the average number NSCs per
clone (R(t)) exponentially decays for p < ½ and increases for p > ½, following R(t) = R0e
-λ(1-2p)t
.
However, in the clonal data the decay in average number of NSCs per clone after P15 is slower
than exponential decay (Figure 4.2D-E). While, for this one population model can reasonably
reproduce the fraction of extinct clones after P15, the average number of NSCs per surviving clone
is substantially overestimated in intermediate times. Further, exponential decay in average NSC
clone size is always predicted if differentiation is favored over division, which is not consistent
with the data (Figure 4.2D), which follows a sum of two exponentials.
Some features of the data are consistent with the existence of multiple populations of neural
stem cells: a slow cycling but renewing population, and another disposable population whereupon
stochastic activation and proliferation, the NSC is lost. An alternative model is to consider two
populations, with differing duplication and differentiation rates. The initial distributions of the
93
slow-renewing population are chosen to the NSC distribution of neuron + glia clones at P15, and
the initial distribution of the fast-renewing population is chosen to be the NSC distribution of
neuron-only clones at P15. After the developmental phase is complete (P30 and later), only clones
containing adult NSCs are surviving. These NSCs can both self-renew and have bursts of
neurogenesis and/or gliogenesis. The observed decrease in correlation towards no correlation (at
P90) is consistent with the existence of disposable and slow-renewing populations of adult NSCs.
By P365, no disposable clones remain, so surviving clones consistent of the long-lived population
containing 1-2 neurons.
Although clone numbers are small, analysis of the cumulative distribution of neuron
number in NSC-containing clones reveals an approximately exponential-like dependence at P30,
similar to extinct clones but with an average size that is larger by a factor of around 2. However,
at longer times, there is a gradual departure from exponential. At P90, there is a preponderance of
oversized clones compared to exponential while, at P180, this trend is reversed with a
preponderance of small clone sizes. Consistently, the average neuron number per clone in NSC-
containing clones shows a complex non-monotonic dependence (Figure S4.3E), with a magnitude
approximately twice larger than that of no NSC-containing clones. A closer analysis shows that
NSC-containing clones show a small reduction in the average neuron number per clone at the
longest time point, while that of no NSC-containing clones shows a rise (Figure 4.3F, S4.3E).
94
Figure 4.4.
Evolution of the
NSC distribution in
model of two
populations of NSC.
For (A-C) the black
line corresponds to
experimental data, red
dotted line
corresponds to model.
The blue (green)
dotted line
corresponds to the
model for only slow-
renewing (disposable)
populations
respectively. (A)
Fraction of clones
containing no NSCs.
(B) Average number
of RGP/NSCs per
clone for surviving
clones. (C) Average
number of NSCs per
clone for all clones. (D) Distribution of NSCs for clones with at least one NSC for the model. (E)
Summary of the developmental dynamics of dentate neural progenitor lineages.
The average NSC per clone over time, following sum of exponentials provides is consistent
with two populations of NSCs which are lost at different rates (Figure 4.4E). The two-population
model can replicate the observed evolution of the NSC clone size distribution better than the one-
population model.
Discussion
NSCs in the developing hippocampus generate neurons as well as glial cells, including
astrocytes and oligodendrocytes; however, the precise behavior of NSCs and the underlying
gliogenic program remain largely unknown. Previous studies have noted the heterogeneity of
95
neural stem cells in the adult brain yet embryonic origins of adult neural progenitor cells in the
mammalian dentate gyrus is yet to be investigated.
Using single cell clonal linage tracing and computational modeling our study reveals
multiple developmental origins of adult neural progenitors in the mouse dentate gyrus and further
suggests an existence of parallel lineages in the mouse hippocampus. It further suggests that one
NSC lineage follows ‘‘continuous’’ model that has been previously suggested (D. A. Berg et al.,
2019) in which a single precursor population continuously and exclusively contributes to dentate
neurogenesis starting from early embryonic stages to adulthood. In contrast second lineage exhibit
similarities with previously described “Sequential” model (A & A, 2009) where NSCs first self-
renew, followed by constant granule neuron generation and glia generation during the postnatal
period, and then the residual NSCs are converted into adult progenitors to generate specific
neurons and astrocytes in the adult brain. However, additional work is required to find the right
parameters, which may make the modelling closer to the experimental data. Together, our findings
offer insight into the origin and behavior of adult neural progenitors during development and aging
and provide a tool to facilitate potential identification and manipulation of these NSCs. Our results
open the possibilities for future studies to investigate precisely how these cells are regulated under
physiological and pathological conditions.
One important question that remain in the field is what is precise lineage relationships of
different neural stem cell subpopulations that exist in adult mouse hippocampus (Bonaguidi et al.,
2016b). Major limitations in studying developmental precursors are a lack of specific tools that
can adapt to a growing tissue. Recent studies using single cell RNA-sequencing and bioinformatics
attempted to reconstruct developmental trajectories but unable to provide definitive lineage
relationships (AB et al., 2018; Hochgerner, Zeisel, Lönnerberg, & Linnarsson, 2018; La Manno et
96
al., 2018). In contrast, our lineage tracing and computational modeling showed that embryonic
Gli+ precursors give rise not only hippocampal granule neurons, but also glia (astrocyte and
oligodendrocytes). Gli1-CreER
T2
mouse line enriches for dentate neural progenitors from E17.5
to adult, allowing us to identify an embryonic origin of adult NSCs and describe lineage
relationships. The Gli1-CreER
T2
mouse line provides a tool to prospectively label and manipulate
precursors to adult dentate neural progenitors in vivo, and computational modeling of dentate
progenitors across development and into adulthood provide a useful resource for future studies.
97
Materials and Methods
Animals and Tamoxifen Administration
All animal procedures were performed in accordance with institutional guidelines of
University of Southern California Keck School of Medicine and protocol (20287) approved by
Institutional Animal Care and Use Committee (IACUC). All mice used in the study were
backcrossed to the C57BL/6 background to ensure the reproducibility of clonal induction with
specific doses of tamoxifen. Animals were housed in a 12-hour light/12-hour dark cycle with free
access to food.
Gli1-CreER
T2
mice (Ahn & Joyner, 2005), were used to clonally label RGLs. The
following genetically modified mice were originally purchased from Jackson Labs: Confetti
f/f
(StrainGt(ROSA)26Sor
tm1(CAG-Brainbow2.1)Cle
/J).
Gli1-CreER
T2
mice were crossed to fluorescent reporter mice for clonal analysis. Gli1-
CreER
T2
::Confetti
f/+
mice were generated by breeding Gli1-CreERT2+/- mice with Confetti
f/f
mice, or by crossing Gli1-CreER
T2+/-
::Confetti
f/f
mice with wild-type C57BL/6 mice.
At least 3 animals were checked for each timepoint to ensure there was no recombination
in the adult SGZ in the absence of tamoxifen. A stock of tamoxifen (14.4 mg/mL) was prepared in
a 5:1 ratio of corn oil to ethanol at 37C with occasional vortexing. A single tamoxifen was
intraperitoneally injected into timed pregnant female Gli1-CreER
T2+/-
::Confetti
f/f
mice to label
embryos at day E17.5 and sacrificed animals at multiple postnatal days for immunohistochemistry
to visualize single cells and their progenies (Table S4.1). A single i.p. injection of tamoxifen
yielded an abundant number of clones that were simultaneously discernable (5-6 clones per mouse
brain, 1 clone per color per hemisphere). Injected animals showed no signs of distress.
98
Primer sets from original publications were used to identify genetically modified mice
(Ahn and Joyner, 2005; Balordi and Fishell, 2007; Kim et al., 2011; Lemberger et al., 2007;
Muzumdar et al., 2007). Genomic tail DNA was isolated in a 25mM NaOH, 0.2 mM EDTA
solution and ran for 35 PCR cycles.
Immunostaining, Confocal Imaging, and Processing
Mice were anesthetized with isoflurane gas and underwent transcardial perfusion with
saline followed by 4% paraformaldehyde. Brains were post-fixed overnight in 4%
paraformaldehyde and then immersed in 30% sucrose for a subsequent 48 hours prior to sectioning.
Brains were sectioned into 45µm coronal sections through the entire dentate gyrus.
Immunohistology was performed with antibodies as previously described (Bonaguidi et al., 2011)
on sections in serial order using custom, in-house staining chambers. Brain sections were washed
in TBS with 0.3% Triton-X100 prior to staining and mounting. Goat anti-GFP (1:1000), rabbit
anti-RFP (1:1000) primary antibodies were used. GFP+ or RFP+ cells were identified with a Zeiss
AxioObserver.A1 microscope and were acquired as a z-stack on a Zeiss LSM700 confocal system
under 40X or 63X magnification (stitching were done at 0.5 overlap). Morphological analysis was
done using Imaris 8.0 Software.
P0/P1 experiment
Pregnant females were anesthetized with isoflurane gas and underwent transcardial
perfusion with saline followed by 4% paraformaldehyde, pups were extracted after perfusion is
complete. Brains were post-fixed overnight in 4% paraformaldehyde and then immersed in 30%
sucrose for a subsequent 48 hours prior to sectioning. Brains were sectioned using cryostat into
25µm sagittal sections through the entire dentate gyrus. Cells were then counted throughout the
entire dentate for a stereological analysis and every section was processed for clonal analysis.
99
GFP+ or RFP+ cells were identified with a Zeiss AxioObserver.A1 microscope and were acquired
as a z-stack on a Zeiss LSM700 confocal system under 40X magnification (stitching were done at
0.5 overlap). Morphological analysis was done using Imaris 8.0 or Zeiss Zen (blue edition) 1.0
Software.
Clonal Analysis
Clonal analysis and categorization of cell types by morphological and immunohistological
criteria were consistent with prior criteria (Bonaguidi et al., 2011). Analyzed dentate volume
included the stratum granulosum (granule cell layer), and SGZ. Serial sections were first screened
for candidate clones, which were defined as possessing at least (1) an NSC (RGL), (2) neuronal
cell(s) in close spatial proximity, or (3) astroglia in close spatial proximity to other astroglia or
neuronal cells. Approximately 5-6 clones per mouse brain, 1 clone per color per hemisphere
allowed for clonal analysis based on prior computer simulations (Bonaguidi et al., 2011). Clones
were randomly induced throughout the dentate.
Dividing clones were marked with two separate nuclei by DAPI under confocal microscopy.
R26R – Confetti is a Stochastic Multicolor Cre-reporter. Upon Cre activation, the neomycin
roadblock is excited, while the Brainbow 2.1 (Livet, J et al. 2007) recombines in a random fashion
to four possible outcomes. GFP is a nuclear, CFP is membrane associated, RFP and YFP are
cytoplasmic. Clones with more than one NSC and progeny were classified by the lineage produced:
neuronal, astroglial, or both. NSC maintenance was assessed as a percentage of clones that
contained at least one NSC. (Figure 4.1C-D)
100
Computational modeling
Suppose the slow-cycling population has duplication rate λp, and differentiation rate λ(1-
p), whereas the fast-cycling population has duplication rate γq, and differentiation rate γ(1-q). First,
we consider the case where the populations are lineage independent. In this case at the population
level the average number NSCs per clone (R(t) = Rslow(t) + Rfast(t)) decays (for p < ½, q < ½) as
R(t) = Rslow(t) + Rfast(t) = Rslow(0) e
-λ(1-2p)t
+ Rfast(0) e
-γ(1-2q)t
.
The probability density function of number of NSCs per clone is described by the Master
equation (Eq. 1). The first line corresponds to the NSC duplication and loss term for the slow-
cycling population, and the second line corresponds the NSC duplication and loss term for the fast-
cycling population.
(1)
An example of results obtained under the model of two populations of NSCs is shown in
(Figure 4A-D). The parameters are chosen by fixing the decay rates λ(1-2p) = 1.67 ± 0.19 (year)^-
1, and γ(1-2q) = (year)^-1, based off the fit of Figure 2D. Then roughly based off (Bottes et al.,
2021) select symmetric division rates p = 0.15, q = 0. The initial distributions of the slow-renewing
population are chosen to the NSC distribution of neuron + glia clones at P15, and the initial
distribution of the fast-renewing population is chosen to be the NSC distribution of neuron-only
clones at P15.
Quantification and statistical analysis
Statistics were performed as indicated in each Figure legends. For all in vivo experiments,
the number of independent experimental replicates are indicated in figure legends, with n
representing experimental replicates (clones). All multiple comparisons were performed with
101
GraphPad Prism’s one-way ANOVA function with Bonferroni’s multiple comparisons test. All
clones observed at each time point were treated as statistically equivalent. No randomization or
blinding was used in the animal studies. Mouse dentate gyri with exceedingly high (>4) clones per
color per hemisphere were omitted from the study. Sample sizes were estimated in accordance to
prior clonal studies (Bonaguidi et al., 2011; Song et al., 2012). Error bars in the study represent
the standard errors in mean frequencies, calculated as Z(1−)/ where p is the frequency of
a given characteristic and n the number of clones considered. Mathematical modeling was
performed by weighted least squares using custom-written MATLAB scripts.
Data and code availability
The codes to implement computational modeling are available on GitHub.
102
Chapter V: Conclusion and Final Thoughts
The United States is experiencing rapid growth of the population over 65 years old. As a
result, age-associated chronic diseases such as cardiovascular, osteoporosis and dementia pose
massive societal and economic challenges (Lopez & Kuller, 2019). The adult brain is especially
susceptible to progressive loss of function during normal aging and is made worse by
neurodegeneration ("Global, regional, and national burden of Alzheimer's disease and other
dementias, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016," 2019;
Lopez & Kuller, 2019). While aging has been traditionally appreciated in chronological terms,
strategies to diagnose and prevent declining function at young ages have become increasingly
popular (Belsky et al., 2015; Gillman, 2005). Biological aging occurs in a gradual and
asynchronous manner where specific cells individually and then collectively deteriorate. The
ability to ensure healthy function depends on individual cells that act in a coordinated manner for
tissue maintenance (Belsky et al., 2015).
Generally, an age-related decline in neurogenesis is thought to result from a smaller
number of NSCs, or their reduced capacity to generate progeny (Artegiani & Calegari, 2012). We
reasoned that cellular heterogeneity within the NSC pool masks the ability to uncover precisely
which mechanisms create NSC dysfunction. We resolved NSC heterogeneity into two radial glia-
like cell subpopulations. We found that clonally labelled Ascl1
#
-NSCs represent a rapidly
depleting neuronal fate-biased population, consistent with a developmental-like non-homeostatic
program (Pilz et al., 2018). Meanwhile, individual NSCs marked by Nestin-CreER
T2
(Nestin
#
) are
slow cycling, generate neurons and astrocytes, and self-renew. These NSCs have not been
observed to replenish Ascl1
#
-NSCs.
103
Next, by combining in vivo single cell clonal lineage tracing, computational modeling
approaches, single cell RNA-sequencing (scRNA-seq) and systems level data science we
comprehensively investigated neural stem cell adaptation and restoration during aging. we have
resolved the behavior of two NSC sub-populations. I identified that while Ascl1-NSCs are
consistently non-homeostatic, we determined that Nestin-NSCs are homeostatic for a period and
then naturally declines in number and remaining NSCs increase their quiescence at all investigated
chronological ages. We determined that NSCs are maintained for months but are pushed out of
homeostasis by lengthening NSC quiescence. Next, I used single-cell RNA-sequencing to provide
a new way to identify cellular aging of stem cells at the molecular level. Remarkably, we identified
numerous pillars of cellular aging including inflammation, cell signaling, epigenetics and DNA
repair – all in NSCs from the mature (5-month-old mouse) brain. I showed that targeting
mechanisms associated with the initial loss of NSC homeostasis can overcome age-related NSC
dysfunction later in life. I used a clinically relevant drug Imatinib (Abl1/2 inhibitor) as a strategy
to overcome NSC cellular aging. Indeed, intracranial infusion into the middle-aged brain was
sufficient to overcome deep NSC quiescence and restore NSC proliferation to younger levels.
In somatic tissues, it is common for one precursor population to be predominantly utilized
under physiological conditions with another cell population recruited in pathological conditions.
In response to tissue damage, a “reserve” cell population can assume a new role and contribute to
repair, despite lacking stem cell properties during normal tissue homeostasis. Surprisingly, our
study reveals the ongoing contribution to cell genesis from two parallel adult neural stem cell
populations that exhibit similar morphology, share certain common lineages do not
compartmentalize along spatial boundaries under physiological conditions, and undergo different
cell fate changes with age and converge in one similarly behave population. In addition, we find
104
that both populations exhibit unexpected plasticity - differentially contribute to cell replacement
after chemical injury. A normally stochastic neural stem cell switches to repeated neurogenic
divisions to regenerate the lost neuronal lineage. Meanwhile, a neurogenic-biased stem cell
undergoes symmetric divisions to replace stem cells killed by chemotherapy. These findings
identify cellular substrates of a coordinated cell response towards healing damaged neural tissue.
Finally, we wanted to reconstruct an identify the history of these lineages, developmental
origins, and their contribution to an early aging phenotype. We combined in vivo single cell clonal
lineage tracing with computational modeling and identified that that neural progenitor cells labeled
by Gli1
#
during late embryonic stage give rise to two parallel lineages, one unipotent population
that divide rapidly and continuously generates only granule neurons, and multipotent population
that is slow cycling and generate neurons, astrocytes and/or oligodendrocytes through
development and adulthood. In addition, we found that one lineage follows “continuous model” in
which a single precursor continuously and exclusively contributes to dentate neurogenesis starting
from embryonic stages to adulthood, while another lineage follows a “sequential model” in which
NSCs successively generate neurons and glia during development, and then retain their neural
stem cell function in the adult brain.
In conclusion, my work has defined neural stem aging from development to old age. To
the best of my knowledge, my work represents the first of its kind in the adult brain. Studying
NSCs at the single cell level uniquely permits comprehensive computational analysis (Teresa
Krieger & Benjamin D. Simons, 2015) of stem cell behavior, including cell cycle kinetics, fate
choices, expansion and depletion to determine the natural origins of neurogenesis decline. Neural
stem cells generate and repair memory parts of the brain. Our innovative approaches bring together
multiple fields including cell biology, genomics, geroscience, mathematics, stem cells and
105
neuroscience. These findings not only significantly advance our basic knowledge of adult neural
stem cells, but also have implications for regenerative medicine and reinterpreting existing data
within the stem cell biology field in general.
106
Supplemental Information
Figure S2.1. Population analysis and clonal lineage-tracing of individual NSCs in the adult
mouse dentate gyrus.
(A) Quantification of the total number of active NSCs in the dentate gyrus across ages. N=4-6
mice; Values represent mean ± SEM. *p<0.05, **p<0.01, ***p<0.001, n.s. - not significant;
ANOVA with Bonferroni post-hoc test. (B) Quantification of NSC homeostasis duration at 6-
month-old. Ascl1#-NSCs are rapidly depleted, while Nestin#-NSCs maintain as stem cells for
months before eventually differentiating. Values represent mean ± SEM. (N = 29-104 clones,
Table S3.2). *p<0.05, **p<0.01, ***p<0.001, n.s. - not significant; two-way ANOVA with
Tukey’s multiple comparisons test.
NSC-NSC
NSC-N
NSC-A
0%
10%
20%
30%
40%
50%
60%
Fate choice of NSC divisions (%)
Nestin
#
2mo --> 30dpi
6mo --> 30dpi
***
12mo --> 30dpi
2mo --> 60dpi
6mo --> 60dpi
12mo --> 60dpi
***
**
***
***
***
**
***
Figure S1
A B C D
E F G
H I J
30dpi
60dpi
0
1
2
3
4
5
6
Average number of cells
per all clones
Nestin#
2-3 mo
6-7 mo
12 mo
***
*
***
***
30dpi
60dpi
0
1
2
3
4
5
6
Average number of cells
per NSC containing clone
Nestin#
2-3 mo
6-7 mo
**
**
12 mo
**
***
6 7 8
0.0
0.2
0.4
0.6
0.8
1.0
Age (months)
Average NSC number
per all clones
Nestin
#
Ascl1
#
*
0 10 20 30 40
0%
20%
40%
60%
80%
100%
DPI
Clonal NSC maitenance (%)
2-3 mo
6-7 mo
Ascl1#
12 mo
20 30 40 50 60 70
0%
20%
40%
60%
80%
100%
DPI
Clonal NSC maitenance (%)
2-3 mo
6-7 mo
Nestin#
12 mo
3
6
9
12
0
2000
4000
6000
8000
10000
Age (Months)
MCM2
+
NSCs/ mm
3
***
***
***
***
n.s
NSC
NSC-NSC
NSC-N
NSC-A
NSC-A-N
no-NSC
0%
20%
40%
60%
80%
Clone composition (%)
1dpi
3dpi
7dpi
Ascl1
#
6mo
n.s n.s
n.s
n.s
n.s
0 0
NSC
NSC-NSC
NSC-N
NSC-A
NSC-A-N
no-NSC
0%
20%
40%
60%
80%
Clone composition (%)
1dpi
7dpi
30dpi
Nestin
#
6mo
n.s
n.s n.s n.s
0 0
GFP GFP
Nestin
#
Ascl1
#
1
1
1
1
2
3
4
3
3
3
4
3
107
(C) Clonal NSC maintenance from Nestin
#
clones acquired across multiple ages (2, 6, 12 months
old animals) at multiple days post tamoxifen injection (Table S3.1-2). Shown is a summary
quantification of the percent of clones which contains NSCs. (D) Clonal NSC maintenance from
Ascl1
#
clones acquired across multiple ages (2, 6, 12 months old animals) at multiple days post
tamoxifen injection (Table S3.1-2). Shown is a summary quantification of the percent of clones
which contains NSCs. (E) Sample confocal images of NSC subpopulations in the adult
hippocampus. (left) Nestin::CreER (Nestin
#
)-labeled multipotential NSCs. (right) Developmental-
like NSCs labeled by Ascl1::CreER (Ascl1
#
). NSC = neural stem cell (1), A= astroglial lineage
(2), intermediate progenitor cells (IPCs) (3) and neurons (4). Scale bar, 10μm. (F) Quantification
of the clone composition for Nestin
#
clones at day 1-, 7- and 30-days post-tamoxifen injection into
6-month-old mice. NSC = radial glia-like neural stem cells; A=Astroglial lineage; N=neuronal
lineage, 0 – no observed phenotype. Values represent mean ± SEM (N = 29-104 clones, see details
in Table S3.4; *p<0.05, **p<0.01, ***p<0.001, n.s. - not significant; ANOVA with Bonferroni
post-hoc test); (G) Quantification of the clone composition for Ascl1
#
clones at day 1-, 3- and 7-
days post-tamoxifen injection into 6-month-old mice. NSC = radial glia-like neural stem cells;
A=Astroglial lineage; N=neuronal lineage, 0 – no observed phenotype. Values represent mean ±
SEM (N = 70-85 clones, Table S3.2; *p<0.05, **p<0.01, ***p<0.001, n.s. - not significant;
ANOVA with Bonferroni post-hoc test); (H) Quantification of NSC self-renewing cell fate
divisions at 30- and 60-days post-tamoxifen injection into 2, 6 and 12 months old Nestin
#
mice.
NSC = radial glia-like neural stem cells; A=Astroglial lineage; N=neuronal lineage. Values
represent mean ± SEM (N = 29-104 clones, Table S3.2; *p<0.05, **p<0.01, ***p<0.001, n.s. - not
significant; ANOVA with Bonferroni post-hoc test); (I) Quantification of average number of cells
per all clones at 30- and 60-days post-tamoxifen injection into 2, 6 and 12 months old Nestin
#
mice. (N = 30-69 clones, Table S3.2; *p<0.05, **p<0.01, ***p<0.001, n.s. - not significant;
ANOVA with Bonferroni post-hoc test). (J) Quantification of average number of cells per NSC-
containing clones at 30- and 60-days post-tamoxifen injection into 2, 6 and 12 months old Nestin
#
mice. (N = 30-69 clones, Table S3.2; *p<0.05, **p<0.01, ***p<0.001, n.s. - not significant;
ANOVA with Bonferroni post-hoc test). Related to Figure 2.1 and 2.2.
108
Figure S2.2. Computational modeling of Nestin
#
-NSC and Ascl1
#
-NSC.
(A) Model of stochastic Nestin
#
-NSC fate dynamics within a small stem cell niche. NSC clonal
content and survival increases in 6 and 12-month-old mice, while neuronal and astroglial content
decreases compared to 2-month-old mice. Red=radial glia-like NSC content, blue=astroglial
content, green=neuronal content; Line=model, dots=experimental data. (B) Least-fit squares
showing correspondence between the simulated model and clonal content from Nestin
#
-NSCs
traced in 2, 6 and 12 months-old mice. (C) Model of a developmental-like stem cell fate program
in Ascl1
#
-NSCs. NSC clonal survival decreases in 6 and 12-month-old mice, while neuronal
content increases compared to 2-month-old mice. Red=radial glia-like NSC content,
blue=astroglial content, green=neuronal content. (D) Least-fit squares showing correspondence
between the simulated model and clonal content from Ascl1
#
-NSCs traced in 2, 6 and 12 months-
old mice. Related to Figures 2.1 and 2.2.
D
● ● ●
●
● ●
●
●
● ● ●
●
● ● ● ● ● ●
� ���
�
�
�
�
���� ����� ����� �������
●
●
●
●
●
●
� ���
�
�
�
�
��� ������/ ����� ����
●
●
●
●
●
●
� ���
�
���� ( ��� )
��� ������� / ��� �����
●
●
●
● ●
●
● ● ●
� ���
�
�
�
�
●
●
●
� ���
�
�
�
�
●
●
●
� ���
�
���� ( ��� )
2 months
Legend
2 months
Ascl1
6 months
A
●
● ●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
●
� ��� ��� ���
�
�
���� ����� ����� �������
●
●
●
●
●
●
� ��� ��� ���
�
�
�
�
��� ������/ ����� ����
●
●
●
●
●
●
� ��� ��� ���
�
���� ( ��� )
��� ������� / ��� �����
●
●
● ●
●
●
● ● ●
� ��� ��� ���
�
�
●
●
●
� ��� ��� ���
�
�
�
�
● ● ●
� ��� ��� ���
�
���� ( ��� )
2 months
Nestin
6 months 12 months
6 months
RGL
neuronal lineage
glial lineage
experiment
simulation
RGL content
1
2
3
4
5
0–2 3–5 6–8 9–11 12–15
neuronal content
2 months 6 months
RGL content
1
2
3
4
5
0–2 3–5 6–8 9–11 12–15
neuronal content
RGL content
1
2
3
4
5
0–2 3–5 6–8 9–11 12–15
glial content
E
B C
������ (� �� )� � ��� ������ ( � �� ) � � ��� ������ (� �� )� �� ��� ������ (� �� )� �� ���
������ (� �� ) � ��� ��� ������ ( � �� )� ��� ��� ������ ( � �� ) � � ��� ������ (� �� )� �� ���
������ ( � �� )� �� ��� ������ ( �� �� ) � �� ��� ������ (�� �� )� �� ���
������ (� �� )� � ��� ������ (� �� ) � � ��� ������ (� �� )� �� ��� ������ (� �� ) � �� ���
������ (� �� )� ��� ��� ������ (� �� )� ��� ��� ������ ( � �� ) � � ��� ������ (� �� ) � �� ���
������ (� �� )� �� ��� ������ (�� �� )� �� ��� ������ ( �� �� ) � �� ���
����( � �� ) � � ��� ����( � �� )� � ��� ����( � �� ) � � ��� ����( � �� )� �� ���
����(� �� )� �� ��� ����(� �� )� ��� ��� ����( � �� ) � � ��� ����( � �� ) � � ���
����(� �� )� �� ��� ����( �� �� ) � � ��� ����(�� �� )� �� ���
● ●
●
●
● ●
� ���
�
�
�
�
● ●
� ���
�
�
�
�
●
●
� ���
�
���� (��� )
●
●
● ●
●
●
� ��� ��� ���
�
�
●
●
� ��� ��� ���
�
�
�
�
●
●
� ��� ��� ���
�
���� (��� )
12 months
Figure S2
109
Figure S2.3. Experimental design of scRNA-seq experiment.
(A) Experimental steps to generate single cell transcriptomes from 2 months-old and 4.5 months-
old Nestin::CFP mice. Scale bar, 50μm. (B) Fluorescence activated cell sorting scheme used to
isolate NestinCFP
+
cells from the dentate gyrus of 2 months-old and 4.5 months-old mice.
Negative controls (C57BL/6 - wild type animals) were used as indicated. Related to Figure 2.3.
B
A
Figure S3
110
Figure S2.4. Differential gene expression analysis.
(A) Principal component plot highlighting quiescent cells marked by log-mean expression of
Aldoc, Aqpr4, Id3, Hes1. (B) Principal component plot highlighting proliferating cells marked by
log-mean expression of PCNA, Mcm6, Mki67. (C) Principal component plot highlighting
intermediate progenitor cells marked by log-mean expression Eomes/Tbr2, Sox11, Stmn1. (D)
Biological coefficient of variation between 2 months-old and 4.5 months-old across average log
counts per million (CPM). A greater degree of relative variation occurs at lower CPM levels justify
a need for edgeR to call differentially expressed genes. (E) Volcano plot of edgeR-called
differentially expressed genes between quiescent NSCs from 2 months-old and 4.5 months-old
mice. FDR-corrected p-values are shown. (F) Mean-mean gene expression scatter plot between
NSCs from 2 months-old and 4.5 months-old mice highlighting differentially expressed genes
(red). (G) String network graph depicting age-related changes upregulated in 4.5-month-old vs 2-
month-old quiescent NSCs. Shown genes have 1+ connections to biological processes. (H) String
network graph depicting age-related changes downregulated in 4.5-month-old vs 2-month-old
quiescent NSCs. Shown genes have 1+ connections to biological processes. (I) Schematic
illustration of the experimental design for intracranial drug infusion of Imatinib and Vehicle
Control. (J) Representative confocal images of c-Abl and Nestin co-expression across multiple
ages (2, 6, 10-month-old). Arrow - Nestin
+
, Arrowhead – c-Abl
+
. Scale bar, 100μm (20μm for the
insert). (K) Representative confocal images of c-Abl and Nestin co-expression in Vehicle-Control
and Imatinib-treated brains. Arrow - Nestin
+
, Arrowhead – c-Abl
+
. Scale bar, 100μm (20μm for
the insert). Related to Figure 2.3 and 2.4.
Biological coefficient
of variation
2mo log2(expr+1) log2(FC)
Volcano Plot 4.5mo/2mo
Quiescent Proliferating IPC
Mean-mean
4.5mo log2(expr+1)
-log10(FDR)
Average log CPM
Figure S4
G H
D E F J
A B C I
c-Abl Nestin DAPI Nestin
Nestin
Nestin
c-Abl
c-Abl
c-Abl
Merge
Merge
Merge
Nestin c-Abl Merge
Nestin c-Abl Merge
c-Abl Nestin DAPI
2mo
6mo
10mo
10mo Imatinib
10mo Vehicle
K
111
Table S2.1. Tamoxifen doses used to achieve clonal recombination among various promoter,
reporter and ages contexts over the analyzed time course.
Related to Figures 2.1-2.2 and S2.1-S2.2.
Promoter Reporter Age
(month
- old)
[Tamoxifen] Chase Number of
clones
Figure(s)
Ascl1::Cre
ER
T2
ROSA-
YFP
2 78 mg/kg 1, 3, 7,
30, 60,
120 dpi
1dpi = 40
3dpi = 83
7dpi = 92
30dpi = 70
60dpi = 59
120dpi = 80
2.1, S2.1
Ascl1::Cre
ER
T2
mT/mG 2 78 mg/kg 1 dpi 1dpi = 45
2.1, S2.1
Ascl1::Cre
ER
T2
mT/mG 6 233.1 mg/kg 3, 7, 30
dpi
3 dpi = 63
7dpi = 101
30dpi = 78
2.1, 2.2,
S2.1-2.2
Ascl1::Cre
ER
T2
Confetti 6 233.1 mg/kg 1 dpi 1 dpi = 30
2.1, 2.2,
S2.1-2.2
Ascl1::Cre
ER
T2
mT/mG 12 233.1 mg/kg 7, 30
dpi
7dpi = 52
30 dpi = 69
2.1, 2.2,
S2.1-2.2
Ascl1::Cre
ER
T2
ROSA-
YFP
12 233.1 mg/kg 30 dpi 30 dpi = 35 2.1, 2.2,
S2.1-2.2
Nestin::Cr
eER
T2
Z/EG 2 62 mg/kg 2, 7,
30, 60,
120 dpi
2dpi = 34
7dpi = 40
30dpi = 42
60dpi = 45
120dpi = 69
365dpi = 30
2.1, S2.1
Nestin::Cr
eER
T2
MADM 2 62 mg/kg 2, 60
dpi
2dpi = 13
60dpi = 24
2.1, S2.1
Nestin::Cr
eER
T2
Confetti 6 66.6 mg/kg 1, 7, 30
dpi
1dpi = 29
7dpi = 68
30dpi = 30
2.1, 2.2,
S2.1-2.2
112
Nestin::Cr
eER
T2
MADM 6 233.1 mg/kg 30, 60
dpi
30dpi = 34
60dpi = 53
2.1, 2.2,
S2.1-2.2
Nestin::Cr
eER
T2
Confetti 12 99.6 mg/kg 30, 60
dpi
30dpi = 68
60dpi = 45
2.1, 2.2,
S2.1-2.2
113
Table S2.2. Number of clones for clonal analysis. Summary of number of all clones, NSC-
containing clones, animals across different ages and tracing time points for clonal analysis.
Related to Figures 2.1-2.2 and S2.1-S2.2.
Total
number of
clones
Total number of
NSC-containing
clones
Total number
of non NSC-
containing
clones
Number of
animals
Nestin
#
-2mo
2 dpi 47 36 11 11
7 dpi 40 34 6 3
30 dpi 54 45 9 6
60 dpi 69 47 22 11
120 dpi 69 28 41 5
365 dpi 30 10 20 4
Nestin
#
-6mo
1 dpi 29 19 10 3
7 dpi 68 44 24 4
30 dpi 64 53 11 5
60 dpi 53 35 19 6
Nestin
#
-12mo
30 dpi 68 46 22 3
60 dpi 46 38 8 3
Clones Animals
Total of Nestin
#
-lineage tracing 637 64
Total
number of
clones
Total number of
NSC-containing
clones
Total number
of non NSC-
containing
clones
Number of
animals
Ascl1
#
-2mo
1dpi 85 71 14 7
3 dpi 83 61 22 6
7 dpi 93 65 28 6
30 dpi 70 39 31 4
60 dpi 59 24 35 4
120 dpi 80 23 57 7
Ascl1
#
-6mo
1dpi 30 15 15 3
3dpi 89 30 59 5
7dpi 101 57 44 9
114
30dpi 78 20 58 6
Ascl1
#
-12mo
7dpi 52 28 24 3
30dpi 104 34 70 6
Clones Animals
Total of Ascl1
#
-lineage tracing 924 66
Total number of clones and animals 1561 130
115
Table S2.3. Parameter list and best fit parameters for the theoretical mode.
Related to Figures 2.1-2.2 and Figures S2.1-S2.2.
Parameter Fit value Description
Ascl1-targeted subpopulation (developmental-like program model)
λ (2m) 0.038 1/d cycle rate at 2 months of age
λ (6m) 0.068 1/d cycle rate at 6 months of age
λ (12m) 0.074 1/d cycle rate at 12 months of age
λ1 (2m) 0.47 1/d activation rate after induction at 2 months of
age
λ1 (6m) 1.2 1/d activation rate after induction at 6 months of
age
λ1 (12m) 58 1/d activation rate after induction at 12 months of
age
R0 1 number of cell division rounds in the
proliferative phase
R1 2 number of cell division rounds in the
neurogenic phase
mN 3.1 avg. amount of neuronal progeny generated
per differentiating division
ωN 0.039 1/d neuronal death rate
q 94 % probability to induce an RGL
Nestin-targeted subpopulation (small niche-like model)
λ (2m) 0.021 1/d cycle rate at 2 months of age
λ (6m) 0.013 1/d cycle rate at 6 months of age
λ (12m) 0.008 1/d cycle rate at 12 months of age
2 niche capacity
p 51 % base probability for duplication (unless niche
is fully occupied)
mN 27 avg. amount of neuronal progeny generated
per differentiating division
mG 1.8 avg. amount of glial progeny generated per
differentiating division
ωN 0.2 1/d neuronal death rate
ωG 0.012 1/d glial death rate
q 79 % probability to induce an RGL (as compared to
a more differentiated cell)
116
Table S2.4. Primer sets that were used for genotyping.
Related to Figure 2.1-2.3 and S2.1-S2.3.
Name Primer (5'-3')
NestinCreER_F TGG CAG GAC ATG CTA CCT C
NestinCreER_R AGG CAA ATT TTG GTG TAC GG
Ascl1_KI_F AAC TTT CCT CCG GGG CTC GTT TC
Ascl1_KI_R CGC CTG GCG ATC CCT GAA CAT G
Rosa10_F CTC TGC TGC CTC CTG GCT TCT
Rosa11_R CGA GGC GGA TCA CAA GCA ATA
Rosa4_R TCA ATG GGC GGG GGT CGT T
Confetti_R CCA GAT GAC TAC CTA TCC TC
Confetti_F_WT CTC CTG GCT TCT GAG GAC C
Confetti_F_spec GAA TTA ATT CCG GTA TAA CTT CG
MADM11_het-homo_F TTC CCT TTC TGC TTC ATC TTG C
MADM11_het-homo_R_WT TGG AGG AGG ACA AAC TGG TCA C
MADM11_het-homo_R_MUT TCA ATG GGC GGG GGT CGT T
117
Figure S3.1. Multiple strategies for targeting and clonal lineage-tracing of RGLs in the
adult mouse dentate gyrus
(A-B) Schematic illustrations of genetic approaches used to lineage-trace individual RGLs in the
adult mouse hippocampus. (C) Sample confocal images of Gli1
#
-RGL and Ascl1
#
-RGL clones at
1 day post-injection (dpi). Immunohistological analysis of GFP
+
RGLs showed 100 ± 0% presence
of neural precursor markers Nestin and GFAP in both Gli1
#
-RGLs (n = 3 animals) and Ascl1
#
-
RGLs (n = 3 animals). Scale bars, 5 μm. (D-E) Quantification of the number of clones labeled (D)
and number of cells within each clone (E) at different time points post tamoxifen injection using
Ascl1
#
and Gli1
#
labeling strategies. Values in (D) represent mean ± SEM (n ≥ 6 dentate gyri for
each time point). Dots in (E) indicate number of cells in each clone at different time points, which
may appear as bars when clones contain the same cell number. (F-G) Sample confocal images of
Cre immunohistology in the dentate gyrus of adult CreER mice. In (F), arrows denote Cre
+
MCM2
-
RGLs in Gli1-CreER
T2
mice and Cre
+
MCM2
+
RGLs in Ascl1- CreER
T2
mice. In (G), open
arrowheads denote Cre
+
Mash1
-
RGLs in Gli1-CreER
T2
mice and arrows denote Cre
+
Mash1
+
RGLs
in Ascl1-CreER
T2
. Scale bars: 100 μm (10 μm for inserts). (H) Comparison of the percentage of
Cre
+
RGLs that were MCM2
+
or Mash1
+
in Gli1-CreER
T2
and Ascl1-CreER
T2
mice. Values
represent mean ± SEM (n = 3 dentate gyri; *p < 0.001; Student’s t-test). Related to Figure 3.1.
118
Figure S3.2. Lineage trees and reporter comparisons from short-term analyses of Gli1
#
-
RGL and Ascl1
#
-RGL clones
(A) List of observed lineage trees for Gli1
#
-RGLs and Ascl1
#
-RGLs at 1, 3 and 7 dpi. The color
scheme denotes the categorization type of lineage trees, which are further subdivided by the
number of RGL divisions. (B-C) Quantification of RGL-containing Gli1
#
-RGL (B) and Ascl1
#
-
RGL (C) clone types at 1, 3 and 7 dpi. Colors represent the same clone type as denoted in (A). (D-
E) RGL fate choices assessed by independent lineage-tracing reporter paradigms. Shown are
summaries of RGL division types for Gli1
#
-RGLs in mT/mG or Z/EG reporter mice at 3 dpi (D)
and for Ascl1
#
-RGLs by Rosa-YFP or mT/mG reporter mice at 1 dpi (E). Values represent mean
±SEM. Related to Figures 3.1 and 3.2.
119
Figure S3.3. Long-term analyses of Gli1
#
-RGL and Ascl1
#
-RGL clones at 30 and 60 dpi
(A) List of observed lineage trees for Gli1
#
-RGLs and Ascl1
#
-RGLs at 30 dpi. Similarly, as in
Figure S3.2A. (B) Quantification of RGL-containing Ascl1
#
-RGL and Gli1
#
-RGL clone type at 30
dpi. Colors represent the same clone type as denoted in (A). (C) Quantification of the frequency
of clone composition types among RGL-retaining clones at 60 dpi for Gli1
#
-RGLs (n = 9 dentate
gyri) and Ascl1
#
-RGLs (n = 7 dentate gyri). Values represent mean ± SEM (*p < 0.05; n.s.: p >
0.1; Student’s t-test). Related to Figure 3.3.
120
Figure S3.4. Computational assessment of RGL properties
(A-C) Plots of RGL quiescence as a function of time (data points and error bars indicate mean ±
SEM; lines represent exponential fits). Activation kinetics (Figure 3.5B) are inferred from the
exponential rate of decay for Nestin
#
-RGLs (A), Gli1
#
-RGLs (B), and Ascl1
#
-RGL (C). Note the
time scale difference. (D-F) Plots of least squares fit of modeling results to the experimentally
observed frequency of clones within Nestin
#
-RGL (D), Gli1
#
-RGL (E) and Ascl1
#
-RGL (F)
lineage-tracing. Bars represent the experimentally observed joint distribution of RGLs and
astroglial cells in clones, while red error bars indicate 95% confidence intervals for the model
prediction. (G-I) Plots of the fraction of RGLs that divided only once as a function of time (data
points and error bars indicate mean ± SEM; lines represent model fits). Cell cycle re-entry time
(TC, Figure 3.5C) reflects the decay rate following the peak for Nestin
#
-RGLs (G), Gli1
#
-RGLs
(H) and Ascl1
#
-RGLs (I). Note the time scale difference. Related to Figure 3.4 and 3.5.
RGL
TA/A
RGL
TA/A
RGL
TA/A
0 20 40 60
0.0
0.2
0.4
0.6
0.8
1.0
Time (dpi)
0 2 4 6
0.0
0.2
0.4
0.6
0.8
1.0
Time (dpi)
0 2 4 6
0.0
0.2
0.4
0.6
0.8
1.0
Time (dpi)
Fraction of
undivided RGL
Fraction of
undivided RGL
Fraction of
undivided RGL
Nestin
#
Gli1
#
Ascl1
#
2
1
0
2
1
0
2
1
0
2
1
0
2
1
0
2
1
0
2 1 0 2 1 0 2 1 0 2 1 0 2 1 0 2 1 0
2 dpi 7 dpi
30 dpi 60 dpi
1 dpi 3 dpi
7 dpi 30 dpi
1 dpi 3 dpi
7 dpi 30 dpi
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
A B C
D E F
G H I
0 20 40 60
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of
RGL divided once
Fraction of
RGL divided once
Fraction of
RGL divided once
Figure S4 (Bonaguidi et al.)
Time (dpi) Time (dpi) Time (dpi)
121
Figure S3.5. Validation of computational models
(A) Fraction of Nestin
#
-RGL clones with the specified number of RGL and astroglial lineage
(TA/A) at different time points (stars) and model prediction (dashes). Values represent mean ±
SEM. (B) Fraction of Nestin
#
-RGL clones that retain at least one RGL at specified time points and
model prediction. (C) Fate choice probabilities obtained by fitting the solution of the master
equation to the experimentally observed Nestin
#
-RGL clone size distributions and fate choice
probabilities deduced from lineage-tracing data. RR: RGL-RGL division, RI: RGL-IPC division,
RA: RGL-astroglia division, II: IPC-IPC division, AA: astroglia-astroglia division, IA: IPC-
astrocyte division. (D) Fraction of Gli1
#
-RGL clones with the specified number of RGL and
astroglial lineage (TA/A) at different time points and model prediction. (E) Fraction of Ascl1
#
-
RGL clones with the specified number of RGL and astroglial lineage (TA/A) at different time
points and model prediction. Related to Figure 3.4 and S3.4.
122
Figure S3.6. Changes of Gli1
#
-RGL and Ascl1
#
-RGL behavior upon AraC-induced injury
(A) Schematic illustration of experimental paradigms. (B-C) Sample confocal images of EdU
staining (B) and GFAP immunohistology (C) at 7 days after completion of AraC treatment. Scale
bars: 100 μm (20 μm for inserts). For GFAP images, samples were processed in parallel and
acquired with same confocal settings for control and AraC groups. (D-E) List of observed lineage
trees for Gli1
#
-RGLs and Ascl1
#
-RGLs (D) and quantification of RGL-containing clone types (E).
Similar as in Figure S3.2A. (F) Summary of fate choices among controls for AraC injury. Fraction
of divisions are consistent between normal and AraC sham conditions for both Gli1
#
-RGLs and
Ascl1
#
-RGLs. Values represent mean ± SEM (n = 3-9 dentate gyri). (G) Quantification of the
number of Ascl1
#
RGL-containing clones after AraC injury. Values represent mean ± SEM (n =
8-9 dentate gyri; *p < 0.001; Student’s t-test). Related to Figure 3.5 and 3.6.
123
Figure S3.7. Nestin
#
-RGL behavior after AraC treatment
(A) Schematic illustration of experimental paradigm. (B) Sample confocal projection and single-
section images of GFP-labeled Nestin
#
-RGL clones 7 days after the stop of AraC treatment. Shown
are examples of Nestin
#
-RGLs that generated multiple cells of the neuronal lineage. Lineage trees
indicating self-renewal modes are shown next to the projection images. Arrows point to cells
within the clone. Scale bars, 10 μm (5 μm for inserts). (C-D) Summary of quantification of
percentages of RGL clones that divided (C, left) re-entered cell cycle (C, right), and of activated
RGL clones that generate new IPCs (D, left), and RGL clonal fate decisions (D, right) after AraC
treatment. The responses of Gli1
#
-RGLs and Ascl1
#
-RGLs to AraC are re-plotted (see Figure 3.6)
for comparison (C-D). Values represent mean ± SEM (n = 4 dentate gyri; *p < 0.05, n.s.: p > 0.1;
Student’s t-test). Related to Figure 3.5.
124
Table S3.1. Tamoxifen doses used to achieve clonal recombination among various promoter,
reporter, and environmental contexts over the analyzed time course
Related to Figures 3.1-3.6.
Promoter Reporter [Tamoxifen] Condition Chase Figure(s)
Gli1::CreER
T2
Z/EG 217 mg/kg Physiological 3, 30, 60 dpi
3.1, 3.3-
3.6
Gli1::CreER
T2
mT/mG 62 mg/kg Physiological
0.5, 1, 3, 7
dpi
3.1, 3.5,
3.6
Gli1::CreER
T2
mT/mG 62 mg/kg
Labelling
under
physiological
conditions
and followed
by injury
14 dpi 3.6
Ascl1::CreER
T2
ROSA-
YFP
78 mg/kg Physiological
1, 3, 30, 60
dpi
3.1- 3.6
Ascl1::CreER
T2
mT/mG
124-217
mg/kg
Physiological 0.5, 1 dpi 3.1, 3.6
Ascl1::CreER
T2
ROSA-
YFP
78 mg/kg
Labelling
under
physiological
conditions
and followed
by injury
7, 14 dpi 3.6
Ascl1::CreER
T2
mT/mG 124 mg/kg
Labelling
under
physiological
conditions
and followed
by injury
7, 14 dpi 3.6
Nestin::CreERT2 Z/EG 62 mg/kg Physiological
2, 7, 30, 60,
120 dpi
3.5 – 3.6
Nestin::CreERT2 Z/EG 62 mg/kg
Labelling
under
physiological
conditions
and followed
by injury
14 dpi S3.7
Nestin::CreERT2 Confetti 62 mg/kg Physiological
12mo –
30dpi, 60
dpi
3.6
125
Figure S4.1. Example of confocal images of the clones depicted at various timepoints
(A) Confocal images of unipotent NSC clones at different days post tamoxifen injection.
NSC=neural stem cell; TA/A=astroglial lineage; IPC/NB/N=neuronal lineage;
OPC/OL=oligodendrocyte lineage. (B) Confocal images of multipotent NSC clones at different
days post tamoxifen injection. NSC=neural stem cell; TA/A=astroglial lineage;
IPC/NB/N=neuronal lineage; OPC/OL=oligodendrocyte lineage. (C) Confocal images of no NSC
clones at different days post tamoxifen injection. NSC=neural stem cell; TA/A=astroglial lineage;
IPC/NB/N=neuronal lineage; OPC/OL=oligodendrocyte lineage.
126
Figure S4.2. Clonal lineage-tracing of individual NSCs in the mouse dentate gyrus during
tissue remodeling.
(A) Quantification of the fraction of clones containing neural stem cells (NSCs) among all clones
at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM (N = 41-96 clones, Table
S1). (B) Quantification of the fraction of clones containing neurons (N) among all clones at 30-,
90-, 180- and 365-days post tamoxifen injection. Mean± SEM (N = 41-96 clones, Table S1). (C)
Quantification of the fraction of clones containing astrocytes (TA) and mature astrocytes (A)
among all clones at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM (N = 41-
96 clones, Table S1).(D) Quantification of the fraction of clones containing oligodendrocyte
progenitor cells (OPC) and mature oligodendrocytes (OL) among all clones 30-, 90-, 180- and
365-days post tamoxifen injection. Mean± SEM (N = 41-96 clones, Table S1).(E) Quantification
of average number of intermediate progenitor cells/neuroblasts (IPCs/NB) and neurons (N) per
NSC-containing clones at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM. (N
= 41-96 clones, Table S1). ANOVA with Tukey’s multiple comparisons test. (F) Quantification
of average number of transition astrocytes (TA) and mature astrocytes (A) per NSC-containing
clones at 30-, 90-, 180- and 365-days post tamoxifen injection. Mean± SEM. (N = 41-96 clones,
Table S1). ANOVA with Tukey’s multiple comparisons test.
127
Figure S4.3. Clonal lineage-tracing of individual NSCs in the mouse dentate gyrus during
postnatal development
(A) Quantification of the fraction of clones containing neurons (N) among all clones at 7, 15, 30
days post tamoxifen injection. Mean± SEM (N = 41-96 clones, Table S1).(B)Quantification of the
fraction of clones containing astrocytes (TA) and mature astrocytes (A) among all clones at 7, 15,
30 days post tamoxifen injection. Mean± SEM (N = 41-96 clones, Table S1).(C) Quantification of
the fraction of clones containing oligodendrocyte progenitor cells (OPC) and mature
oligodendrocytes (OL) among all clones at 7, 15, 30 days post tamoxifen injection. Mean± SEM
(N = 41-96 clones, Table S1).(D) Quantification of average number of intermediate progenitor
cells/neuroblasts (IPCs/NB) and neurons (N) per NSC-containing clones at 7, 15, 30 days post
tamoxifen injection. Mean± SEM. (N = 41-96 clones, Table S1). ANOVA with Tukey’s multiple
comparisons test.(E) Quantification of average number of transition astrocytes (TA) and mature
astrocytes (A) per NSC-containing clones at 7, 15, 30 days post tamoxifen injection. Mean± SEM.
(N = 41-96 clones, Table S1). ANOVA with Tukey’s multiple comparisons test.(F) Quantification
of average number of oligodendrocyte progenitor cells (OPC) and mature oligodendrocytes (OL)
per NSC-containing clones at 7, 15, 30 days post tamoxifen injection. Mean± SEM. (N = 41-96
clones, Table S1). ANOVA with Tukey’s multiple comparisons test.
128
Table S4.1. Number of clones for clonal analysis.
Total
number of
clones
Total number of
NSC-containing
clones
Total number
of non NSC-
containing
clones
Number of
animals
Gli1
#
E17.5
P0/P1 144 labeled
cells
4
P7 41 24 17 8
P15 81 35 46 15
P30 96 37 59 13
P90 78 29 49 12
P180 85 24 61 13
P365 60 9 51 11
Clones Animals
Total of Gli1
#
-lineage tracing 441 76
129
References
A, K., & A, A.-B. (2009). The glial nature of embryonic and adult neural stem cells. Annual review
of neuroscience, 32.
AB, R., CM, R., RA, M., A, K., P, S., Z, Y., et al. (2018). Single-cell profiling of the developing
mouse brain and spinal cord with split-pool barcoding. Science (New York, N.Y.),
360(6385).
Ahn, S., & Joyner, A. L. (2005). In vivo analysis of quiescent adult neural stem cells responding
to Sonic hedgehog. Nature, 437(7060), 894-897.
Aimone, J. B., Li, Y., Lee, S. W., Clemenson, G. D., Deng, W., & Gage, F. H. (2014). Regulation
and function of adult neurogenesis: from genes to cognition. Physiol Rev, 94(4), 991-1026.
Andersen, J., Urban, N., Achimastou, A., Ito, A., Simic, M., Ullom, K., et al. (2014). A
transcriptional mechanism integrating inputs from extracellular signals to activate
hippocampal stem cells. Neuron, 83(5), 1085-1097.
Artegiani, B., & Calegari, F. (2012). Age-related cognitive decline: can neural stem cells help us?
Aging (Albany NY), 4(3), 176-186.
Balordi, F., & Fishell, G. (2007). Mosaic removal of hedgehog signaling in the adult SVZ reveals
that the residual wild-type stem cells have a limited capacity for self-renewal. J Neurosci,
27(52), 14248-14259.
Belsky, D. W., Caspi, A., Houts, R., Cohen, H. J., Corcoran, D. L., Danese, A., et al. (2015).
Quantification of biological aging in young adults. Proc Natl Acad Sci U S A, 112(30),
E4104-4110.
Ben Abdallah, N. M., Slomianka, L., Vyssotski, A. L., & Lipp, H. P. (2010). Early age-related
changes in adult hippocampal neurogenesis in C57 mice. Neurobiol Aging, 31(1), 151-161.
130
Berg, D. A., Su, Y., Jimenez-Cyrus, D., Patel, A., Huang, N., Morizet, D., et al. (2019). A Common
Embryonic Origin of Stem Cells Drives Developmental and Adult Neurogenesis. Cell,
177(3), 654-668.e615.
Berg, D. K., Yoon, K. J., Will, B., Xiao, A. Y., Kim, N. S., Christian, K. M., et al. (2015). Tbr2-
expressing intermediate progenitor cells in the adult mouse hippocampus are unipotent
neuronal precursors with limited amplification capacity under homeostasis. Front Biol
(Beijing), 10, 262-271.
Blanpain, C., & Fuchs, E. (2014). Stem cell plasticity. Plasticity of epithelial stem cells in tissue
regeneration. Science, 344(6189), 1242281.
Blanpain, C., & Simons, B. D. (2013). Unravelling stem cell dynamics by lineage tracing. Nat Rev
Mol Cell Biol, 14(8), 489-502.
Bonaguidi, M. A., Song, J., Ming, G. L., & Song, H. (2012). A unifying hypothesis on mammalian
neural stem cell properties in the adult hippocampus. Curr Opin Neurobiol, 22(5), 754-
761.
Bonaguidi, M. A., Stadel, R. P., Berg, D. A., Sun, J., Ming, G. L., & Song, H. (2016). Diversity of
Neural Precursors in the Adult Mammalian Brain. Cold Spring Harb Perspect Biol, 8(4).
Bonaguidi, M. A., Wheeler, M. A., Shapiro, J. S., Stadel, R. P., Sun, G. J., Ming, G. L., et al.
(2011). In vivo clonal analysis reveals self-renewing and multipotent adult neural stem cell
characteristics. Cell, 145(7), 1142-1155.
Bond, A. M., Ming, G. L., & Song, H. (2015). Adult Mammalian Neural Stem Cells and
Neurogenesis: Five Decades Later. Cell Stem Cell, 17(4), 385-395.
131
Bottes, S., Jaeger, B. N., Pilz, G. A., Jörg, D. J., Cole, J. D., Kruse, M., et al. (2021). Long-term
self-renewing stem cells in the adult mouse hippocampus identified by intravital imaging.
Nat Neurosci, 24(2), 225-233.
Buczacki, S. J., Zecchini, H. I., Nicholson, A. M., Russell, R., Vermeulen, L., Kemp, R., et al.
(2013). Intestinal label-retaining cells are secretory precursors expressing Lgr5. Nature,
495(7439), 65-69.
Calzolari, F., Michel, J., Baumgart, E. V., Theis, F., Gotz, M., & Ninkovic, J. (2015). Fast clonal
expansion and limited neural stem cell self-renewal in the adult subependymal zone. Nat
Neurosci, 18(4), 490-492.
Christian, K. M., Song, H., & Ming, G. L. (2014). Functions and dysfunctions of adult
hippocampal neurogenesis. Annu Rev Neurosci, 37, 243-262.
Clayton, E., Doupe, D. P., Klein, A. M., Winton, D. J., Simons, B. D., & Jones, P. H. (2007). A
single type of progenitor cell maintains normal epidermis. Nature, 446(7132), 185-189.
Codega, P., Silva-Vargas, V., Paul, A., Maldonado-Soto, A. R., Deleo, A. M., Pastrana, E., et al.
(2014). Prospective identification and purification of quiescent adult neural stem cells from
their in vivo niche. Neuron, 82(3), 545-559.
Collins, C. A., Olsen, I., Zammit, P. S., Heslop, L., Petrie, A., Partridge, T. A., et al. (2005). Stem
cell function, self-renewal, and behavioral heterogeneity of cells from the adult muscle
satellite cell niche. Cell, 122(2), 289-301.
DeCarolis, N. A., Mechanic, M., Petrik, D., Carlton, A., Ables, J. L., Malhotra, S., et al. (2013).
In vivo contribution of nestin- and GLAST-lineage cells to adult hippocampal
neurogenesis. Hippocampus, 23(8), 708-719.
132
Deng, W., Aimone, J. B., & Gage, F. H. (2010). New neurons and new memories: how does adult
hippocampal neurogenesis affect learning and memory? Nat Rev Neurosci, 11(5), 339-350.
Doetsch, F., Caille, I., Lim, D. A., Garcia-Verdugo, J. M., & Alvarez-Buylla, A. (1999).
Subventricular zone astrocytes are neural stem cells in the adult mammalian brain. Cell,
97(6), 703-716.
Dranovsky, A., & Leonardo, E. D. (2012). Is there a role for young hippocampal neurons in
adaptation to stress? Behav Brain Res, 227(2), 371-375.
Dranovsky, A., Picchini, A. M., Moadel, T., Sisti, A. C., Yamada, A., Kimura, S., et al. (2011).
Experience dictates stem cell fate in the adult hippocampus. Neuron, 70(5), 908-923.
Ehm, O., Göritz, C., Covic, M., Schäffner, I., Schwarz, T. J., Karaca, E., et al. (2010). RBPJkappa-
dependent signaling is essential for long-term maintenance of neural stem cells in the adult
hippocampus. J Neurosci, 30(41), 13794-13807.
Ema, H., Morita, Y., & Suda, T. (2014). Heterogeneity and hierarchy of hematopoietic stem cells.
Exp Hematol, 42(2), 74-82 e72.
Encinas, J. M., Michurina, T. V., Peunova, N., Park, J. H., Tordo, J., Peterson, D. A., et al. (2011).
Division-coupled astrocytic differentiation and age-related depletion of neural stem cells
in the adult hippocampus. Cell Stem Cell, 8(5), 566-579.
Etzrodt, M., Endele, M., & Schroeder, T. (2014). Quantitative single-cell approaches to stem cell
research. Cell Stem Cell, 15(5), 546-558.
G, L., H, K., SR, C., & SJ, P. (2009). Identification of a transient subpial neurogenic zone in the
developing dentate gyrus and its regulation by Cxcl12 and reelin signaling. Development
(Cambridge, England), 136(2).
133
G, L., L, F., G, F., & SJ, P. (2013). The ventral hippocampus is the embryonic origin for adult
neural stem cells in the dentate gyrus. Neuron, 78(4).
Gage, F. H. (2000). Mammalian neural stem cells. Science, 287(5457), 1433-1438.
Ge, S., Goh, E. L., Sailor, K. A., Kitabatake, Y., Ming, G. L., & Song, H. (2006). GABA regulates
synaptic integration of newly generated neurons in the adult brain. Nature, 439(7076), 589-
593.
Gebara, E., Bonaguidi, M. A., Beckervordersandforth, R., Sultan, S., Udry, F., Gijs, P. J., et al.
(2016). Heterogeneity of Radial Glia-Like Cells in the Adult Hippocampus. Stem Cells.
Gillman, P. K. (2005). Monoamine oxidase inhibitors, opioid analgesics and serotonin toxicity. Br
J Anaesth, 95(4), 434-441.
Global, regional, and national burden of Alzheimer's disease and other dementias, 1990-2016: a
systematic analysis for the Global Burden of Disease Study 2016. (2019). Lancet Neurol,
18(1), 88-106.
Gontier, G., Iyer, M., Shea, J. M., Bieri, G., Wheatley, E. G., Ramalho-Santos, M., et al. (2018).
Tet2 Rescues Age-Related Regenerative Decline and Enhances Cognitive Function in the
Adult Mouse Brain. Cell Rep, 22(8), 1974-1981.
Hansen, N., & Auger, A. (2014). Principled Design of Continuous Stochastic Search: From Theory
to Practice. In Y. Borenstein & A. Moraglio (Eds.), Theory and Principled Methods for the
Design of Metaheuristics (pp. 145-180). Berlin, Heidelberg: Springer Berlin Heidelberg.
Hara, K., Nakagawa, T., Enomoto, H., Suzuki, M., Yamamoto, M., Simons, B. D., et al. (2014).
Mouse spermatogenic stem cells continually interconvert between equipotent singly
isolated and syncytial states. Cell Stem Cell, 14(5), 658-672.
134
Hayes, N. L., & Nowakowski, R. S. (2002). Dynamics of cell proliferation in the adult dentate
gyrus of two inbred strains of mice. Brain Res Dev Brain Res, 134(1-2), 77-85.
Hochgerner, H., Zeisel, A., Lonnerberg, P., & Linnarsson, S. (2018). Conserved properties of
dentate gyrus neurogenesis across postnatal development revealed by single-cell RNA
sequencing. Nat Neurosci, 21(2), 290-299.
Hochgerner, H., Zeisel, A., Lönnerberg, P., & Linnarsson, S. (2018). Conserved properties of
dentate gyrus neurogenesis across postnatal development revealed by single-cell RNA
sequencing. Nat Neurosci, 21(2), 290-299.
Hodge, R. D., Kowalczyk, T. D., Wolf, S. A., Encinas, J. M., Rippey, C., Enikolopov, G., et al.
(2008). Intermediate progenitors in adult hippocampal neurogenesis: Tbr2 expression and
coordinate regulation of neuronal output. J Neurosci, 28(14), 3707-3717.
Ibrayeva, A., Bay, M., Pu, E., Jörg, D., Peng, L., Jun, H., et al. (2019). Early Stem Cell Aging in
the Mature Brain. bioRxiv, 654608.
Imayoshi, I., Isomura, A., Harima, Y., Kawaguchi, K., Kori, H., Miyachi, H., et al. (2013).
Oscillatory control of factors determining multipotency and fate in mouse neural
progenitors. Science, 342(6163), 1203-1208.
Jang, M. H., Bonaguidi, M. A., Kitabatake, Y., Sun, J., Song, J., Kang, E., et al. (2013). Secreted
frizzled-related protein 3 regulates activity-dependent adult hippocampal neurogenesis.
Cell Stem Cell, 12(2), 215-223.
Jones, K. M., Sarić, N., Russell, J. P., Andoniadou, C. L., Scambler, P. J., & Basson, M. A. (2015).
CHD7 maintains neural stem cell quiescence and prevents premature stem cell depletion
in the adult hippocampus. Stem Cells, 33(1), 196-210.
135
Kalamakis, G., Brüne, D., Ravichandran, S., Bolz, J., Fan, W., Ziebell, F., et al. (2019). Quiescence
Modulates Stem Cell Maintenance and Regenerative Capacity in the Aging Brain. Cell,
176(6), 1407-1419.e1414.
Kim, E. J., Ables, J. L., Dickel, L. K., Eisch, A. J., & Johnson, J. E. (2011). Ascl1 (Mash1) defines
cells with long-term neurogenic potential in subgranular and subventricular zones in adult
mouse brain. PLoS One, 6(3), e18472.
Klein, A. M., & Simons, B. D. (2011). Universal patterns of stem cell fate in cycling adult tissues.
Development, 138(15), 3103-3111.
Knoth, R., Singec, I., Ditter, M., Pantazis, G., Capetian, P., Meyer, R. P., et al. (2010). Murine
features of neurogenesis in the human hippocampus across the lifespan from 0 to 100 years.
PLoS One, 5(1), e8809.
Kretzschmar, K., & Watt, F. M. (2012). Lineage tracing. Cell, 148(1-2), 33-45.
Krieger, T., & Simons, B. D. (2015). Dynamic stem cell heterogeneity. Development, 142(8),
1396-1406.
Krieger, T., & Simons, B. D. (2015). Dynamic stem cell heterogeneity. Development, 142(8),
1396-1406.
Kuhn, H. G., Dickinson-Anson, H., & Gage, F. H. (1996). Neurogenesis in the dentate gyrus of
the adult rat: age-related decrease of neuronal progenitor proliferation. J Neurosci, 16(6),
2027-2033.
Kuhn, H. G., Toda, T., & Gage, F. H. (2018). Adult Hippocampal Neurogenesis: A Coming-of-
Age Story. J Neurosci, 38(49), 10401-10410.
La Manno, G., Soldatov, R., Zeisel, A., Braun, E., Hochgerner, H., Petukhov, V., et al. (2018).
RNA velocity of single cells. [OriginalPaper]. Nature, 560(7719), 494-498.
136
LC, F., SB, R., JI, P., K, O., R, R., CL, C., et al. (2015). Embryonic Origin of Postnatal Neural
Stem Cells. Cell, 161(7).
Leeman, D. S., Hebestreit, K., Ruetz, T., Webb, A. E., McKay, A., Pollina, E. A., et al. (2018).
Lysosome activation clears aggregates and enhances quiescent neural stem cell activation
during aging. Science, 359(6381), 1277-1283.
Lemberger, T., Parlato, R., Dassesse, D., Westphal, M., Casanova, E., Turiault, M., et al. (2007).
Expression of Cre recombinase in dopaminoceptive neurons. BMC Neurosci, 8, 4.
Leung, C. T., Coulombe, P. A., & Reed, R. R. (2007). Contribution of olfactory neural stem cells
to tissue maintenance and regeneration. Nat Neurosci, 10(6), 720-726.
Li, L., & Clevers, H. (2010). Coexistence of quiescent and active adult stem cells in mammals.
Science, 327(5965), 542-545.
Liu, N., Matsumura, H., Kato, T., Ichinose, S., Takada, A., Namiki, T., et al. (2019). Stem cell
competition orchestrates skin homeostasis and ageing. Nature, 568(7752), 344-350.
Lopez, O. L., & Kuller, L. H. (2019). Epidemiology of aging and associated cognitive disorders:
Prevalence and incidence of Alzheimer's disease and other dementias. Handb Clin Neurol,
167, 139-148.
Martín-Suárez, S., Valero, J., Muro-García, T., & Encinas, J. M. (2019). Phenotypical and
functional heterogeneity of neural stem cells in the aged hippocampus. Aging Cell, 18(4),
e12958.
Mascre, G., Dekoninck, S., Drogat, B., Youssef, K. K., Brohee, S., Sotiropoulou, P. A., et al.
(2012). Distinct contribution of stem and progenitor cells to epidermal maintenance.
Nature, 489(7415), 257-262.
137
Merkle, F. T., Mirzadeh, Z., & Alvarez-Buylla, A. (2007). Mosaic organization of neural stem
cells in the adult brain. Science, 317(5836), 381-384.
Mich, J. K., Signer, R. A., Nakada, D., Pineda, A., Burgess, R. J., Vue, T. Y., et al. (2014).
Prospective identification of functionally distinct stem cells and neurosphere-initiating
cells in adult mouse forebrain. Elife, 3, e02669.
Moreno-Jiménez, E. P., Flor-García, M., Terreros-Roncal, J., Rábano, A., Cafini, F., Pallas-
Bazarra, N., et al. (2019). Adult hippocampal neurogenesis is abundant in neurologically
healthy subjects and drops sharply in patients with Alzheimer's disease. Nat Med, 25(4),
554-560.
Morgenstern, N. A., Lombardi, G., & Schinder, A. F. (2008). Newborn granule cells in the ageing
dentate gyrus. The Journal of Physiology, 586(16), 3751-3757.
Muller-Sieburg, C. E., Sieburg, H. B., Bernitz, J. M., & Cattarossi, G. (2012). Stem cell
heterogeneity: implications for aging and regenerative medicine. Blood, 119(17), 3900-
3907.
Muzumdar, M. D., Tasic, B., Miyamichi, K., Li, L., & Luo, L. (2007). A global double-fluorescent
Cre reporter mouse. Genesis, 45(9), 593-605.
Navarro Negredo, P., Yeo, R. W., & Brunet, A. (2020). Aging and Rejuvenation of Neural Stem
Cells and Their Niches. Cell Stem Cell, 27(2), 202-223.
Novak, A., Guo, C., Yang, W., Nagy, A., & Lobe, C. G. (2000). Z/EG, a double reporter mouse
line that expresses enhanced green fluorescent protein upon Cre-mediated excision.
Genesis, 28(3-4), 147-155.
Pilz, G.-A., Bottes, S., Betizeau, M., Jörg, D. J., Carta, S., Simons, B. D., et al. (2018). Live
imaging of neurogenesis in the adult mouse hippocampus. Science, 359(6376), 658-662.
138
Renault, V. M., Rafalski, V. A., Morgan, A. A., Salih, D. A., Brett, J. O., Webb, A. E., et al. (2009).
FoxO3 regulates neural stem cell homeostasis. Cell Stem Cell, 5(5), 527-539.
Ritsma, L., Ellenbroek, S. I., Zomer, A., Snippert, H. J., de Sauvage, F. J., Simons, B. D., et al.
(2014). Intestinal crypt homeostasis revealed at single-stem-cell level by in vivo live
imaging. Nature, 507(7492), 362-365.
Rocheteau, P., Gayraud-Morel, B., Siegl-Cachedenier, I., Blasco, M. A., & Tajbakhsh, S. (2012).
A subpopulation of adult skeletal muscle stem cells retains all template DNA strands after
cell division. Cell, 148(1-2), 112-125.
Rompolas, P., Mesa, K. R., & Greco, V. (2013). Spatial organization within a niche as a
determinant of stem-cell fate. Nature, 502(7472), 513-518.
S, F., H, M., T, W., H, K., N, Y., Y, H., et al. (2015). Slowly dividing neural progenitors are an
embryonic origin of adult neural stem cells. Nature neuroscience, 18(5).
Schaum, N., Lehallier, B., Hahn, O., Pálovics, R., Hosseinzadeh, S., Lee, S. E., et al. (2020).
Ageing hallmarks exhibit organ-specific temporal signatures. Nature, 583(7817), 596-602.
Sierra, A., Encinas, J. M., Deudero, J. J., Chancey, J. H., Enikolopov, G., Overstreet-Wadiche, L.
S., et al. (2010). Microglia shape adult hippocampal neurogenesis through apoptosis-
coupled phagocytosis. Cell Stem Cell, 7(4), 483-495.
Sierra, A., Martín-Suárez, S., Valcárcel-Martín, R., Pascual-Brazo, J., Aelvoet, S. A., Abiega, O.,
et al. (2015). Neuronal hyperactivity accelerates depletion of neural stem cells and impairs
hippocampal neurogenesis. Cell Stem Cell, 16(5), 488-503.
Solanas, G., & Benitah, S. A. (2013). Regenerating the skin: a task for the heterogeneous stem cell
pool and surrounding niche. Nat Rev Mol Cell Biol, 14(11), 737-748.
139
Song, J., Sun, J., Moss, J., Wen, Z., Sun, G. J., Hsu, D., et al. (2013). Parvalbumin interneurons
mediate neuronal circuitry-neurogenesis coupling in the adult hippocampus. Nat Neurosci,
16(12), 1728-1730.
Song, J., Zhong, C., Bonaguidi, M. A., Sun, G. J., Hsu, D., Gu, Y., et al. (2012). Neuronal circuitry
mechanism regulating adult quiescent neural stem-cell fate decision. Nature, 489(7414),
150-154.
Spalding, K. L., Bergmann, O., Alkass, K., Bernard, S., Salehpour, M., Huttner, H. B., et al. (2013).
Dynamics of hippocampal neurogenesis in adult humans. Cell, 153(6), 1219-1227.
Srinivas, S., Watanabe, T., Lin, C. S., William, C. M., Tanabe, Y., Jessell, T. M., et al. (2001). Cre
reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus.
BMC Dev Biol, 1, 4.
Stange, D. E., Koo, B. K., Huch, M., Sibbel, G., Basak, O., Lyubimova, A., et al. (2013).
Differentiated Troy+ chief cells act as reserve stem cells to generate all lineages of the
stomach epithelium. Cell, 155(2), 357-368.
Steiner, B., Kronenberg, G., Jessberger, S., Brandt, M. D., Reuter, K., & Kempermann, G. (2004).
Differential regulation of gliogenesis in the context of adult hippocampal neurogenesis in
mice. Glia, 46(1), 41-52.
Sun, G. J., Sailor, K. A., Mahmood, Q. A., Chavali, N., Christian, K. M., Song, H., et al. (2013).
Seamless reconstruction of intact adult-born neurons by serial end-block imaging reveals
complex axonal guidance and development in the adult hippocampus. J Neurosci, 33(28),
11400-11411.
140
Urban, N., van den Berg, D. L., Forget, A., Andersen, J., Demmers, J. A., Hunt, C., et al. (2016).
Return to quiescence of mouse neural stem cells by degradation of a proactivation protein.
Science, 353(6296), 292-295.
van der Wath, R. C., Wilson, A., Laurenti, E., Trumpp, A., & Lio, P. (2009). Estimating dormant
and active hematopoietic stem cell kinetics through extensive modeling of
bromodeoxyuridine label-retaining cell dynamics. PLoS One, 4(9), e6972.
Weissman, I. L., Anderson, D. J., & Gage, F. H. (2001). Stem and progenitor cells: origins,
phenotypes, lineage commitments, and transdifferentiations. Annu Rev Cell Dev Biol, 17,
387-403.
Zhao, C., Deng, W., & Gage, F. H. (2008). Mechanisms and functional implications of adult
neurogenesis. Cell, 132(4), 645-660.
Zhao, C., Teng, E. M., Summers, R. G., Jr., Ming, G. L., & Gage, F. H. (2006). Distinct
morphological stages of dentate granule neuron maturation in the adult mouse
hippocampus. J Neurosci, 26(1), 3-11.
Ziebell, F., Dehler, S., Martin-Villalba, A., & Marciniak-Czochra, A. (2018). Revealing age-
related changes of adult hippocampal neurogenesis using mathematical models.
Development, 145(1), dev153544.
Abstract (if available)
Abstract
Brain plasticity underlies our ability to maintain cognitive ability and the capability to efficiently interact with society. Such flexibility is well known to decline with age but determining how to identify and prevent it at young ages remains unknown. It is now recognized that in adults quiescent neural stem cells (NSCs) generate newborn neurons and astrocytes to modify existing neural circuits. Despite this, neurogenesis and hippocampus function are markedly lower in older animals. Unique among somatic tissues containing stem cells, the brain experiences a decline in cell genesis early in young adult rodents and middle ages in humans. Moreover, numerous studies indicate that a decline in adult stem cell function can drive aging-related diseases.
Aged animals have significantly less neural stem cell numbers, stem cell proliferation, neuronal differentiation and newborn neuron survival compared to younger animals (Encinas et al., 2011; Kuhn, Dickinson-Anson, & Gage, 1996; Ziebell, Dehler, Martin-Villalba, & Marciniak-Czochra, 2018). Further, NSCs in old animals exhibit hallmarks of cellular aging including deficits in proteostasis and receive high levels of inflammation (Kalamakis et al., 2019; Leeman et al., 2018). Yet, the hippocampus experiences a loss of neurogenesis early in the mature brain of rodents (Ben Abdallah, Slomianka, Vyssotski, & Lipp, 2010; Morgenstern, Lombardi, & Schinder, 2008) and by middle-age in humans (Knoth et al., 2010; Moreno-Jiménez et al., 2019; Spalding et al., 2013). This decline is accompanied by epigenetic loss of DNA demethylation (Gontier et al., 2018), suggesting NSCs could become dysregulated early during chronological aging.
Therefore, the central goals of my doctoral work were to identify cellular and molecular mechanisms that underline cellular aging of adult NSC populations, as well as understand the origins and heterogeneity of these NSC populations. I’ve applied new approaches in endogenous single cell lineage tracing, computation modeling and single cell genomic profiling to address questions that have remained unanswered for more than twenty years: what primarily drives neural cell genesis decline during physiological aging? What are the origins of different NSC populations? How do they contribute to the whole physiological decline? What are the cellular and molecular mechanisms that are responsible? Can targeted neural circuit remodeling restore stem cell function to healthy, younger levels? A better understanding of neurogenesis during development and aging could serve as a platform to develop novel therapeutic strategies against the different age-related pathological and physiological neurological disorders.
Recent studies revealed the first demonstration of endogenous adult mammalian NSC properties at the single cell level, which are now recognized as the new standard for the field of stem cell research (Bonaguidi et al., 2011). During my doctoral studies I have developed various lineage tracing strategies to track individual NSCs and their evolution within the mouse hippocampus. I was able to establish single cell RNA sequencing (scRNA-seq) protocol for capturing and deep sequencing rare subpopulations of adult neural stem cells.
In my doctoral work I present a new concept of stem cell aging where stem cell function declines due to cellular and molecular changes that compromise their homeostasis. While aging is typically examined among old organisms, biological aging occurs in a gradual and asynchronous manner throughout the body (Schaum et al., 2020). By combining in vivo single cell clonal lineage tracing, computational modeling approaches, scRNA-seq and systems level data science we comprehensively investigated neural stem cell adaptation and restoration during development and aging. I identified that NSCs undergo early aging by defining when, why and how NSCs lose homeostasis. I showed that targeting mechanisms associated with the initial loss of NSC homeostasis can overcome age-related NSC dysfunction later in life. I used a clinically relevant drug Imatinib (Abl1/2 inhibitor) as a strategy to overcome NSC cellular aging. Indeed, intracranial infusion into the middle-aged brain was sufficient to overcome deep NSC quiescence and restore NSC proliferation to younger levels. My study elucidated cellular and molecular origins of neurogenesis decline in the middle-age adult and may serve as a new mammalian stem cell model to study cellular aging. These findings provide a broadly useful resource to prioritize individual genes or complementary gene families to create new directions towards age-related regenerative medicine throughout the body.
Lastly, I developed a transgenic labeling approach in mice driven by Gli1# regulatory elements combined with stochastic multicolor Confetti reporter (CFP, YFP, RFP) to trace the entire process of DG neurogenesis from a single cell from development through adulthood. For performing in vivo single cell lineage tracing, I inject time-pregnant female Gli1-CreER::Confetti mice with 14 mg/kg tamoxifen to label pups at E17.5 and sacrificed them at multiple postnatal timepoints. I found that multipotent NSC clones emerge by the end of the hippocampal postnatal ontogenesis. In addition, neurogenesis expands within NSC-containing clones into the young adult. I also was able to establish the sequential order of neuron, oligodendrocyte and astrocyte production in the adult mouse hippocampus. In doing so, I revealed developmental properties of neural precursors and their transition into adult neural stem cells.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
ROOT: a novel pharmacotranscritomic pipeline for rescuing age-associated functional decline of hippocampal adult neurogenesis
PDF
Natural divergence of traits in species of mice reveals novel molecular mechanisms of cellular senescence
PDF
A single cell time course of senescence uncovers discrete cell trajectories and transcriptional heterogeneity
PDF
Investigating brain aging and neurodegenerative diseases through omics data
PDF
Identification and characterization of adult stem cells in the oral cavity
PDF
Identifying and measuring cell-intrinsic and cell-extrinsic factors influencing aging
PDF
The overlap between mTOR signaling, rapamycin and cellular senescence
PDF
Genomic and phenotypic novelties in the Southeast Asian house mouse
PDF
Neural fate regulated by extrinsic signaling and epigenetics
PDF
Computational approaches to identify genetic regulators of aging and late-life mortality
PDF
Dissecting the heterogeneity of mouse hematopoietic stem cells in vivo
PDF
Tissue interactions & molecular pathways in specification of the ectomesenchyme from cranial neural crest
PDF
Characterization of senescent cell heterogeneity using cell culture models
PDF
Signaling networks in complex brain disorders
PDF
Modeling neurodegenerative diseases using induced pluripotent stem cells and identifying therapeutic targets
PDF
Mediating the homeostatic response during light & cytotoxic induced retinal damage
PDF
Molecular basis of mouse epiblast stem cell and human embryonic stem cell self‐renewal
PDF
Signaling mechanisms governing intestinal regeneration and gut-glia cross-talk in Drosophila
PDF
Novel roles for Maf1 in embryonic stem cell differentiation and adipogenesis
PDF
Regional localization and regulation of hematopoietic stem cells in the bone marrow stem cell niche
Asset Metadata
Creator
Ibrayeva, Albina
(author)
Core Title
Longitudinal assessment of neural stem-cell aging
School
Leonard Davis School of Gerontology
Degree
Doctor of Philosophy
Degree Program
Biology of Aging
Degree Conferral Date
2022-05
Publication Date
03/07/2022
Defense Date
03/04/2022
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
aging,cellular aging,Development,homeostasis,lineage tracing,neural stem cells,OAI-PMH Harvest,RNA-seq
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Bonaguidi, Michael A. (
committee chair
), Benayoun, Berenice A. (
committee member
), Campisi, Judith (
committee member
), Finch, Caleb E. (
committee member
), Quadrato, Giorgia (
committee member
)
Creator Email
aibrayev@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC110768080
Unique identifier
UC110768080
Legacy Identifier
etd-IbrayevaAl-10429
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Ibrayeva, Albina
Type
texts
Source
20220308-usctheses-batch-915
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
cellular aging
homeostasis
lineage tracing
neural stem cells
RNA-seq