University of Southern California Dissertations and Theses

Investigating molecular roadblocks to enhance direct cellular reprogramming

(USC Thesis Other)

Investigating molecular roadblocks to enhance direct cellular reprogramming

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Investigating Molecular Roadblocks
to Enhance Direct Cellular Reprogramming

A Dissertation Presented By
Kimberley Nicole Wunder

To the Faculty of the USC Graduate School
In Partial Fulfillment of the Requirements for the
Degree Doctor of Philosophy in
Development, Stem Cells, and Regenerative Medicine

University of Southern California
Los Angeles, California
December 2019

ii
Acknowledgments
“It’s the quality of one’s convictions that determines success.” – Harry Potter
Seven years later, I’m still the same Harry Potter nerd I was when I started this crazy journey,
but now I’m finishing with a few more letters behind my name. I truly believe that my heart and
soul, sweat, blood and tears, God’s answered prayers, and belief in myself have brought me to
this point. This thesis one of my greatest life successes, and the culmination of years of work
that would not have been possible without the guidance, support, and love of many people.
First and foremost, to Nana and Pop, whom I would like to dedicate this thesis to. Nana
and Pop Doyle have supported me in every endeavor of life, and this PhD experience is no
exception. Our weekend phone calls were my time to vent and talk, to seek advice, and to ask
for support and prayers. Thank you for attending one of my very first conference talks in San
Diego. Thank you for always asking, “Do you want anything, do you need anything?” Thank you
for the Chatham Candy Manor sugar highs that kept me working through lab. Thank you for
serving as examples of persistence, faith, and love in professional and personal life. Nana and
Pop, you are the reason I didn’t quit or ever give up, and the reason I made it through this PhD.
I hope I have made you proud.
To Mom and Dad, thank you for letting me crash your weekends at the condo from my
first weekend here. They truly saved me and kept me grounded about what’s most important in
life during this crazy journey. Thank you for always sending me back with food (because we
know I’m not the best cook), being my voice of reason, picking up the phone when I needed to
vent or cry, letting me practice my science talks (even if you couldn’t understand the third or fifth
syllables), and for always believing in me. I wouldn’t be the woman or scientist I am without you.
To Ryan, my monkeyman. Thank you doesn’t adequately express all that you’ve given to
me on this journey. Thank you for being willing to go the distance, literally and figuratively, to
allow me to pursue this dream. Thank you for driving 100s of miles in the middle of the night to
iii
surprise me, for supporting me, and understanding that my cells don’t sleep on weekends.
Thank you for moving here, even though many nights I come home crying, upset, or frustrated
and you usually have to cook. Thank you for lifting me up when I’ve fallen, believing in me and
my dreams, and giving me the love, support, and friendship that only a husband could.
To my family, including all of my siblings, nieces and nephews, and Mom and Dad
Wunder, thank you for supporting me, listening to me, guiding me, laughing with me, crying with
me, and loving me throughout this journey. Every weekend home, every trip to San Diego, every
holiday, and every FaceTime and phone call has kept me going when I didn’t think I could.
To my friends near and far, I owe you a great debt for all the ways you’ve supported me.
Yayettes, thank you for always being there for and supporting me. Veronica, thank you for
always picking up the phone to check in. Lisa, thank you for saying “yes” to being my friend at
our orientation, and then “yes” again to becoming roomies. From concerts to wine nights to
weddings, we’ve shared amazing memories together. Roomies forever! Louise, I am so grateful
you joined our lab and became my instant best friend. Our love of kitties, trip Down Under,
happy hour dates, and half-marathon craziness are memories I’ll have forever. I am so thankful
for you! “Uncle” Dion, thank you for always being there to listen, get drinks, have a good time,
and for becoming one of my lifelong friends. Finally, thank you to Krissy Rose, my greatest
cheerleader and best friend of all, for always lifting me up and encouraging me to finish my PhD.
Science is a team effort, and I could not have made it without my Ichida lab “family
members” both past and present. From helping feed my cells, to guiding me through math
problems, to showing me the ropes on the microscope, to providing feedback and ideas during
lab meetings, the things I’ve learned from my lab mates are too many to count. I hope I’ve
taught you all as much as you’ve taught me. I’m forever grateful for your support and friendship.
To Katie Galloway, my partner on this project and real-life friend and mentor. When we
got scooped on Christmas Eve in 2014 and recruited you to join this project, I never expected
the friendship and partnership we would build together. Thank you for always providing critical
iv
insight about the project, showing me new ways to think about things, critiquing my often
incorrect hypotheses, and encouraging the motto “if you can, flow it!”. Moreso, thank you for
serving as an example of grace and poise during the challenging times of lab and life. You’re a
wonderful example of work-life balance, and someone I aspire to be like. I can’t wait to see the
things you accomplish in your new lab at MIT. Team p53DD, we make fetch happen!
To my committee members, Neil Segil and Qi-Long Ying, thank you for being there from
the beginning of my graduate career. Thank you for giving advice and critical feedback that
provided new direction and ideas for this project, and thank you for your support on this PhD
journey and beyond.
To my mentor, Justin Ichida, thank you for taking a risk and letting me join your new lab
at USC. You have always challenged me to think critically and carefully, traits that will greatly
benefit me in future career endeavors. You’ve pushed me to work harder than I ever have, but
the success and achievement of this project have been so worth it. Thank you for providing
intellectual stimulation, bench side support, and for giving me opportunities to present my work
to the world (both scientists and the lay public). Thank you for the encouragement and guidance
I needed to grow and change into the scientist I am today. I am truly grateful for all that I’ve
learned during my time in your lab.
If you asked me seven years ago where I saw myself today, I never would have thought
that life would bring me here to Los Angeles, the City of Angels and a million dreams. Truthfully,
I didn’t think that I would “survive” or last out here in a new city in a new state by myself. It may
have taken time and patience I certainly didn’t have, but I found a home and made a life here I
never thought possible. I proved to myself, and probably most of my family and friends, that I
can do hard things, that I can persevere, and that I did it! I’m finally Doctor Wunder!

“I can do all things through Christ who strengthens me.” – Philippians 4:13
v
Table of Contents
Acknowledgments ……………………………………………………………………………………..ii

Table of Contents ……………………………………………………………………………………....v

List of Figures.…….…………………………………………………………………………………….vi

Chapter 1: Introduction to Cellular Reprogramming……………...…………………………….. 1
1.1 Brief history of cellular reprogramming…………………………………………………………….2
I. SCNT…………………………………………………………………………………………....3
II. Induced pluripotency……………………………………………………………………........6
III. Direct lineage conversion/reprogramming, transdifferentiation………………………....7
1.2 Challenges to reprogramming.……………………………………………………………...........11
I. Limitations of reprogramming efficiency…………………………………………………...11
II. Epigenetic remodeling……………………………………………………………..............13
III. The role of Trp53 in reprogramming……………………………………………………...15
IV. Elite, deterministic vs. non-deterministic, stochastic reprogramming…………………18
V. Heterogeneity……………………………………………………………...........................20
VI. Functional maturity, aging and relevance for regenerative medicine and disease
modeling………………………………………………………................…….......................22
1.3 Thesis organization……………………………………………………………............................25

Chapter 2: Identification of a genetic and chemical cocktail drives robust conversion,
enhances acquisition of target cell identity………………………………...…………………….26

Chapter 3: Genetic and chemical cocktail enhances conversion by mitigating conflicts
between transcription and proliferation……..…………………………………………………… 42

Chapter 4: Conclusion………….…………………………………………………………………….82
1.1 Summary of findings……………………………………………………………...........................83
1.2 Future directions and work…………………………………………………………….................84
I. DDRR – alternative, transient approaches………………………………………………..84
a. Trp53 inhibition…………………………………………………………….............85
b. Mutant RAS……………………………………………………………..................85
II. The HHC state, cell cycle/proliferation, and genome integrity………………………….87
a. Dynamics and existence of the HHC state………………………………………87
b. Cell cycle and proliferation………………………………………………………...88
c. Genome integrity…………………………………………………………...............89
III. Other applications for DDRR and HHC-mediated systems…………………………….90

Appendix I: Materials and Methods………….……………………………………………….…….93

Appendix II: References………….……………………………………………..…………………..103

vi
List of Figures

Chapter 1:
1.1 – Depiction of the Waddington epigenetic landscape

1.2 – Early history of cell types generated via direct lineage reprogramming
1.3 – Up- and downstream signaling pathways of tumor suppressor Trp53 as reported in
reprogramming

1.4 – Transdifferentiation from fibroblast to induced motor neuron (iMN)

Chapter 2:
2.1 – Transcription factor overexpression induces genomic stress
2.2 – Effect of Mbd3 and Gatad2a on iMN reprogramming
2.3 – Genetic and chemical factors relieve genomic stress and reprogramming block
2.4 – DDRR accelerates shift towards neuronal transcriptional state
2.5 – iMN populations partition into separate clusters, exhibit distinct patterns of gene
expression
2.6 – DDRR enhances adoption of the induced motor neuron transcriptional program
2.7 – Addition of DD accelerates morphological maturation
2.8 – The DDRR cocktail boosts reprogramming across multiple mouse paradigms
2.9 – The DDRR cocktail enhances reprogramming across starting cell types and species
2.10 – DD enhances maturity of human-derived iMNs

Chapter 3:
3.1 – Reprogramming factors induce a state of hypertranscription
3.2 – Hyperproliferating cells convert to iMNs at significantly higher rates than non-
hyperproliferating cells

3.3 – Early proliferation drives iMN reprogramming
3.4 – DDRR expands rare hypertranscribing, hyperproliferating (HHC) population
vii
3.5 – Sustained transcription in hyperproliferative cells drives conversion
3.6 – DDRR increases reprogrammable population
3.7 – Hypertrannscribing, hyperproliferating cells reprogram at near-deterministic rates
3.8 – DDRR drives activation of Hb9::GFP, morphological remodeling to iMNs
3.9 – Activation of neuronal genes differentiates successful from unsuccessful reprogramming
3.10 – DDRR sustains transgene expression in hyperproliferating cells during reprogramming
3.11 – DDRR does not promote reprogramming by increasing viral integrations

3.12 – DDRR accelerates reprogramming cells across a conserved trajectory

3.13 – DDRR upregulates topoisomerases to enable combined hypertranscription and
hyperproliferation

3.14 – DDRR-mediated topoisomerase expression drives robust reprogramming

3.15 – Topoisomerases enable DDRR-mediated reduction of negative supercoiling

3.16 – DDRR rescues transcription factor-induced negative supercoiling, reduces DNA torsion

3.17 – DDRR reduces formation of R-loops via topoisomerases

3.18 – DDRR restores replication fork processivity

3.19 – DDRR sustains high levels of transcription in actively synthesizing cells via
topoisomerases

3.20 – Model of topoisomerase-mediated reprogramming through hypertranscribing,
hyperproliferating cells

1

Chapter 1

Introduction to Cellular Reprogramming

2

Cellular reprogramming has fundamentally reshaped our understanding of development,
biology, and the promise of personalized regenerative medicine. The plasticity of cells and their
ability to change fate, whether naturally or through forced de- or transdifferentiation, transforms
how the field of biomedical research is approaching questions about biology and disease.
Serving as cellular models of diseased cell types, reprogrammed cells enable high-throughput
drug screening on patient-derived cells, accelerating the identification of therapeutics for
incurable diseases like amyotrophic lateral sclerosis. Additionally, reprogramming provides a
context to understand principles of cells fate transitions such as pathological transformation in
cancer. These are just two examples of the potential opportunities in cellular reprogramming for
biomedical and translational research. The advent of reprogramming strategies generated an
array of methods each with unique potential to address complex biological questions.
Reprogramming technologies are diverse in technique, approach, and outcome, each
accompanied by its own benefits and limitations. To enable accurate modeling and study of
cellular and molecular biology and disease, recent work in the field of reprogramming has
focused on making protocols more robust, reliable, and efficient. Somatic cell nuclear transfer,
induced pluripotency, and direct lineage reprogramming are three key areas of cellular
reprogramming whose histories and shared challenges lay the groundwork and foundation of
this thesis.

1.1 Brief History of Cellular Reprogramming
Since the late 19
th
– early 20
th
century biologists have investigated cellular stability,
plasticity, and differentiation ability. Looking to understand tissue homeostasis, development,
and regeneration, biologists wondered if development and differentiation was terminal. Was the
transition from an undifferentiated to a more differentiated cellular state irreversible and
inescapable (Takahashi 2012)? As early as the 1920s, several novel studies proved cell fate to
be flexible and reversible in amphibian systems (Spemann 1938; Briggs and King 1952; Gurdon
3

et al, 1962). In fact, cell fate transition was achieved not only in mammalian systems between
related germ lineages, for instance from glia to neurons (Heins et al, 2002), but also between
more distantly related lineages as demonstrated in generating induced pluripotent stem from
fibroblast cells (Takahashi and Yamanaka 2006). Many of these studies will be described in-
depth in this Introduction. These studies revolutionized the Waddington epigenetic landscape
and our thoughts on cellular plasticity, demonstrating that differentiation can be achieved
through “downhill,” “upwards,” and “sideways” cellular transitions (Fig. 1.1A-C, respectively).
Figure 1.1. Depiction of the Waddington epigenetic landscape. (A) In normal development, a pluripotent cell (blue) “rolls” down
the hill of development until it reaches its terminal differentiated fate (red). (B) In pluripotent reprogramming, a terminally-
differentiated cell (red) is able to traverse back “up” to a pluripotent fate (blue). (C) In direct reprogramming, differentiated cells (red)
can change fate to a cell of a different lineage (purple) by traversing “sideways” along the landscape instead of returning to a
pluripotent state.

I. Somatic Cell Nuclear Transfer (SCNT)
Studies in the early 1920s by Hans Spemann and Hilde Mangold first demonstrated
cellular plasticity with the concept of induction and the “Spemann-Mangold organizer.” Induction
is the idea that cell fate and differentiation can be dictated and/or influenced by signals from
surrounding cells (Spemann and Mangold 1924). They showed that an area of the embryo, i.e.
the “organizer,” was capable of directing cellular differentiation of other parts of the embryo. This
idea remains at the cornerstone of embryology today. Later, Spemann’s baby hair and
salamander egg experiments demonstrated nuclear totipotency, specifically, in the generation of
twin salamanders from a differentiated nucleus. (Spemann 1938). This work, and his proposition
that older embryos may have similar differentiation and potency potential, led to the first
4

successful somatic cell nuclear transfer experiments by Briggs and King in 1952. Their
experiments were fundamental in validating that differentiated nuclei, blastula more specifically,
can in fact be transplanted to enucleated eggs and returned to a totipotent state that then gives
rise to developed tadpoles (Briggs and King 1952).
Just ten years later, John Gurdon expanded on Briggs’ and King’s research. In his Nobel
Prize-winning work in nuclear transfer, Gurdon sought to answer if cellular plasticity was limited
to “younger,” less differentiated cells, or a general phenomena that included more developed,
differentiated cells. Gurdon collected differentiated intestinal epithelium from developed tadpoles
and transferred them to enucleated Xenopus frog eggs. The differentiated, transplanted nuclei
were able to give rise to fully developed frogs (Gurdon 1962). These groundbreaking
experiments proved that differentiated, somatic nuclei, when transplanted to enucleated eggs,
have the plasticity and ability to return to a totipotent state that can then give rise to many
differentiated tissues in the developed organism. Gurdon’s discoveries proved that
differentiation is reversible. Totipotent, self-renewal abilities can be restored to differentiated
cells and nuclei. Additionally, these de-differentiated cells do contain the necessary genetic
information and material to give rise to all other somatic cells in a developed organism (in this
case, a frog).
While a monumental concept, neither Briggs and King’s nor Gurdon’s experiments
proved such cellular plasticity existed in mammalian cells. This changed in 1996 with the cloning
of “Dolly,” the sheep. Nuclei derived from cultured adult sheep mammary gland cells
transplanted into enucleated sheep eggs enabled the generation of an entire mammalian
organism (Campbell et al; 1996). With the birth of Dolly came the realization that mammalian
cells, like their amphibian counterparts, are also capable of dedifferentiation. Since Dolly, a wide
variety of animal species have been derived using SCNT including mice, pigs, rats, dogs, and
horses (Campbell et al, 2007). Most recently, this technology has been used to generate human
ESC-like cells. Tachibana and colleagues first demonstrated the use of SCNT in generating
5

human nuclear transfer derived embryonic stem cells (NT-ESCs) in 2013 (Tachibana et al,
2013), and many more studies have emerged since then.
Somatic cell nuclear transfer holds much promise for biology and medicine, particularly
in the areas of agriculture, conservation, and study and treatment of human disease. However,
SCNT is a technique that suffers from high costs, rigorous methodology, and varied
reprogramming outcomes. Differences in enucleation, oocyte maturity, embryo quality, and
culture conditions are several areas of variation that, subsequently, lead to varied outcomes
(Campbell et al, 2007). Most problematic, many SCNT-derived organisms suffer from mild to
severe phenotypic abnormalities, owing in part to many of the aforementioned technical
differences. To accurately model and study animal and human disease and development,
phenotypic abnormalities due to technical variation is very problematic. They could potentially
skew what may be observed as a disease or developmentally relevant finding in the in vitro
cells, thus making it difficult to draw meaningful conclusions without separating technological
artifact from actual phenotype. Additionally, high-throughput scalability is a major challenge with
such a technically complicated methodology. To accurately model human disease, large cohorts
of control and patient samples, or many replicates of fewer control and patient samples, must be
assessed in tandem for phenotypes to provide any disease-relevant meaning. Some progress
has been made in standardizing aspects of the methodology, for example modifying the donor
cells and recipient oocytes and/or changing culture media to include factors capable of
supporting proper gene expression patterns (Campbell et al, 2007). However, for SCNT to be a
reliable model of animal and human development and disease, robust, reproducible generation
of “normal” cells that only present with phenotypes under disease conditions are important to
explore and optimize.

6

II. Induced Pluripotency
The groundbreaking studies by Gurdon and others in the field of SCNT demonstrated
unexpected cellular plasticity of differentiated cells. After the demonstration of SCNT in
mammals, developmental biologists pondered if such plasticity could be shown without use of
nuclear transfer with eggs. Thomson and colleagues derived human embryonic stem cells
(ESCs) from the inner cell mass of embryos in 1998. These ESCs exhibited long-term
proliferation without differentiation as well as germ layer competency, or contribution to all three
germ lineages (Thomson et al, 1998). These findings were demonstration that indeed, plasticity
of differentiated cells was not limited to nuclear transfer techniques. However, use of human
embryos for biomedical research slowed due to ethical and moral concerns as well as the
breakthrough in inducing pluripotency from somatic cells.
Yamanaka and colleagues first discovered that mouse fibroblasts could be differentiated
back to an embryonic stem cell (ESC)-like state without use of eggs or, more important, human
embryos in 2006. After identifying a panel of 24 genes associated with ESC maintenance and
stability, each gene was introduced into mouse embryonic fibroblasts via retroviral transduction.
While individual genes did not elicit an ESC-like phenotype (observed via activation of Fbxo15),
co-transduction of all 24 genes did induce an ESC-like phenotype 30 days post-transduction.
Subsequent drop out experiments led to the eventual identification of a core cocktail of four
factors (Oct4, Sox2, Klf4, and c-Myc, i.e. “Yamanaka factors”) deemed essential in deriving this
pluripotent cellular state. Termed induced pluripotent stem cells (iPSCs), iPSCs expressed
many hallmark genes associated with ESCs, exhibited morphological similarities to ESCs, and
contributed to several different tissues when injected into mouse blastocysts, demonstration of
germ layer competency (Takahashi and Yamanaka 2006). Just a year later, Yamanaka and
colleagues discovered self-renewal abilities in human cells in their creation of iPSCs from adult
human fibroblasts using these same “Yamanaka factors” (Takahashi et al, 2007). These studies
7

revealed that key transcriptional pluripotency networks could be harnessed as tools to induce
cellular fate switching without nuclear transfer, use of eggs, and human embryos.
Since these studies were introduced over a decade ago, iPSCs have been generated
from many different species and starting cell types (Sing et al, 2015; Stadfield and
Hochedlinger, 2010), demonstrating the evolutionary conservation of this pluripotent, self-
renewal transcriptional program. iPSCs are now being used to study and model nearly every
disease from cancer, to neurodegenerative disorders, to diabetes, to kidney disease.
Furthermore, scientists are pursuing such studies on a patient-specific level, using a person’s
own skin or blood cells to create iPSCs with which to model their particular disease. This Nobel
Prize-winning work was a turning point in stem cell and developmental biology, and opened the
door to the reality of patient-specific disease modeling and regenerative medicine research.

III. Direct Lineage Conversion/Reprogramming, Transdifferentiation
The advent of iPSC technology revolutionized biomedical research. After discovering
self-renewal abilities of mouse and human cells, the reprogramming field was emboldened to
examine if direct differentiation between somatic cells was possible. Previous work established
transcriptional networks and programs as one route to inducing lineage fate switching between
somatic cells, i.e. direct lineage conversion/reprogramming or transdifferentiation.
Transdifferentiation relies on lineage-specific transcription factors that drive and/or maintain cell
fate and cell-type-specific gene expression to induce cell fate switching from one somatic cell to
another. Importantly, this technology could directly generate cell types currently inaccessible for
study of disease and development. Davis and colleagues demonstrated the first example of
transdifferentiation from fibroblasts to muscle fibers with simple overexpression of transcription
factor MyoD (Davis et al, 1987). Additional studies showed that conversion between related
lineages could be achieved through transdifferentiation, for example from glial to neuronal cells
8

(Heins et al, 2002) and pancreatic to liver cells (Shen et al, 2000). However, it was unclear if
fate conversion between more distant germ lineages was possible.
Vierbuchen and colleagues were first to use transcription factor-mediated
transdifferentiation in 2010 to create neurons from fibroblasts, cell types derived from distinct
germ lineages. Utilizing a strategy similar to Takahashi and Yamanaka, their group screened a
candidate pool of 19 neuronal-specific genes for activation of a TauEGFP reporter in transgenic
mouse embryonic fibroblasts (MEFs). Subsequent screening and drop out studies revealed a
core cocktail of only three transcription factors (Brn2, Ascl1, and Myt1l, (i.e.. BAM) were
necessary to induce EGFP reporter activation in as little as 12 days after transduction.
Electrophysiology studies further confirmed these to be functional induced neuron-like cells (i.e.
iNs) (Vierbuchen at al, 2010). Soon after, other groups discovered that transdifferentiation from
fibroblasts to induced cardiomyocytes (Ieda et al, 2010) and induced hepatocytes (Sekiya and
Suzuki 2011) were also possible using this transcription factor-mediated approach to
reprogramming.
The discovery of direct lineage reprogramming accelerated rapidly after these early
studies given the promise, possibility, and speed of producing previously inaccessible cells from
patients for study and treatment of disease. Speed represented a major advantage of this
technology. Induced cells could be generated in a few days to a few weeks compared to the
weeks of culture that derivation and maintenance of iPSCs required. Hundreds of publications
have since used this strategy to derive induced cell types from both mouse and human starting
cells (Fig. 1.2). These induced cell types are being used to study disease pathogenesis,
mechanisms of disease onset, and pursue personalized therapeutic screening in unique,
patient-specific genomic contexts.
9

Figure 1.2. Early history of cell types generated via direct lineage reprogramming. Since transdifferentiation of fibroblasts
myoblasts was discovered (left), cell types of the brain, heart, and others have been generated using this technology. Studies listed
above the line refer to mouse, and studies below the line refer to human. Unless noted, above listed are in vitro studies.

a. Induced motor neurons
Studying diseases of the central nervous system has been limited by access to relevant
cell types. Generating cells of the central nervous system using transdifferentiation provides the
opportunity to model neurodegenerative and other complex diseases of the brain. Neurological
disorders and neurodegenerative diseases are often characterized by phenotypic and functional
problems of a particular neuronal cell type. ALS is a devastating disease characterized by the
selective loss of spinal motor neurons in the body. These cells are responsible for muscle
movement and contraction in the body, and their loss leads to difficulty breathing and eating,
paralysis of the face and limbs, and often death. Direct lineage reprogramming represents an
ideal technology to create these cells with which to gain deeper understanding of their
properties and the disease.
Given the remarkable success of lineage or cell type-specific transcription factor pools in
inducing transdifferentiation (e.g. cardiomyocytes, hepatocytes), Son and colleagues tried a
similar approach to create spinal motor neurons in vitro (Son et al, 2011). After identifying a
group of 11 candidate genes with established roles in the development, function, and
specification of spinal motor and other neuronal subtypes, the 11 genes were retrovirally
10

transduced into Hb9::GFP transgenic MEFs. Hb9, a motor neuron-specific gene, drives GFP
expression, thus allowing conversion to a motor neuron fate to be monitored via GFP reporter
activation.
From the 11 genes that successfully activated the GFP reporter, subsequent drop out
experiments revealed a minimal cocktail of 7 factors were required for generating Hb9::GFP+
induced motor neurons (iMNs). Characterization revealed that iMNs properly expressed several
pan-neuronal markers (eg. Map2, Tuj1), exhibited similar gene expression patterns to primary
mouse embryonic motor neurons, and performed as functional motor neurons in
electrophysiology and spinal cord integration experiments. More importantly, when subjected to
an ALS-like disease environment in co-culture experiments with SOD1G93A mutant glial cells,
iMNs exhibited significant reduction in survival compared to counterparts co-cultured with
wildtype glia (Son et al, 2011).
Recent work has expanded to human iMNs and demonstrated ALS-like pathologies of
patient-derived iMNs in vitro (Shi et al, 2018). Shi and colleagues demonstrated that C9ORF72
ALS patient induced motor neurons present hallmark disease characteristics in a dish and, more
important, could be used to identify novel mechanisms of disease pathogenesis (Shi et al,
2018). Patient and control cells were differentiated into iMNs and characterized for survival and
functional differences under ALS-like environmental conditions. For example, neurons were
subjected to excess glutamate signaling conditions, mimicking high glutamate levels present in
patients’ CSF. The C9ORF72 patient-derived iMNs exhibited significantly reduced survival in
these conditions compared to control-derived iMNs. Further, this survival was linked to reduced
levels of the C9ORF72 protein. Restoration of the protein rescued this survival deficit.
Additionally, functional assays revealed these cells display excitotoxicity in response to excess
glutamate signaling, as well as impaired vesicle trafficking. Together, these properties impair
motor neuron ability to function and respond to stimuli, leading to their eventual degeneration.
Studies with reprogrammed cells illuminated for the first time that both gain and loss of
11

C9ORF72 plays a role in ALS pathogenesis. These findings reveal the utility of direct
reprogramming to effectively mimic ALS disease phenotypes and, more interesting, to identify
novel cause(s) of disease pathogenesis. This study has been just one of many to demonstrate
that induced cells generated in vitro might serve as reliable proxies to model human disease.

1.2 Challenges to reprogramming
With exciting potential to impact disease modeling, reprogramming technologies have
been limited in key areas: poor and/or incomplete reprogramming, non-deterministic
reprogramming events, and inability to generate functionally mature, phenotypically-relevant
induced cells. The crux of this thesis has been to identify ways to overcome the primary
roadblocks to direct reprogramming. To explore this, it is important to note several crucial
advancements that have been made or proposed to overcome the challenges of these
techniques collectively, as much basis for this thesis was grounded in these previous findings.

I. Limitations of reprogramming efficiency
Despite the progress made towards using iPSC technology to study human development
and disease, it is not without its challenges and limitations. Many of the first studies in the iPSC
field report reprogramming efficiencies as less than 1% (Takahashi and Yamanaka 2006;
Takahashi et al, 2007), not nearly enough to use for disease modeling which can require 100s-
1000s of cells, and not necessarily all of one cell type. Additionally, the “first generation” of
iPSCs made by Takahashi and Yamanaka were found to be only partially reprogrammed.
Evidence of this inefficient conversion is inability of the reprogrammed cells to sustain
expression of endogenous “Yamanaka factor” genes, for example through demethylation of
endogenous Oct4. This hinted at the idea that simple overexpression of key transcriptional
regulators may be insufficient to support a complete cell transition, and that remodeling a more
12

complex epigenetic landscape during reprogramming should be considered to support fully
efficient transitions. This will be discussed at greater length in the proceeding section.
Additionally, re-activation of the viral transgenes in somatic cells has hindered their
developmental and differentiation potential, often leading to tumor formation when transplanted
into mice (Stadfield and Hochledinger 2010; Takahashi and Yamanaka 2006). The risk of
teratoma and tumor formation has thus severely undermined the potential use of iPSCs for cell
or gene therapies in patients. Biologists have vigorously pursued ways to improve this core
cocktail of “Yamanaka factors” to eliminate these problems. Removal of factors like oncogenic
c-Myc (Nakagawa et al, 2008), and/or addition of small molecule compounds to supplement or,
sometimes, to replace the factors themselves, have modestly improved reprogramming
efficiency and iPSC quality (Hou et al, 2013; Huangfu et al, 2008a and b; Ichida et al, 2009;
Lyssiotis et al, 2009; Shi et al, 2018). Additionally, use of inducible lentiviral constructs, non-
integrating viruses, and recombinant proteins are other methods employed to reduce the risk of
viral integrins affecting iPSC reprogramming and quality.
Poor reprogramming efficiency is also frequently reported in reprogramming to a variety
of somatic cell types (eg. ~10% conversion to iMNs (Son et al, 2011)). To model complex
neurodegenerative diseases characterized by altered neuronal phenotypes, enough cells must
be created so that accurate inferences about potential disease-like phenotypes can be drawn.
Unlike iPSCs that are capable of expansion, most direct reprogramming paradigms do not
undergo a documented highly proliferative transitive intermediary stage, though. Extensive work
has thus been done to improve direct reprogramming efficiencies. Groups have discovered that,
similar to iPSC reprogramming, cell cycle and senescence play regulatory roles in neuron
conversion. Inhibition of cellular senescence, induced by silencing p16/
Ink4
a and p19/
Arf
, and
overexpression of human telomerase reverse transcriptase were sufficient to improve
conversion of human fibroblasts to functional neurons (Sun et al, 2014). Interestingly, induction
of cell cycle arrest during the G1 phase of the cell cycle improved conversion of human
13

fibroblasts to induced dopaminergic neurons (iDA) (Jiang et al, 2015). Previous studies have
revealed DNA synthesis is crucial in achieving successful cell fusion-mediated reprogramming
to allow for nucleotide incorporation in reprogramming cells. This may explain why, immediately
after S-phase and nucleotide incorporation of the reprogramming factors, arrest in G1 was
sufficient to enhance reprogramming.
Several groups have also explored chromatin modifications as ways in which to boost
these low reprogramming efficiencies. Addition of the neuronal specific micro-RNAs miR-9/9*
and miR-124 to defined cocktails of reprogramming factors (eg. Isl1 and Lhx3) have significantly
increased neuronal reprogramming (Abernathy et al, 2017). Abernathy and colleagues propose
that these miRs induce a chromatin switch to an environment more permissive to neuronal
reprogramming. Interestingly, other groups have also found that addition of microRNAs enhance
reprogramming efficiency, particularly to neuronal subtypes (Jiang et al, 2015; Yoo et al, 2011).
Epigenetic modifications will be discussed at greater depth in the proceeding section.
Together, these studies highlight that challenges to robust reprogramming remain
undiscovered. Progress has been made identifying some shared roadblocks limiting
reprogramming to iPSCs and somatic cell types (these will be discussed at length in the
proceeding sections). However, generating large numbers of highly homogeneous cellular
subtypes remain the major obstacle to using these technologies for in vitro and in vivo
regenerative medicine and disease modeling.

II. Epigenetic remodeling
Upon initial exploration of mechanisms hindering robust reprogramming, biologists
proposed that efficient reprogramming might require modification beyond simple transcriptional
regulation. Groups have since worked to identify and target components of the epigenome that,
when either inhibited or activated, improve reprogramming. Early studies focused on exploration
and use of pioneer factors to induce reprogramming-permissive chromatin states. Pioneer
14

factors are transcription factors that are able to engage silenced genomic regions usually
inaccessible by other transcription factors (Iwafuchi-Doi and Zaret 2014). They have been
described as crucial components of reprogramming cocktails due to their ability to create
genomic and epigenetic landscapes permissive for other reprogramming factors to bind and
engage. For instance, Oct4, Sox2, and Klf4 have been identified as iPSC pioneer factors due to
their ability to bind closed chromatin and thus promote reprogramming (Soufi et al, 2012). Work
has further refined how pioneer factors exert their functions during reprogramming, with Oct4
recently described as recruiting chromatin remodeling protein BRG1 to support additional
binding and expression of other factors (King and Klose 2017). Additionally, both Ascl1 and
Neurog2 have been reported as pioneer factors in direct reprogramming to neurons due to their
ability to bind closed chromatin regions (Smith et al, 2016; Wapinski et al, 2013). Greater
exploration of the role for Ascl1 in neuron reprogramming describe it inducing accessibility
changes in ~80% of its downstream genomic loci within just the first few days of reprogramming
(Wapinski et al, 2017). This single major switch in chromatin state activates many endogenous
neuronal loci followed by subsequent maturation of the induced neurons (Wapinski et al, 2017).
Together, these studies reveal crucial roles for some transcription factors in inducing
reprogramming-permissive changes to the epigenetic landscape.
Additionally, biologists have explored chemically removing or modifying repressive
marks that hinder transcription factors binding to their target DNA sequences. When inhibiting
histone deacetylases (HDACs) using valproic acid (Huangfu et al, 2008a and 2008b), or adding
epigenetic modifiers like DZnep (Hou et al, 2013), iPSC reprogramming efficiency increases up
to 100-fold. Additionally, knocking down Mbd3 or other members of the nucleosome remodeling
and deacetylase (NuRD) repressor complex (eg. Gatad2a, Chd4), leads to 100%, near
deterministic reprogramming (Mor et al, 2018; dos Santos et al, 2015; Rais et al, 2013).
However, while these approaches have benefited iPSC reprogramming, they have not been
demonstrated to universally promote reprogramming in other contexts, for instance in direct
15

reprogramming, suggesting that use of such general chromatin-modifying small molecules is
limited to certain cellular contexts.
Biologists have also explored microRNAs (miRNA/miR) as other approaches to
modifying epigenetic landscapes during reprogramming. miRNAs miR-9/9* and miR-124 are
responsible for silencing neural precursor BAF-chromatin remodeling complex BAF53a and
inducing a switch to BAF53b, a neuron-specific complex (Yoo et al, 2011). Yoo and colleagues
first described use of miR9/9* and miR-124 as capable of inducing direct reprogramming from
fibroblast to neuron. Additionally, these neuronal specific miRNAs miR-9/9* and miR-124 have
been added to defined cocktails of reprogramming factors (eg. Isl1 and Lhx3), increasing
reprogramming to induced spinal motor neurons at upwards of 80% efficiency (Abernathy et al,
2017). Abernathy and colleagues propose that addition of these miRs induce a chromatin switch
to an environment more permissive to reprogramming through extensive remodeling,
specifically to a pan-neuronal program. Interestingly, other groups have also found that addition
of miRs enhance reprogramming efficiency, particularly to neuronal subtypes (Jiang et al, 2015).
While promising for direct reprogramming to neurons, use of microRNAs to promote
reprogramming is cell-type specific and thus not generally applicable across paradigms.
Great progress has been made in identifying ways to modify or modulate the epigenomic
landscape to promote iPSC and direct reprogramming. However, it is important to note that
using small molecule inhibitors of epigenetic components may yield unintended consequences
due to their non-specificity and off-target effects. Additionally, many of these findings and
studies are not generally applicable across different reprogramming paradigms,

III. The role of Trp53 in reprogramming
Perhaps the most well-established and studied roadblock to iPSC reprogramming is the
Trp53 pathway and its interacting partners and targets. The physiological role of Trp53 is to
maintain genomic stability, “guarding” from mutations and other harmful aberrations by halting
16

cell cycle or promoting cellular senescence, inducing DNA repair, or promoting apoptosis.
Owing to the fact that several of the “Yamanaka factors” are oncogenic or have oncogenic
potential (eg. c-Myc, Klf4, and Lin28) (Zhao and Xu 2010), or are proposed to inhibit cell cycle
and proliferation, it was perhaps not surprising that several groups showed its inhibition
significantly enhanced reprogramming efficiency (Zhao et al, 2008; Hanna et al, 2009; Hong et
al, 2009; Kawamura et al, 2009; Li et al, 2009; Marión et al, 2009; Utikal et al, 2009).
Researchers described many ways in which Trp53 serves as a negative regulator of
reprogramming. Several groups described induction of cell cycle arrest via interaction with cyclin
dependent kinase inhibitor p21 (Hong et al, 2009; Kawamura et al, 2009), or induction of cellular
senescence via activation of the Ink4/Arf locus (activated by the p53-p21 axis) (Li et al, 2009;
Utikal et al, 2009). Still other groups propose that apoptosis, another key function of p53,
activated by accumulated DNA damage, is what limits reprogramming (Marión et al, 2009). In
either sustained or transient conditions of Trp53 inhibition (using shRNAs, siRNAs, or small
molecule inhibitors like pifithrin), reprogramming to iPSCs increased upwards of 100-fold.
However, genomic integrity of the Trp53-deficient cells was compromised, as colonies
frequently exhibited chromosomal aberrations and were prone to malignant transformation.
Study and modeling of human disease requires genomically stable and karyotypically normal
cells. However, transient, non-integrating Trp53 inhibition may be sufficient to suppress its
activity while still soliciting a robust reprogramming response. Collectively, these studies
highlight critical roles for Trp53 in maintaining genomic stability of reprogramming cells,
including induction of apoptosis, cell cycle arrest, and DNA repair (Fig. 1.3). They also suggest
an interesting link to cancer, given that initiation and development of many are characterized by
inactivation or mutation of Trp53.
17

Figure 1.3. Up- and downstream signaling activities of tumor suppressor Trp53 as reported in reprogramming. Trp53 is
activated by a range of signals. Highlighted in green are published or proposed ways that Trp53 is induced during reprogramming.
Highlighted in red are published downstream activities for Trp53 during reprogramming.

Surprisingly, little work has been done exploring roles for p53 in influencing direct
reprogramming. However, a few groups have proposed ways in which p53 negatively regulates
neuronal reprogramming. Jiang and colleagues found that, upon transduction of the
reprogramming lentiviruses with an shRNA targeting p53, they observed significantly improved
conversion to iDAs compared to cultures generated with wildtype p53 (Jiang et al, 2015). They
propose that p53 normally acts to sustain the existing transcriptional network of the starting
fibroblast cells, suggesting its inhibition modulates transition to that of an iDA by downregulating
the starting transcriptional network. Additionally, a study has shown that simple inhibition of p53
with an shRNA is sufficient to induce a complete fate switch from fibroblast to induced neuron
(as well as oligodendrocytes and astrocytes) (Zhou et al, 2014). They suggest that by inhibiting
p53 and culturing their cells with neuronal media, they achieve conversion from fibroblast to
neuron via upregulation of neurogenic transcription factors Ascl1, Brn2, and Neurod2. However,
the latter results are unconvincing, as attempts by our group to repeat these findings were
unsuccessful in achieving such fate switching.

18

IV. Elite, deterministic vs. non-deterministic, stochastic reprogramming
Several groups have hypothesized reasons for the seemingly random, inefficient nature
of iPSC reprogramming. Some of the first models were proposed by Yamanaka and suggested
that iPSC reprogramming occurs through either an elite model or a stochastic model
(Yamanaka 2009).
The “elite” model suggests that all, or only a small group of cells are amenable to viral
infection and conversion, driving most of the successful reprogramming events. However,
several lines of evidence suggest this not to be the case. One reason is that a wide variety of
starting cell types with differing proliferation and differentiation properties have been
successfully reprogrammed to iPSCs (Eminli et al, 2009; Hanna et al, 2008; Stadfield et al,
2008; Zhou et al, 2008; Takahashi and Yamanaka 2006; Takahashi et al, 2007). Additionally,
reprogramming has since been achieved without use of integrating viruses as was first reported
(Stadfield et al, 2008; Okita et al, 2008; Yu et al, 2009). This model would suggest that, despite
diversity in methodologies influencing cell fate change, the same group(s) of cells respond to
reprogramming stimuli. However, it is highly unlikely that the same cells are always predestined
to respond to such varied reprogramming approaches. Given the diversity of starting cell types
and techniques utilized, together these studies suggest against a “one cell fits all” idea of elite
reprogramming.
An alternative model of reprogramming, termed stochastic or non-deterministic, has also
been explored. In this model, most or all cells have the potential to respond to reprogramming
signals, but a bottleneck or other refractory event prevents most from converting. In the early
days of the iPSC field, fine-tuned control of viral integration, stoichiometry, and expression were
limited and thought to be contributing factors to poor reprogramming. Despite improvements
with inducible constructs or removal of viruses altogether (Wernig et al, 2009; Stadfield et al,
2008; Okita et al, 2008; Yu et al, 2009), reprogramming efficiencies remain low and
inconsistent, suggesting there remain elements of randomness to the process beyond viral
19

infection. In contrast, an interesting idea proposes that perhaps reprogramming is not a single-
step process, but instead an accumulation or “drift” in cell state that accelerates over time
(Hanna et al, 2009). Hanna and colleagues describe a reprogramming process wherein cellular
state, defined by gene expression, epigenome, and cell cycle, are modified and accelerate over
reprogramming time. These “acceleration” events may allow more partially reprogrammed cells
to give rise to daughter cells (with independent probability of reprogramming), and/or assist in
cell cycle-mediated epigenetic remodeling (Hanna et al, 2010; Hanna et al, 2009; Yamanaka
2009). An independent study also suggested that accelerated re-entry into cell cycle is a
hallmark characteristic of cells that go on to reprogram, perhaps for similar reasons as
mentioned above (Smith et al, 2010). These studies touch on interesting links between cell
cycle, proliferation, and epigenetic reprogramming to a pluripotent state.
There are other interesting possibilities that might be more faithful explanations of iPSC
reprogramming as suggested by the Jaenisch and Lu groups (Buganim et al, 2012; Guo et al,
2014a). Using a single cell transcriptional profiling approach, Buganim and colleagues found
that reprogramming is characterized by early stochastic events, with gene expression varying in
the reprogramming cells, followed by a non-stochastic late stage, characterized by hierarchical
activation of pluripotency genes (Buganim et al, 2012). These findings suggest that while early
events are unpredictable, late reprogramming events take on characteristic and predictable
gene expression changes.
Additionally, Guo and colleagues described reprogramming as a stochastic process
driven by a group of privileged cells that display an “ultrafast” cell cycle. While not identical,
similar phenomena were described by other groups, that acceleration of cell proliferation is an
early driver of reprogramming (Hanna et al, 2009; Smith et al, 2010), perhaps owing to
increased probability not only of daughter cells to reprogram, but also for epigenetic remodeling
to occur. Interestingly, independent studies demonstrate that certain phases of cell cycle, G1
and synthesis in particular, are deemed more conducive to different reprogramming paradigms
20

(Pauklin and Vallier 2013; Tsubouchi et al, 2013). Thus, expedited cell cycle re-entry may
increase the chances that cells are able to reach these “reprogrammable” phases. Additionally,
several groups describe roles for p53 inhibition in accelerating proliferation and/or expanding
the pool of proliferating cells (Guo et al, 2014a; Hanna et al, 2010; Smith et al, 2010; Hanna et
al, 2009), though these events are not necessarily sufficient for reprogramming on their own.
However, they lend mechanistic insight into how p53 inhibition acts on cell cycle to promote
reprogramming, perhaps by driving more cells into a “privileged,” reprogrammable state.
The dynamics of iPSC reprogramming remain elusive. However, it seems this paradigm
may in reality be neither an “either or” scenario, but instead a blend of both stochastic and
deterministic models. While populations of starting cells might stochastically give rise to iPSCs,
“privileged,” fast-cycling cells are an elite group within the population that undergo predictable
changes in gene expression that ultimately give rise to most of the resulting converted cells. In
harnessing tools to modulate cell cycle and cell proliferation, particularly to accelerate re-entry
into cell cycle, perhaps slower cycling cells normally unequipped to reprogram will be endowed
with “privileged” abilities more capable of reprogramming.

V. Heterogeneity
Heterogeneity of the reprogrammed population serves as another roadblock to robust
direct reprogramming and its use in biomedical research. For instance, to study a disease like
ALS, characterized by selective and specific loss of spinal motor neurons, contamination of
other cell types generated during direct conversion will negatively influence how results of iMN
phenotype and function are interpreted. Put differently, the “signal” of the iMNs should far
outweigh the “noise” from non-iMN cells in the dish to accurately depict and model disease.
Thus, assessing and reducing the heterogeneity of reprogrammed cultures compared to their
bona fide counterparts is paramount to yield meaningful, biologically relevant results.
21

Many groups have reported high levels of heterogeneity in a variety of reprogramming
paradigms (Jiang et al, 2015; Cahan et al, 2014; Morris et al, 2014; Marro et al, 2011; Ieda et al,
2010). The most notable revelations came with the introduction of CellNet from the Collins and
Daley labs in late 2014. These groups discovered that induced cells generated via directed
differentiation of pluripotent cells are better able to establish gene regulatory networks (GRNs)
of target cells than cells derived via transdifferentiation. For instance, induced neurons sustain
relatively high expression of genes associated with a fibroblast GRN and dysregulation of other
fibroblast-associated regulators (Cahan et al, 2014). More striking, they claim the biggest
roadblock to robust conversion is a cell’s ability to effectively silence or shut down the starting
GRN in the transition from starting to target cell. While these initial studies utilized only
microarray data to make these assessments, they have since developed ways to utilize CellNet
with RNA-sequencing data. This will undoubtedly increase the power of these analyses,
providing deeper understanding of transcriptional interplay during reprogramming.
Interestingly, the Deng and Pei groups highlight a similar phenomenon, that
downregulation of the starting fibroblast program is essential in chemical-mediated
reprogramming to neurons (Hu et al, 2015; Li et al, 2015). Though a slightly different context, it
is of note that, despite the methodology with which cells are converted from one somatic cell to
another, downregulation or silencing of the starting cell transcriptional program is important for
successful fate transition (Fig. 1.4).

Figure 1.4. Transdifferentiation from fibroblast to induced motor neuron (iMN). Studies suggest that robust fate transitions
may be improved in downregulating the gene regulatory network (GRN) of the starting cell (eg. fibroblast) while upregulating the
GRN of the target cell (eg. iMNs).

22

Single cell RNA-sequencing has further strengthened our understanding of cellular
heterogeneity before, during, and after transdifferentiation. This technology has allowed groups
to dissect reprogramming trajectories from fibroblasts to cardiomyocytes (Liu et al, 2017),
neurons (Treutlein et al, 2016), and endoderm progenitors (Biddy et al, 2018) to name a few.
These studies reveal that distinct subpopulations of cells appear during reprogramming, some
of which diverge to alternate reprogramming trajectories instead of towards their target identity
(Liu et al, 2017; Treutlein et al, 2016). Deeper exploration of gene expression in cells on
“successful” versus “aberrant” reprogramming trajectories will help elucidate molecular
mechanisms underlying both and stimulate exploration of ways to re-direct the aberrant back
towards the target fate.
Reprogramming heterogeneity with regard to cell state, morphology, gene expression,
and functionality are all considerations to make when seeking ways to improve converted
cultures. Through deeper analyses of transcriptional programs and profiles with single cell RNA-
sequencing, and perhaps targeting starting networks to silence, these current problems will
improve, making the “signal” of target induced cells greater than the “noise” of non-target cells.

VI. Functional maturity, aging, and relevance for regenerative medicine and disease modeling
Recapitulating “aging” or diseases of the elderly are prominent challenges facing the
iPSC and direct reprogramming fields. Disease like Parkinson’s, amyotrophic lateral sclerosis,
and some cancers take decades to present in patients, time that 2D in vitro culture systems
inadequately model. Further, cell-cell interactions and the microenvironment of the affected cells
and tissues aren’t captured in 2D culture. Groups have since begun culturing disease-relevant
iPSCs under environmental stressors related to disease, for example oxidative stress, or using
organoids to culture multiple likely interacting cell types together. Further, iPSCs, in their
transition back through an embryonic cell-like state, experience erasure of their epigenetic
signatures and lose memory of their starting cell age (Mertens et al, 2015; Maherali et al, 2007;
23

Chambers and Studer 2007). To model and depict a disease of the aged, scientists must find
ways to retain or restore these aging signatures and patterns in their in vitro models. A study by
Ho and colleagues suggests that transcriptional profiling of “aged” motor neurons may provide
insights about targets for gene and small molecule overexpression that could induce such
signatures in iPSC-derived induced motor neurons, specifically (Ho et al, 2016).
Creating mature, functional cells is critical to faithfully modeling human development and
disease in vitro. As with iPSC reprogramming, generating cells that mimic age and phenotype of
disease cells will yield the most reliable results and analyses of disease mechanisms,
pathologies, and drug discovery. Unlike iPSC reprogramming, direct reprogramming does
indeed permit somatic cells to retain their age in transition to new cell types. As Mertens and
colleagues demonstrate, neurons derived directly from fibroblasts, not through a pluripotent
state, retain signatures of their aged fibroblast starting cell state (Mertens et al, 2015),
suggesting these induced neurons will serve as more faithful proxies for “aging” diseases.
Additionally, groups have utilized RNA-sequencing to identify factors present in “aged,”
or older spinal motor neurons (Ho et al, 2106). Though iPSC-derived motor neurons exhibit
transcriptional similarities more closely related to younger, embryonic motor neuron
counterparts, there are sets of genes that might be “targetable” to induce more aged signatures
(Ho et al, 2016). Xu and colleagues have similarly suggested that combinations of cell type-
specific determination and maturation factors may be crucial in generating functionally and
phenotypically mature induced cells for biomedical research (Xu et al, 2015). Sequencing
studies may reveal “master genes” capable of both, though this is yet to be determined for many
cell types. Perhaps the most robust method of maturation would involve downregulation of the
starting GRN, with subsequent upregulation or overexpression of the target determination and
maturation GRN.
Many diseases involve the interplay of multiple cell types for their development and/or
progression. Spinal motor neurons interact with glial cells, muscle cells, astrocytes, etc, in vivo.
24

Thus, modeling under appropriate disease states should involve environmental and cell-cell
interaction conditions, as well. For instance, the ALS-associated mutation SOD1G93A has been
described to have non cell-autonomous effects when mutated in glial cells in the presence of
induced motor neurons (Di Giorgio et al, 2007). Not only will maturation of the target cell-of-
interest be critical for reliable disease modeling, but also culture in the presence of other
disease relevant phenotypes.
Finally, reprogramming holds much promise for use in regenerative medicine of the
central nervous system (CNS). Local or surrounding cells of the damaged CNS tissue may
serve as ideal targets for in vivo reprogramming due to their proximity to the injury site, lineage-
relatedness to the injured to deteriorated neuronal cells, as well as their reported plasticity after
injury (Gascón et al, 2017). Studies have reported that both glial cells and astrocytes have been
reprogrammed into neurons in vivo using lenti-, retro-, and adeno-associated viral approaches
(Gascón et al, 2017). However, neurons generated have been difficult to define based on
specific subtype. Additionally, it has been proposed that glial cells and astrocytes, which
represent large numbers of target-able cells, might be reprogrammed into neurons using non-
invasive small molecule/chemical approaches (Gascón et al, 2017). However, there are many
important considerations for in vivo approaches to regeneration. These include but are not
limited to: targeting specific cell types for reprogramming, environmental response to
reprogramming, timing of reprogramming after injury, evaluating integration, identity, and
function of induced neurons generated, prolonging survival of in vivo generated neurons, and
reprogramming at highly homogeneous efficiencies (Gascón et al, 2017).
Despite the progress made to elucidate “maturity” factors important in target cell types,
much proof-of-concept work is yet to be done to validate that such overexpression or targeting
methods are effective at inducing “aging” signatures. Environmental and genomic “aging”
conditions in combination with appropriate co-culture will likely serve as the most reliable and
robust mimics for disease-in-a-dish. Finally, work done to understand molecular roadblocks to in
25

vitro reprogramming may carry-over and be applicable to in vivo strategies for use in
regenerative medicine and cell therapy.

1.3 Thesis Organization
In this thesis, we aim to elucidate the barriers to reprogramming and outline strategies to
improve direct cellular reprogramming for modeling of human development and disease. While
progress has been made in identifying cell type-specific barriers to direct reprogramming,
general features hindering reprogramming more generally are limited. Given that low efficiency
and high heterogeneity represent the major roadblocks to improvement for achieving robust,
reliable results using this technology, we will begin with these two barriers. We utilized
reprogramming to induced motor neurons for most of this work, a well-characterized neuronal
subtype used extensively in our lab. We have found several similarities of iPSC reprogramming
are conserved in transdifferentiation, for instance roles for cell cycle, proliferation, and Trp53 in
preventing successful reprogramming events. Our findings suggest that transdifferentiation to
induced motor neurons is likely a stochastic process governed by a privileged population of cells
able to maintain high levels of both transcription and replication. This population can be
expanded, and these privileged cells give rise to most successful reprogramming events,
reprogramming at near-deterministic rates. Harnessing these findings, we improve
reprogramming efficiency 100-fold and enable generation of more robust, homogeneous
induced motor neurons for the study of neurodegenerative disease like ALS. Our findings apply
across species and reprogramming cocktails to post-mitotic cells. Additionally, our results may
be applicable to other reprogramming paradigms, and may serve as ideal models for
transformation and cancer, as well.

26

Chapter 2

Identification of a genetic and chemical cocktail drives robust
conversion, enhances acquisition of target cell identity

27

Introduction
Cellular reprogramming through forced overexpression of transcription factor cocktails
redirects the transcriptional state of a cell, converting its fate into an increasing list of somatic
lineages (Xu et al., 2015). By providing access to rare, inaccessible somatic cell types in unique
genomic contexts, reprogramming massively expands the potential for in vitro disease
modeling. Investigations into neurologic diseases, previously limited by the supply of relevant
human cells and the weak fidelity of murine models, have advanced through cellular
reprogramming. Insights into the molecular changes that occur in disease contexts are already
generating novel therapeutic strategies (Ma et al., 2018; Shi et al., 2018; Wainger et al., 2014;
Wen et al., 2014; Zhao et al., 2015). Because direct lineage conversion can preserve epigenetic
signatures derived from the starting cells, this approach has the potential to lead to a better
understanding of enigmatic processes such as aging (Kim et al., 2018; Mertens et al., 2015;
Tang et al., 2017). In addition, several proof-of-concept studies have demonstrated the
therapeutic potential in vivo cellular reprogramming to regenerate damaged tissue (Guo et al.,
2014b; Kurita et al., 2018; Srivastava and DeWitt, 2016; Zhou et al., 2008).
However, epigenetic barriers limit reprogramming between somatic lineages to rare
events (Guo et al., 2014b; Lee et al., 2018; Son et al., 2011; Wapinski et al., 2013; Yoo et al.,
2011; Zhou et al., 2008; Zhou et al., 2016) and cause incomplete conversion of gene regulatory
networks in reprogrammed cells (Cahan et al., 2014). Efforts to identify epigenetic factors
limiting reprogramming have focused primarily on iPSC generation, and many of these findings,
such as interactions of epigenetic complexes with Oct4, Sox2, and Klf4, appear to be specific to
28

iPSC reprogramming (dos Santos et al., 2014; Mor et al., 2018; Papp and Plath, 2013; Rais et
al., 2013; Soufi et al., 2012). Indeed, inclusion of epigenetic modifiers such as valproic acid and
trichostatin A increase reprogramming to iPSCs, but has limited utility in other lineage
conversions (Huangfu et al., 2008a; Huangfu et al., 2008b; Ma et al., 2013). In addition,
modulating the Gatad2a-Mbd3/NuRD complex shows opposing effects on reprogramming in
different contexts (dos Santos et al., 2014; Mor et al., 2018; Rais et al., 2013). Moreover, factors
shown to reduce epigenetic blockades to reprogramming into somatic lineages are often
lineage-specific (Lee et al., 2018; Yoo et al., 2011; Zhou et al., 2016). Therefore, the
identification of more general mechanisms that overcome barriers preventing reprogramming
into somatic lineages could have significant translational utility.

Results
Transcription factor overexpression induces genomic stress
We focused on the motor neuron lineage because it is a well-defined neuronal subtype
with established markers and reporters. Utilizing mouse embryonic fibroblasts (MEFs) isolated
from Hb9::GFP transgenic mice, we generated induced motor neurons (iMNs) by viral
overexpression of six transcription factors (Ascl1, Brn2, Myt1l, Ngn2, Isl1, Lhx3) (6F) as
described previously (Son et al., 2011). We noticed that a surprisingly large number of iMNs
were binucleated (~10%, Fig. 2.1A). This suggested that cells were dividing during
reprogramming and had undergone incomplete cytokinesis. The possibility that cells divided
during reprogramming into a post-mitotic cell type such as neurons was surprising since
previous studies by ourselves and others did not detect large numbers of dividing cells during
induced neuron reprogramming (Son et al., 2011; Vierbuchen et al., 2010). However, careful
review of these studies suggested that the BrdU-labeling approach used in these studies might
29

not have been sensitive enough to detect all cell division events. Moreover, the number of
successful induced neuron reprogramming events was similar to the number of dividing cells in
these studies, indicating that in principle, most successful reprogramming events could have
arisen from cells that divided after transcription factor overexpression.
Failed cytokinesis can result from DNA replication stress and impaired DNA replication,
in which case chromatin bridges form between separating nuclei during mitosis (Keszthelyi et
al., 2016; Wang et al., 2010). In addition, micronuclei can emerge as the chromatin bridges
resolve and break the bridging DNA (Keszthelyi et al., 2016; Wang et al., 2010). To determine if
MEFs undergo genomic stress during reprogramming, we identified mitotic cells based on the
morphology of DAPI
+
nuclei as has been previously reported (Slattery et al, 2012). Anaphase-
telophase cells with one or more non-integrated DNA fragments were determined as having
micronuclei. Anaphase-telophase cells with one or more DNA strands between the
separating/separated daughter cells were determined as having a bridge. Consistent with
impaired DNA replication leading to failed cytokinesis, cultures transduced with the iMN factors
contained DAPI
+
micronuclei and chromatin bridges in about 30% of the mitotic anaphase-
telophase cells at two and four days post-infection (dpi), respectively (Fig. 2.1B-E). In contrast,
mitotic cells rarely contained chromatin bridges or micronuclei in MEFs transduced with a
retrovirus encoding a puromycin resistance gene (Control-Puro) or in MEFs not infected with the
iMN factors (Fig. 2.1B-E), These results indicate that cell division does occur during induced
motor neuron reprogramming, and that transcription factor overexpression induces DNA
replication stress and significantly elevates rates of chromatin bridges, micronuclei formation,
and binucleated neurons.

30

Fig 2.1. Transcription factor overexpression induces genomic stress. (A) Representative image of a binucleated iMN at 14 dpi.
Scale bar represents 10 µm. (B) Representative image of a mitotic cell with a micronucleus during reprogramming at 2 dpi. Arrow
points to micronucleus. Scale bar represents 5 µm. For all experiments in Figure 1, mitotic cells were identified based on
morphology of DAPI+ nuclei as previously described (Slattery et al, 2012). (C) Percentage of mitotic anaphase-telophase cells with
at least one micronucleus at 2 dpi for Control-Puro and 6F conditions. Anaphase-telophase cells with one or more non-integrated
DNA fragments were determined as having micronuclei. n = 150-175 cells from 3-6 independent conversions per condition.
Percentage +/- 95% confidence interval. Significance determined using a Chi-square test to compare the frequency in encountering
a mitotic cell with a micronucleus between conditions. (D) Representative image of a mitotic cell with a chromatin bridge during
reprogramming at 4 dpi. Arrow points to bridge. Scale bar represents 10 µm. (E) Percentage of mitotic anaphase-telophase cells
with a least one chromatin bridge at 4 dpi for Control-Puro and 6F conditions. Anaphase-telophase cells with one or more DNA
strands between the separating/separated daughter cells were determined as having a bridge. n = 63-100 cells from 3-6
independent conversions per condition. Percentage +/- 95% confidence interval per condition. Significance determined using a Chi-
square test to compare the frequency in encountering a mitotic cell with a chromatin bridge between conditions. Significance
summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01.

Identification of a genetic and chemical cocktail that massively increases reprogramming
To identify factors that can overcome epigenetic barriers preventing lineage conversion
into somatic cell types, we screened small molecule kinase inhibitors, epigenetic modifiers, and
oncogenes for the ability to increase the efficiency of MEF-to-iMN reprogramming. Suppression
of Mbd3/NuRD enables deterministic iPSC reprogramming (Rais et al, 2013; Mor et al, 2018).
Mor et al. showed that it is the Gatad2a-Mbd3/NuRD complex specifically, as opposed to other
complexes such as Gatad2b- or Mta2-Mbd3/NuRD that enables deterministic iPSC
reprogramming (Mor et al, 2018). In partial agreement with the iPSC studies, we found that
Mbd3 suppression increased the iMN reprogramming efficiency (Fig. 2.2A, B). However, in
A. C. B.
DAPI
DAPI Hb9::GFP
6F
6F
14 dpi 2 dpi
0
20
40
30
10
6F
*
(%) mitotic cells
with micronucleus
Control-Puro
2 dpi
E. D.
DAPI
6F
4 dpi
0
20
40
30
10
6F Control-Puro
**
(%) mitotic cells with
chromatin bridge
4 dpi
50
31

contrast to the iPSC studies in which Gatad2a suppression provided equal or better iPSC
reprogramming rates than Mbd3 knockdown (Mor et al, 2018), we found that Gatad2a
suppression did not significantly increase iMN reprogramming (Fig. 2.2C). These results
indicate that Mbd3 suppression facilitates both iPSC and iMN reprogramming and that there
may be shared mechanisms by which Mbd3 suppression enhances reprogramming in both
systems, such as transcription levels and gene activation (Mor et al, 2018). However, our
findings suggest that the Gatad2a-Mbd3/NuRD complex does not regulate iMN reprogramming
as strongly as it regulates iPSC reprogramming, highlighting mechanistic differences between
the two systems.

Fig. 2.2. Effect of Mbd3 and Gatad2a on iMN reprogramming. (A) Yield of iMNs for 6F conditions in presence of scrambled or
Mbd3 shRNAs at 14 dpi. n = 8-23 independent conversions per condition. Median +/- interquartile range. Kruskal-Wallis test. (B)
Yield of iMNs for 6F condition in presence of scrambled or titration of Mbd3-A shRNA at 14 dpi. n = 4-5 independent conversions
per condition. Mean +/- s.e.m. One-way ANOVA. (C) Yield of iMNs for 6F condition in presence of scrambled or Gatad2a shRNAs at
14 dpi. n = 9-16 independent conversions per condition. Median +/- interquartile range. Kruskal-Wallis test. Significance summary: p
> 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

In contrast to the modest gain in iMN reprogramming elicited by Mbd3 suppression, a
combination of RepSox, a TGFb inhibitor (Ichida et al., 2009), a Ras mutant (hRasG12V)
(Kotsantis et al., 2016), and p53DD (DD), a p53 mutant lacking a DNA-binding domain
(Bowman et al., 1996) (6FDDRR, Fig. 2.3A), increased the number of iMNs at the end of
reprogramming by 100-fold (Fig. 2.3B-D). Interestingly, the addition of DD, RepSox, and
hRasG12V to reprogramming cultures significantly reduced the percentage of mitotic cells with
micronuclei and chromatin bridges (Fig. 2.3E, F), as well as the number of binucleated iMNs
(Fig. 2.3G). These results show that there is a strong correlation between reducing genomic
stress during reprogramming and increased iMN formation.
32

Fig 2.3. Genetic and chemical factors relieve genomic stress and reprogramming block. (A) Legend of genetic and chemical
combinations used for three primary conditions employed in conversion. 6F, 6 transcription factors only. 6FDD, 6 transcription
factors and p53DD, a p53 mutant. 6FDDRR, 6 transcription factors and a combination of p53DD, hRasG12V, a mutant of hRas, and
RepSox, a TGF-β inhibitor. (B) Representative image of 6F-iMNs at 14 dpi. Scale bar represents 100 µm. (C) Representative image
of 6FDDRR-iMNs at 14 dpi. Scale bar represents 100 µm. (D) Yield of iMNs in 6F, 6FDD, or 6FDDRR conditions at 14 dpi.
Conversion yield determined by counting iMNs (Hb9::GFP+ cells with neuronal morphology) divided by the number of cells seeded
for conversion. n = 10-20 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (E) Percentage of mitotic
anaphase-telophase cells with a micronucleus at 2 dpi for 6F and 6FDDRR conditions. n = 100 cells from 3 independent
conversions per condition. Percentage +/- 95% confidence interval. Significance determined using a Chi-square test to compare the
frequency in encountering a mitotic cell with a micronucleus between conditions. (F) Percentage of mitotic anaphase-telophase cells
with a chromatin bridge at 4 dpi for 6F and 6FDDRR conditions. n = 100 cells from 3 independent conversions per condition.
Percentage +/- 95% confidence interval. Significance determined using a Chi-square test to compare the frequency in encountering
a mitotic cell with a chromatin bridge between conditions. (G) Percentage of binucleated iMNs for 6F and 6FDDRR conditions at 14
dpi. n = 6 independent conversions. Mean +/- s.e.m. Unpaired t-test. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01,
∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

DDRR drives acquisition of the induced motor neuron transcriptional program,
accelerates maturation
To determine if the genetic and chemical factors used to amplify the reprogramming
efficiency affect the resulting iMNs, we compared the molecular and functional properties of
33

iMNs generated in the 6F, 6FDD, and 6FDDRR conditions. First, we evaluated the
transcriptional state of Hb9::GFP+ cells from all three conditions by RNAseq to determine how
the populations vary globally. Hb9::GFP+ cells collected from all three conditions clustered
together compared to the starting cell population of MEFs (Fig. 2.4A). Cells from 6F and
6FDDRR conditions were relatively similar, although we observed small variations between the
two populations (Fig. 2.4A-C). The addition of DDRR led to the upregulation of slightly more
genes than those that were downregulated (Fig. 2.4B). Consistent with the increased efficiency
of reaching the terminal iMN state in 6FDDRR condition, Hb9::GFP+ cells showed
downregulation of genes associated with fibroblasts such as extracellular matrix organization
compared to 6F only (Fig. 2.4B, C). DDRR upregulated regulatory targets involved in neuron
projection development as well as apoptosis, cell cycle, and migration, reflecting processes
expected to be modulated by overexpression of hRasG12V (Fig. 2.4C). Taken together these
data indicate that DDRR accelerates the transcriptional shift away from the fibroblast state
without loss of fidelity to the induced motor neuron profile generated by 6F conditions.
Fig 2.4. DDRR accelerates shift towards neuronal transcriptional state. (A) Heat map of RNAseq collected from Hb9::GFP+
cells from different conditions on 17 dpi compared to starting MEFs across the 1186 genes that are differentially expressed between
MEFs and Hb9::GFP+ cells. n = 3 independent conversions per condition. (B) Volcano plot comparison of genes up- (blue) or
downregulated (red) in 6FDDRR versus 6F Hb9::GFP+ cells at 17 dpi. (C) List of gene ontology (GO) terms for genes upregulated
(top, blue) or downregulated (bottom, red) in 6FDDRR cells compared to 6F at 17 dpi.

To increase the resolution of comparison of iMN across protocols and to primary MNs,
we collected iMNs from 6F, 6FDDRR, and 6FDDRR+Top1 and primary embryonic motor
34

neurons at E12.5 for single-cell transcriptional profiling. We observed that each iMN condition
grouped into multiple clusters each with a larger Map2+ population and a smaller Col1a1+
cluster (Fig. 2.5A, B, Fig. 2.6A, B). Immunostaining confirmed that Hb9::GFP
+
cells possess
high MAP2 levels (Fig. 2.6C). By single cell RNAseq analysis, the majority of cells grouped into
the Map2+, neuronal population for each condition (Fig. 2.6D). Thus, most cells analyzed in the
single cell RNA sequencing experiment were Hb9::GFP
+
neurons.
Fig. 2.5. iMN populations partition into separate clusters, exhibit distinct patterns of gene expression (A) tSNE plot with
clusters of iMNs identified by neuronal gene signature and other Hb9::GFP+ cells captured at 14 dpi in reprogramming.
“6F,””6FDDRR,” and “Top1 + 6FDDRR” clusters comprise cells with non-neuronal gene expression profiles. (B) Gene expression for
clusters in (A) separate neuronal iMN clusters with high Map2 (middle) expression from non-neuronal clusters with high Col1a1 (top)
and lower Isl1 (bottom) expression.

We observed small differences in gene expression between neuronal iMN clusters.
Neuronal cluster markers were limited to 20-30 genes per condition that could be identified as
significantly differentially expressed by 1.5-fold between the conditions. In examining marker
genes, we found that differences in neurosignaling pathways characterized different clusters
and conditions (Fig. 2.6E). Gene ontology analysis of markers associated with neuronal clusters
35

identified differences in neuronal properties including neurosignaling and cell cycle across the
iMN clusters (Fig. 2.6F).
Fig 2.6. DDRR enhances adoption of the induced motor neuron transcriptional program (A) tSNE projection of Hb9::GFP+
embryonic motor neurons (embMNs) collected at 12.5 dpi and iMNs generated by three different cocktails (6F, 6FDDRR, and
6FDDRR+Top1) colored by individual condition. embMNs were bioinformatically identified by Isl1 expression to distinguish from
other Hb9::GFP+ populations. (B) Relative expression colored by intensity of Col1a1, Isl1, Map2, and Chat over the populations in
the tSNE in (A). (C) Representative images of Hb9::GFP+ 6F- or 6FDDRR-iMNs immunostained for MAP2 at 17 dpi. Scale bars
represent 5 µm. (D) Percentage of the Hb9::GFP+ cell population with neuronal gene expression profile for 6F, 6FDDRR, and
6FDDRR+Top1 conditions at 17 dpi. (E) Relative expression of neurosignaling genes (i.e.. Scg2, Chgb, Sncg, Snca) colored by
intensity over the populations in the tSNE in (A). (F) List of gene ontology (GO) terms for marker genes upregulated in iMN clusters
capture differences between different iMN populations as well as embMNs. Significance summary: ∗∗∗p ≤ 0.001.

Accelerating maturation of lineage-converted cells remains one of the preeminent
challenges limiting translational studies (Corti et al., 2015; Marchetto et al., 2010). To determine
if addition of DDRR accelerates maturation of the resulting cells, we examined the
morphological and electrophysiological properties of iMNs. In vivo, neurons adopt different
morphologies with varying polarity (e.g. unipolar, bipolar, multipolar). These morphologies are
unique to their function and developmental window and impact signal processing (Takazawa et
al., 2012; Vrieseling and Arber, 2006). Mature spinal motor neurons are multipolar (Takazawa et
al., 2012; Vrieseling and Arber, 2006). We found that inclusion of DD significantly increased the
percentage of multipolar neurons generated (Fig. 2.7A). These results suggest that inclusion of
36

DD increases or accelerates the maturation of motor neurons generated by direct lineage
conversion.
The chief function of mature motor neurons involves receiving and transmitting
electrophysiological signals. Given that neuronal morphology influences electrophysiological
behaviors (Takazawa et al., 2012; Vrieseling and Arber, 2006), we next examined how DD
impacted electrophysiological function of iMNs. We examined the ability of iMNs to adapt to
repetitive stimulation. Upon repetitive stimulation, mature neurons display spike-frequency
adaptation (SFA), increasing the time interval between spikes and SFA ratio (Takazawa et al.,
2012). Unlike 6F iMNs, 6FDD iMNs displayed spike frequency adaptation, with a spike
frequency adaptation ratio several-fold higher than that of 6F iMNs (Fig. 2.7B-D). The SFA ratio
from 6FDD iMNs measured 5-fold higher than that reported for iPSC-derived iMNs with
prolonged culture maturation (Takazawa et al., 2012). Taken together, the morphological and
electrophysiological data indicate that addition of DDRR during reprogramming results in the
production of iMNs that possess greater functional maturity.

37

Fig 2.7. Addition of DD accelerates morphological maturation. (A) Percentage of multipolar iMNs derived from MEFs
in 6F and 6FDD conditions at 14 dpi. n = 6-7 independent conversions per condition. Mean +/- s.e.m. Unpaired t-test. (B) Spike
frequency adaptation (SFA) ratio evoked APs of mouse iMNs in 6F and 6FDD conditions at 14 dpi. n = 7-8 cells from 3 independent
conversions per condition. Median +/- interquartile range. Mann-Whitney test. (C-D) Representative action potentials evoked in
mouse iMNs by a positive current injection (indicated by solid bar across bottom) illustrating SFA over the course of the stimulus of
iMNs in 6FDD (M) and 6F (L) conditions at 14 dpi. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001

Chemical and genetic factors enable efficient reprogramming across cell types and
species
Given the robust increase in iMN reprogramming upon inclusion of DD and DDRR, we
sought to determine the generality of this phenomenon in other reprogramming schemes. To do
so, we employed different protocols to generate an array of post-mitotic cell types. Inclusion of
DD or DDRR increased reprogramming of MEFs into induced neurons (iNs) via Ascl1, Brn2,
Mytl1L (Vierbuchen et al., 2010), induced dopaminergic neurons (iDANs) via Ascl1, Brn2,
Mytl1L, Lmx1A, and FoxA2 (Pfisterer et al., 2011), and induced hair cells (iHCs) via Atoh1,
Gata3, and Brn3C (Costa et al., 2015) (Fig. 2.8A-D).

38

Fig 2.8. The DDRR cocktail boosts reprogramming across multiple mouse paradigms. (A) Yield of induced neurons (iNs) for
different conditions including control with 3 factors only (e.g. Brn2, Ascl1, Myt1l), 3FDD, and 3FDDRR counted by MAP2+ cells at 17
dpi over number of cells seeded. Conversion yield quantified as previously described. n = 6-7 independent conversions per
condition. Mean +/- s.e.m. One-way ANOVA. (B) Yield of induced dopaminergic neurons (iDANs) for different conditions including
control with 5 factors only (e.g. Brn2, Ascl1, Myt1l, Lmx1A, FoxA2), 5FDD, and 5FDDRR counted by MAP2+ cells at 17 dpi.
Conversion yield quantified as described previously. n = 6-8 independent conversions per condition. Mean +/- s.e.m. One-way
ANOVA. (C) Yield of induced inner ear hair cells (iHCs) for different conditions including control with 3 factors only (e.g. Brn3C,
Atoh1, Gfi1), 3FDD, and 3FDDRR counted by Atoh1::nGFP+ cells at 17 dpi. Conversion yield quantified as described previously. n
= 3-16 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (D) Representative images of 3F-iHCs and
3FDD-iHCs immunostained with Myosin VIIa at 17 dpi. Scale bar represents 100 µm. Significance summary: p > 0.05 (ns), ∗p ≤
0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Additionally, the reprogramming increase extended across species and age in the
starting cell populations to include mouse adult tail tip fibroblasts and myoblasts into iMNs (Fig.
2.9A, B) and conversion of human adult fibroblasts into iMNs (Son et al., 2011) (Fig. 2.9C). For
human iMNs, we utilized a seven-factor human iMN reprogramming cocktail which includes
addition of NEUROD1 to the 6F cocktail (7F) (Son et al, 2011). These results indicate that
addition of DDRR promotes the reprogramming of somatic cells from different ages and species
into post-mitotic lineages.

39

Fig 2.9. The DDRR cocktail enhances reprogramming across starting cell types and species. (A) Yield of iMNs generated
from adult tail tip fibroblasts with factors alone (6F) compared to yield in presence of p53DD (6FDD) (both conditions with RepSox)
and 6FDDRR at 28 dpi. Conversion yield quantified as described previously. n = 4-9 independent conversions per condition. Mean
+/- s.e.m. One-way ANOVA. (B) Yield of iMNs generated from Hb9::GFP+ adult mouse muscle explants with factors alone (6F)
compared to yield in the presence of p53DD (6FDD) at 28 dpi. Conversion yield quantified as described previously. n = 4-5
independent conversions per condition. Mean +/- s.e.m. Unpaired t-test. (C) Yield of iMNs generated from human fibroblasts with
factors alone (7F) compared to yield in presence of p53DD (7FDD) (both conditions with RepSox) counted by MAP2+ cells at 35
dpi. Conversion yield quantified as described previously. n = 4-6 independent conversions per condition. Median +/- interquartile
range. Mann-Whitney test. (D) Percentage of multipolar iMNs derived from primary human fibroblasts in 7F and 7FDD conditions at
35 dpi. n = 3 independent conversions per condition. Mean +/- s.e.m. Unpaired t-test. Significance summary: p > 0.05 (ns), ∗p ≤
0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

To examine the human fibroblast-derived iMNs, we evaluated morphology and
electrophysiology with and without inclusion of DD. As for iMNs derived from human fibroblasts,
we found that inclusion of DD significantly increased the percentage of multipolar neurons
generated (Fig. 2.9D). Using patch clamp electrophysiology, we measured sodium and
potassium currents in human iMNs generated with and without DD. 7FDD human iMNs
displayed significantly faster sodium and potassium currents compared to 7F iMNs (Fig. 2.10A,
40

B). Similarly, 7FDD iMNs displayed tighter, more mature action potentials (Fig. 2.10C, D). Both
metrics indicate more mature ion channel organization across the surface of 7FDD iMNs. Taken
together we demonstrate that DD and DDRR increase reprogramming rate and maturity across
protocols and starting cells types.

Fig 2.10. DD enhances maturity of human-derived iMNs (A, B) Representative step voltage depolarizations results in functional
sodium channels (fast inward currents) as well as functional potassium channels (slower outward currents) in human 7F (A) or
7FDD-iMNs (B) at 35 dpi, confirming functional ion channels. (C, D) Representative action potentials evoked by a step current
injection in current-clamp configuration for human 7F (C) or 7FDD-iMNs at 35 dpi (D).

Discussion
Transdifferentiation has enabled access to previously inaccessible cell types for study of
development and disease. Neurodegenerative diseases are characterized by loss of function of
one or more neuronal cell types inaccessible in patients, making transdifferentiation an
attractive tool with which to generate and study these complex disorders. However,
reprogramming remains a rare, inefficient event and exploration of roadblocks have been
limited. Through screening of small molecules and genetic perturbations to enhance fibroblast-
41

to-iMN reprogramming, we find addition of a p53 mutant, p53DD, a mutant of hRas, hRasG12V,
and a small molecule TGF-β inhibitor, RepSox, increases iMN conversion nearly 100-fold. We
find that addition of this cocktail significantly enhances conversion across multiple starting cell
types, ages, lineages, and species, as well. This suggests that both p53 and/or hRas are
generally shared roadblocks between direct reprogramming systems.
Attaining a mature somatic cell state remains a major difficulty limiting the translational
utility of reprogrammed or stem cell-derived cells (Marchetto et al., 2010; Vierbuchen and
Wernig, 2012). We show that this combination of genetic perturbations and small molecules
reduces genomic stress and heterogeneity of the resulting cells. Additionally, DDRR accelerates
and enables cells to achieve greater functional maturity in the reprogrammed state. We
hypothesize that enhanced maturity results from a greater ability of converting cells to remodel
their GRN and proteome due to better dilution of the starting cell components, allowing more
complete transition to the target state. Our study suggests that barriers to robust direct
reprogramming are overcome with addition of our perturbations, allowing more efficient
conversion and acquisition of the iMN identity.
Previous studies in the iPSC field uncovered p53 plays inhibitory roles by promoting cell
cycle exit, cellular senescence, and apoptosis. Additionally, other studies have suggested that
epigenetic roadblocks like the Mbd3-Gatad2a/NuRD complex can be modulated to increase
iPSC conversion at near-deterministic rates. However, while we observe similar increased
conversion upon p53 inhibition, we did not observe similar phenomena with the Mbd3-
Gatad2a/NuRD complex in this direct reprogramming system, suggesting that mechanisms
preventing and regulating conversion are both shared and distinct between iPSC and direct
conversion. Building on established roles for p53 and hRas in biology and the iPSC field,
exploration of the mechanisms underlying the DDRR affect will provide deeper understanding of
molecular roadblocks to robust direct reprogramming.
42

Chapter 3

Genetic and chemical cocktail enhances conversion by mitigating
conflicts between transcription and proliferation

43

Introduction
Transdifferentiation offers unprecedented access to inaccessible cell types for disease
modeling and studies of development. However, utilization of this technology has been limited
due to poor reprogramming efficiency and heterogeneity of converted cultures. A cocktail of
small molecules and genetic perturbations significantly enhances conversion to induced motor
neurons and other lineages. We sought to identify and understand the mechanisms underlying
the universal roadblocks to reprogramming that are overcome with this cocktail.
To this end, we examined systems-level constraints limiting the conversion of fibroblasts
into motor neurons, as well as other paradigms. We find that during lineage conversion, addition
of the reprogramming factors sharply increases the rate of transcription in cells and reduces the
rate of DNA synthesis and cell division, highlighting the existence of tradeoffs between
transcription and cell replication during the conversion process. Most cells display either a high
rate of transcription and limited proliferation or a high rate of proliferation and limited
transcription, with both cell states being largely refractory to reprogramming. However, we
identify a privileged population of cells capable of both high proliferation and high transcription
rates that contribute to the majority of reprogramming events. This indicates that a high rate of
proliferation is not sufficient for efficient reprogramming, and that it must be coupled with high
rates of transcription. Using our defined cocktail of genetic and chemical factors, we expand the
hypertranscribing, hyperproliferating cell (HHC) population and achieve induced motor neuron
reprogramming at near-deterministic rates.
Transcription and DNA synthesis interfere directly through collisions of transcription and
replication machinery, as well as indirectly by generating inhibitory DNA structures and
topologies (e.g. R-loops, supercoiling). Through transcriptional profiling, we identify
topoisomerases, enzymes that curate DNA supercoiling introduced during these processes of
transcription and DNA replication, as key regulators supporting the emergence and expansion
44

of these privileged HHCs. Our results indicate that relieving biophysical constraints governing
transcription and replication overcomes the epigenetic barriers to reprogramming.

Results
Hypertranscription and hyperproliferation drive neuronal reprogramming
Transcription and DNA replication antagonize each other by increasing torsional strain
and steric interference on genomic DNA (Keszthelyi et al., 2016; Kotsantis et al., 2016; Merrikh
et al., 2011; Tuduri et al., 2009). Based on our observation that iMN transcription factor
overexpression induced high rates of chromatin bridges and micronucleus formation, we
hypothesized that interference between DNA replication and elevated transcriptional activity
may prevent efficient transcription factor-mediated reprogramming. To test this hypothesis, we
first determined if transcription factor overexpression caused an overall increase in the rate of
transcription in MEFs by measuring 5-ethynyl uridine (EU) incorporation (Fig. 3.1A). Indeed,
MEFs transduced with the iMN factors showed a significant increase in mean EU intensity over
non-transduced MEFs as measured by FACS at 2 dpi (Fig. 3.1B). We defined the relative
transcription rate as the mean EU incorporation of 6F-transduced MEFs relative to the mean
non-transduced MEFs on the same day. Relative transcription rate of 6F-transduced MEFs
increased 50% from 1 to 2 dpi (Fig. 3.1C). Nuclear EU signal in 6F MEFs increased both within
and outside of nucleoli by 2 dpi compared to 1 dpi (Fig. 3.1D), suggesting that both RNA
polymerase I- and RNA polymerase II-dependent transcription is elevated. Because this
increased EU incorporation reflected an increase in the overall rate of transcription in MEFs, we
termed this a state of “hypertranscription.”
45

Figure 3.1. Reprogramming factors induce a state of hypertranscription. (A) Representative image of EU-click labeling and
nucleolin immunostaining in 6F MEFs at 1 and 2 dpi. Scale bar represents 10 µm. (B) FACS plot showing the relative EU
incorporation in viable cells from 6F-infected MEF cultures at 1 and 2 dpi. Cell viability was determined based on the FSC and SSC
profiles in FACS analysis. (C) Relative transcription rate measured by EU incorporation via flow cytometry at 1 and 2 dpi in 6F-
infected MEFs compared to uninfected control. The mean EU intensity of non-transduced MEFs was defined as a relative
transcription rate of 1. Only viable cells, determined based on their FSC and SSC profile via FACS, were included in the analysis. n
= 5 independent transductions per condition. Mean +/- s.e.m. Unpaired t-test (D) Mean EU intensity within or excluding nucleoli in
6F MEFs at 1 and 2 dpi. n = 56 cells (nucleolar EU) or 57 cells (non-nucleolar EU) from 3 independent conversions per condition.
Median +/- interquartile range. Mann-Whitney test between 1 and 2 dpi samples within each nuclear compartment. Significance
summary: ∗∗∗∗p ≤ 0.0001.

To determine if hypertranscription and cell proliferation act antagonistically during
reprogramming, we evaluated the impact of transcription factor overexpression on cell
proliferation. We measured cell proliferation by labeling fibroblasts with the stable dye CFSE 24
hours after transduction and flow sorting 72 hours later (Fig. 3.2A). Dilution of the CFSE dye
occurs via cell division, resulting in dim, fast-cycling cells and brightly-stained, slowly-dividing
cells (Fig 3.2A). In accordance with published studies, we defined “hyperproliferation” or “fast-
cycling” cells as those showing a two-fold increase in division rate (i.e. an eight-fold decrease in
CFSE intensity) compared to the average of the control population, which was comprised of
Control-Puro MEFs (Guo et al., 2014a). As we had hypothesized, retroviral overexpression of
the iMN reprogramming factors reduced the percentage of hyperproliferating cells ten-fold
46

compared to Control-Puro MEFs (Fig. 3.2B). At 4 dpi, staining for Ki67, a marker of proliferative
cells, confirmed that 6F reduced the proliferative population compared to control-infected cells
and DDRR restored the population (Fig. 3.2C). Transducing with DsRed retrovirus did not
reduce proliferation, suggesting that this effect was dependent on the transgenic factors being
transcription factors (Fig. 3.2D). We observed a similar trend with subsets of reprogramming
transcription factors. Transduction with Ascl1 alone or Brn2, Ascl1, and Myt1l (BAM)
significantly reduced the percentage of hyperproliferative cells relative to Control-Puro cells (Fig.
3.2E, F). These results indicate that transcription factor overexpression reduces cell proliferation
during reprogramming.
Figure 3.2. Hyperproliferating cells convert to iMNs at significantly higher rates than non-hyperproliferating cells. (A)
Schematic of CSFE-based flow sorting and re-plating of populations for reprogramming quantification assays. (B) Representative
histograms of CFSE intensity for Control-Puro and 6F-infected cells measured by flow cytometry at 4 dpi. “HyperP” stands for
hyperproliferating cells. Hyperproliferating cells were defined as cells showing a two-fold increase in division rate (i.e. an eight-fold
decrease in CFSE intensity) compared to the average of the control population, which was comprised of Control-Puro MEFs. AU
defined as arbitrary units. (C) Percentage of Ki67+ cells in Control-Puro, 6F, or 6FDDRR conditions measured via flow cytometry at
4 dpi. n = 3-4 independent transductions per condition. Mean +/- s.e.m. One-way ANOVA. (D) Representative histogram of CFSE
intensity measured by flow cytometry for MEFs infected with DsRed or 6F at 4 dpi. (E) Representative histogram of CFSE intensity
measured by flow cytometry for Control-Puro-infected cells, Ascl1-infected cells, or Brn2+Ascl1+Myt1l-infected cells (BAM) at 4 dpi.
“HyperP” stands for hyperproliferating cells. (F) Effect of addition of Ascl1 or neuronal reprogramming factors (BAM (e.g. Brn2,
Ascl1, Myt1l)) on the percentage of hyperproliferating cells measured by flow cytometry at 4 dpi. n = 3-6 independent transductions
per condition. Mean +/- s.e.m. One-way ANOVA. (G) Representative histograms of CFSE intensity for Control-Puro, 6F, and
6FDDRR conditions by flow cytometry at 4 dpi with gates showing CFSE-Low (e.g. hyperproliferating cells (HyperP)) and CFSE-
High (e.g. slow cycling cells). (H) Yield of iMNs from 6F, 6FDD, or 6FDDRR reprogramming populations sorted by CFSE-intensity
(i.e. CFSE-Low and CFSE-High) at 4 dpi. Percent yield determined by counting total iMNs generated normalized by total number of
cells counted per population at 4 dpi. n = 4-23 independent conversions per condition. For 6F and 6FDD: Median +/- interquartile
range and Mann-Whitney test between the CFSE High and Low groups in each transduction condition. For 6FDDRR: Mean +/-
s.e.m. Unpaired t-test between CFSE High and Low groups. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤
0.001, and ∗∗∗∗p ≤ 0.0001.

47

To determine if hypertranscription and hyperproliferation are key drivers of
reprogramming, we first asked if hyperproliferating cells convert into motor neurons more
efficiently than slow-cycling cells. Consistent with the idea that hyperproliferating cells
reprogrammed more efficiently, adding DD, RepSox, and hRasG12V to iMN reprogramming
cultures greatly increased the number of hyperproliferating cells (Fig. 3.2G). To definitively
determine if hyperproliferative cells convert into iMNs at higher rates than less proliferative cells,
we labeled reprogramming cultures with CFSE at 1 dpi, prospectively isolated hyperproliferative
and non-hyperproliferative cells from converting cultures by flow cytometry 72 hours after CFSE
labeling (at 4 dpi), re-plated them, and measured their ability to form iMNs between 14-17 dpi
(Fig. 3.2A). In the 6F, 6F + DD (6FDD) and 6FDDRR conditions, hyperproliferative cells (e.g.
cells that hyperproliferated from 1-4 dpi) that were prospectively-isolated and re-plated went on
to form iMNs at significantly higher rates than slow-cycling cells (Fig. 3.2H). Additionally, we
observed that the percentage of hyperproliferating cells and iMN yield from MEFs dropped with
passaging (Fig. 3.3A, B). Moreover, reducing cell division by treating reprogramming cultures
with mitomycin C or overexpressing p21 sharply reduced iMN conversion (Fig. 3.3C, D).
Mitomycin C inhibited iMN formation only when treatment started at early time points post-
transduction, indicating that only cell division early in the conversion process promotes
reprogramming (Fig. 3.3C). Importantly, hyperproliferative cells only reprogrammed with
substantially greater efficiently in the 6FDD and 6FDDRR conditions, suggesting that DD,
RepSox, and hRasG12V provided hyperproliferative cells with additional properties that enabled
efficient reprogramming that the 6F condition did not (Fig. 3.2H). Cleaved caspase-3 staining at
2, 4, and 8 dpi in 6F and 6FDDRR conditions ruled out the possibility that DDRR simply reduced
rates of apoptosis during reprogramming (Fig. 3.3E). These results demonstrate that cells that
hyperproliferate early in reprogramming convert into motor neurons with high efficiency, but this
ability is dependent on an additional property provided by DD, RepSox, and hRasG12V.

48

Figure 3.3. Early proliferation drives iMN reprogramming. (A) Percentage of hyperproliferating cells for MEFs passage 1-6
measured via CFSE using flow cytometry. Hyperproliferating cells were defined as cells showing a two-fold increase in division rate
(i.e. an eight-fold decrease in CFSE intensity) compared to the average of the control population, which was comprised of passage
1 MEFs. n = 3-5 biological replicates per condition. Mean +/- s.e.m. One-way ANOVA. (B) Effect of MEF passage on iMN yield for
6F condition. n = 3- 4 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (C) Schematic of timeline for
mitomycin C treatment and effect on conversion yield for the 6FDD condition at 14 dpi. n = 3 independent conversions per condition.
Mean +/- s.e.m. One-way ANOVA. (D) Conversion yield in 6F and 6FDD conditions in absence or presence of p21 overexpression
at 14 dpi. n = 5-7 independent conversions per condition. Mean +/- s.e.m. Unpaired t-test of dsRed vs p21 within each
reprogramming cocktail. (E) Percentage of cleaved-caspase-3+ cells at 2, 4, and 8 dpi in Control-Puro, 6F, or 6FDDRR conditions
measured via flow cytometry. n = 3-5 independent transductions per condition. Mean +/- s.e.m. One-way ANOVA between
conditions at each day. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

To determine if DDRR enables 6F-infected hyperproliferative cells to reprogram
efficiently by providing them the ability to maintain hypertranscription, we added an EU-labeling
step at 4 dpi of our quantitative CFSE/flow-cytometry assay (Fig. 3.4A) and measured both
cellular proliferation and transcription rates during reprogramming in the presence or absence of
DD or DDRR. We set our gating by using the 6F condition and defined hypertranscribing cells
49

as cells in the top half of EU intensity within the hyperproliferating population (Fig. 3.4B).
Consistent with the notion that hyperproliferation and hypertranscription antagonize each other,
hyperproliferating cells displayed significantly reduced transcription levels when transduced with
the six iMN factors (Fig. 3.4B, C). These data suggest a model in which cells are limited in total
nucleic acid synthesis rates (i.e. transcription and replication rates). Controlling for differences
based on condition, we would expect that inhibition of DNA synthesis should increase RNA
synthesis. An 18-hour aphidicolin treatment dramatically reduced the percent cells in S-phase
measured by EdU incorporation at 4 dpi compared to DMSO-treated control (Fig. 3.5A, B).
Fig 3.4, DDRR expands rare hypertranscribing, hyperproliferating (HHC) population (A) Schematic of CFSE-EU assay for
measuring transcription and proliferation rates in converting cells via flow cytometry at 4 dpi. Transcription rates measured by 5-
ethynyl uridine (EU) incorporation during 1 hr incubation with 1mM EU followed by “click” reaction with fluorescent dye to visualize
EU incorporation. CFSE assay performed as described previously. (B) Representative dot plot of CFSE intensity and fluorescently
labeled-EU for Control-Puro (grey), 6F (green), and 6FDDRR (red). Histograms of CFSE and EU intensity adjacent to dot plot.
Quadrant to demark hypertranscribing, hyperproliferating cells (HHCs) set by reference to 6F condition. Hyperproliferating and slow
cycling cells set by selecting CFSE value in 6F condition to allow the dimmest 15%. High EU values set by top half of 6F condition,
resulting in ~7% HHCs in 6F. (C) Relative transcription rate measured by EU incorporation via flow cytometry at 4 dpi of the whole
population (All Cells) compared to hyperproliferative cells measured in 6F-infected MEFs. n = 10 independent transductions per
condition. Mean +/- s.e.m. One-way ANOVA. (E) Percentage of HHCs for Control-Puro, 6F, and 6FDDRR conditions as assayed. n
= 11-16 independent conversion per condition. Median +/- interquartile range. Kruskal-Wallis Test. Significance summary: p > 0.05
(ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.
50

As expected, aphidicolin treatment resulted in a reduction (~40%) in total cell count in the
6FDDRR condition (Fig. 3.5C). However, the fold-decrease of in cell count was smaller than the
fold-reduction in the percentage of cells in S-phase (Fig. 3.5A, B) indicating that aphidicolin
treatment specifically inhibited DNA synthesis (Fig. 3.5A-C). Consistent with our model of
antagonism between DNA synthesis and transcription, aphidicolin treatment significantly
increased the rate of RNA synthesis for Control-Puro and 6FDDRR cells at 4 dpi (Fig. 3.5D).
Consistent with our hypothesis that DDRR promotes reprogramming by supporting
higher transcription rates in hyperproliferative cells, DDRR increased the transcription rate of
6F-infected hyperproliferative cells, resulting in a larger population of hypertranscribing,
hyperproliferating cells, or HHCs (Fig. 3.4B, D, hypertranscribing and hyperproliferating cells
defined as above in Fig. 3.2B and 3.4B, respectively). This is consistent with known effects of
these factors, as hRasG12V was previously shown to globally increase transcription human
mammary epithelial cells (Kotsantis et al., 2016) and TGFb signaling has been shown to
suppress transcription levels (Elkon et al., 2015).
To determine if the increased transcription rates in hyperproliferative cells may explain
the increased conversion efficiency in the presence of DDRR, we measured iMN
reprogramming with or without the RNA polymerase inhibitor α-amanitin. At 4 dpi, treatment with
1 µM α-amanitin for 18 hrs did not reduce viable cell count or viability as measured by FSC and
SSC gating via FACS (Fig. 3.5E). Consistent with the notion that a high rate of transcription was
required for the DDRR effect, using α-amanitin treatment at 4 dpi to reduce the average
transcription rate in 6FDDRR cells by 20% as determined by the EU-click FACS assay
significantly reduced reprogramming (Fig. 3.5F, G). Importantly, α-amanitin treatment did not
reduce viable cell count under these conditions, indicating that cell toxicity was not responsible
for the decrease in reprogramming (Fig. 3.5E G). Conversely, overexpressing TATA-binding
protein (TBP), which increases transcription in fibroblasts (Kotsandis et al, 2016), significantly
51

increased iMN reprogramming in the 6F condition (Fig. 3.5H). This result indicates that proteins
that act as transcriptional drivers can increase iMN reprogramming. Thus, there is a strong
correlation between hypertranscription, hyperproliferation, and the rate of iMN conversion, and
DDRR increases the number of hypertranscribing, hyperproliferating cells after 6F infection.

Figure 3.5. Sustained transcription in hyperproliferative cells drives conversion. (A) Scatter plot showing effect of 18-hour 1
µM aphidicolin treatment on cells as measured by EdU incorporation and DAPI via flow cytometry at 4 dpi. (B) Percentage of cells in
S-phase in 6FDDRR condition with DMSO or Aphidicolin treatment for 18 hrs and measured with EdU incorporation via flow
cytometry at 4 dpi. n = 4 independent transductions per condition. Mean +/- s.e.m. Unpaired t-test. (C) Effect of aphidicolin treatment
on total number of viable cells in 6FDDRR condition at 4 dpi. Viable cells were defined based on their FSC and SSC profile via
FACS. n = 5-6 independent transductions per condition. Mean +/- s.e.m. Unpaired t-test. (D) Percent relative transcription rate
increase upon inhibition of DNA synthesis with aphidicolin treatment at 4 dpi in Control-Puro and 6FDDRR. Relative transcription
rate determined by difference between rates with and without aphidicolin treatment normalized to without for each transduction
condition. n = 3 independent transductions per condition. Mean +/- s.e.m. Unpaired t-test between with and without aphidicolin
treatment for each transduction condition. (E) Percent viable cells in 6FDDRR condition at 4 dpi following 18-hrs treatment with
water or α-Amanitin. Viable cells were defined based on their forward scatter (FSC) and side scatter (SSC) profile via FACS. n = 6
independent transductions per condition. Mean +/- s.e.m. Unpaired t-test. (F) Relative transcription rate following 18-hour α-
Amanitin treatment in 6FDDRR conditions as measured by EU incorporation via flow cytometry at 4 dpi. The mean EU intensity of
6FDDRR cells treated with water was defined as a relative transcription rate of 1. n = 6 independent transductions per condition.
Mean +/- s.e.m. Unpaired t-test. (G) Effect of α-Amanitin treatment on the yield of iMNs in 6FDDRR condition at 14 dpi. n = 8
independent conversions per condition. Mean +/- s.e.m. Unpaired t-test. (H) Yield of iMNs in 6F condition with TBP overexpression
at 14 dpi. n = 8-21 independent conversions per condition. Mean +/- s.e.m. Unpaired t-test. Significance summary: p > 0.05 (ns), ∗p
≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Given the high density of Hb9::GFP+ cells in 6FDDRR conditions (Fig. 2.2C, D), we
reasoned that our previous image-based estimates of iMN reprogramming represented an
underestimate of yield. To improve our quantification of reprogramming efficiency in the
6FDDRR condition, we generated 6F or 6FDDRR iMNs in the absence of primary glia and
exhaustively quantified cell number by flow cytometry. Quantifying the cell population for
Hb9::GFP+ cells at the end of conversion, we analyzed total yield of Hb9::GFP+ cells based on
52

starting number of cells and the Hb9::GFP+ fraction of the whole culture at 17 dpi. While the 6F
alone condition resulted in fewer than 10 iMNs per 100 MEFs plated, 6FDDRR yielded about
300 iMNs per 100 MEFs plated (Fig. 3.6A), a 30-fold increase in yield.
To take into account the fact that MEFs proliferate beyond the initial amount plated and
to determine the true efficiency of iMN conversion, we also quantified the percent of Hb9::GFP+
cells out of the total number of cells at the end of the conversion process. In the absence of
DDRR, 90% of cells failed to activate Hb9::GFP. With DDRR, 30% of the population activated
Hb9::GFP (Fig. 3.6B). Taking into account that HHCs represent approximately 20-30% of the
whole population in the 6FDDRR condition (Fig. 3.6B) and comprise the majority of
reprogrammable cells (see Fig. 3.7A-C below), 30% of MEFs activating Hb9::GFP represents
near-deterministic reprogramming of the HHC population. Together, these data suggest that
DDRR boosts the reprogrammable population by increasing the number of cells capable of
exhibiting maintaining hypertranscription and hyperproliferation early in the conversion process.

Figure 3.6. DDRR increases reprogrammable population. (A) Yield of Hb9::GFP+ cells for 6F and 6FDDRR conditions counted
via flow cytometry at 17 dpi and normalized to number of seeded cells. n = 7-8 independent conversions per condition. Mean +/-
s.e.m. Unpaired t-test. (B) Yield of Hb9::GFP+ cells for 6F and 6FDDRR conditions normalized to the total cell number at 17 dpi.
Cells were quantified via flow cytometry at 17 dpi. n = 7-8 independent conversions per condition. Mean +/- s.e.m. Unpaired t-test.
Significance summary: ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Next, we sought to definitively determine if HHCs possessed privileged reprogramming
relative to hyperproliferative, but non-hypertranscribing cells. To test if HHCs identified at 4 dpi
possess greater reprogramming potential relative to hyperproliferative cells with lower
53

transcription rates, we performed a reprogramming experiment with prospective labeling of
HHCs and non-HHCs (Fig. 3.7A). First, we CFSE labeled Hb9::GFP MEFs at 1 dpi and pulse-
labeled cells with EU for 4 hours at 4 dpi, and then used flow cytometry to isolate all CFSE-low,
hyperproliferative cells after EU-labeling. We again defined CFSE-low, hyperproliferative cells
as cells showing a two-fold increase in division rate (i.e. an 8-fold decrease in CFSE intensity)
compared to the average of the control population, which was comprised of 6F MEFs (Fig.
3.2B). Given that we isolate the CFSE-low population and CFSE dims 10-fold from 4 to 8 dpi
(Fig. 3.7B), we could detect Hb9::GFP signal over background CFSE. We then re-plated these
cells for another 4 days, until 8 dpi, when Hb9::GFP reporter activity is observable (Fig. 3.7A).
At 8 dpi, we fixed and “clicked” EU-labeled cells and used FACS to measure EU incorporation
that had occurred in the hyperproliferative cells at day 4 (Fig. 3.7A). From 4 to 8 dpi, EU signal
remains stable dropping by only ~10% (Fig. 3.7C). Using FACS analysis, we identified HHCs by
high EU levels (top quartile of EU intensity evaluated in hyperproliferative cells), and analyzed
them for Hb9::GFP expression (Fig. 3.7A). We used cells with an EU intensity in the top quartile
of hyperproliferative cells instead of the top half to most stringently examine hypertranscribing
cells and avoid confounding loss due to dilution of EU.
Of the HHC population, over 40% expressed Hb9::GFP at 8 dpi, while only 13% of the
non-HHC population expressed Hb9::GFP (Fig. 3.7E). Thus, HHCs were 3 times more likely to
activate Hb9::GFP relative to hyperproliferative, but non-hypertranscribing cells. Because our
combined imaging/qRT-PCR analyses showed that the GFP intensity of Hb9::GFP+ cells was
strongly correlated with neuronal morphology and gene expression (Fig. 3.7D, Fig. 3.8A, 3.9A),
we examined the EU intensity of bright Hb9::GFP+ cells (bright was defined as cells in the top
50% of Hb9::GFP intensity by FACS). We observed that 90% of bright Hb9::GFP+ cells had an
EU intensity in the top quartile of EU intensity evaluated in all cells, meaning that of the
Hb9::GFP+ cells that advanced to the terminal neuronal stage of reprogramming, the vast
54

majority of them originated from HHCs (Fig. 3.7F). Thus, these prospective isolation studies
indicate that HHCs possess significantly greater reprogramming potential than non-HHCs,
including cells that are hyperproliferative but do not hypertranscribe. Taken together, our data
indicate that the inability of most cells to sustain hypertranscription and hyperproliferation early
in conversion limits reprogramming to rare cell populations. By increasing the population of cells
capable of mediating both processes, we improve reprogramming to near-deterministic rates.
Fig 3.7 Hypertranscribing, hyperproliferating cells reprogram at near-deterministic rates. (A) Schematic of CFSE-EU-pulse
label assay to sort and label HHCs at 4 dpi followed by evaluation of Hb9::GFP intensity at 8 dpi. (B) CFSE intensity of cells treated
with CFSE at 1 dpi and flow sorted at 4, 6, and 8 dpi in 6FDDRR conditions. n = 3 independent transductions per condition. Mean
+/- s.e.m. One-way ANOVA. AU = arbitrary units. (C) EU intensity in cells pulsed with EU for 4 hours at 4 dpi and fixed at 4, 6, and 8
dpi in 6FDDRR condition. EU intensity measured by flow cytometry. n = 3 independent transductions per condition. Mean +/- s.e.m.
One-way ANOVA. AU = arbitrary units. (D) GFP intensity of Hb9::GFP+ cells at 14 dpi correlates with neuronal morphology,
increasing from fibroblast to neuronal cells (left to right). Scale bar represents 100 µm. (E) Percentage of Hb9::GFP+ cells in
6FDDDR conditions for various gated populations. Cells gated for low EU intensity (EU-Low), defined as cells having an EU
intensity in the lowest three quartiles, and high EU-intensity (EU-High), defined as cells having an EU intensity in the highest
quartile, at 8 dpi compared to all viable cells, both EU-high and EU-low (All Cells). n = 7-8 independent conversions. Mean +/- s.e.m.
One-way ANOVA. (F) Percentage of replated hyperproliferating cells in 6FDDRR conditions gated for high EU intensity (cells having
an EU intensity in the top quartile as measured by FACS) at 8 dpi. Hb9::GFP+ Bright cells (i.e. Hb9::GFP intensity in the top half of
all viable Hb9+ cells) display enrichment of EU-High cells. n = 4-8 independent conversions. Median +/- interquartile range. Kruskal-
Wallis test. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Sustained transgene expression differentiates complete from partial reprogramming.
Previous analysis of gene regulatory networks (GRNs) identified that components of the
fibroblast GRN remain active within induced neurons (iNs) (Cahan et al., 2014; Morris et al.,
2014). One interpretation of this finding is that induced neurons fail to adopt a fully neuronal
transcriptional program. Alternatively, mechanisms that limit induced neuron reprogramming
55

may arrest cells at intermediate states leading to heterogeneous cultures comprised of fully
neuronal and partially neuronal cells.
Using a combination of live imaging and cross-sectional analysis across multiple
endpoints, we identified a post-mitotic intermediate state characterized by Hb9::GFP reporter
activation and retention of a fibroblast morphology (Fig. 3.8A, top panel). This state frequently
preceded Hb9::GFP+ iMN formation (Fig. 3.8B), and in iMN conversions using 6F alone, at least
50% of Hb9::GFP+ cells remained trapped in this state characterized by a fibroblast
morphology.
We hypothesized that by enabling hypertranscription in hyperproliferative cells, DD and
DDRR might accelerate the transition from Hb9::GFP+ intermediates to iMNs with full adoption
of the motor neuron transcriptional state. To test this, we evaluated the two rate-limiting steps of
reprogramming, the activation of Hb9::GFP and the morphological remodeling to iMNs via
longitudinal tracking across the two-week window of conversion. In the presence of DD,
Hb9::GFP+ intermediates were four times more likely to adopt a neuronal morphology and fully
convert into iMNs compared to 6F alone cells (Fig. 3.8C, n=65-80 cells in 6F condition and
n=1200-1400 in 6FDD condition). To confirm these results, we counted Hb9::GFP+ cells by flow
cytometry to quantify the rate of Hb9::GFP activation at 8 dpi and used imaging to determine the
morphology of Hb9::GFP+ cells at 17 dpi for the 6F, 6FDD, and 6FDDRR conditions (Fig. 3.8D,
E). Similar to our longitudinal tracking data, the rates of Hb9::GFP activation and conversion into
iMNs correlate with the population size of HHCs. With 6F alone, less than 1% of cells activate
Hb9::GFP, while 8% and 40% activate Hb9::GFP in the DD and DDRR conditions, respectively
(Fig. 3.8D). Additionally, 50% of Hb9::GFP cells remained trapped in the fibroblastic
intermediate state with 6F alone (Fig. 3.8E). In contrast, addition of DDRR to the 6F cocktail
resulted in 90% of Hb9::GFP+ cells becoming iMNs at 17 dpi (Fig. 3.8E).
56

Fig 3.8. DDRR drives activation of Hb9::GFP, morphological remodeling to iMNs. (A) Representative images of Hb9::GFP+
cells with fibroblast (top) or neuronal (bottom) morphology at 17 dpi. Scale bars represent 20 µm. (B) Time series of Hb9::GFP+
intermediate to iMN. (C) Longitudinal tracking of cells to measure the rate at which Hb9::GFP+ intermediates adopt neuronal
morphology. n = 65-80 cells in the 6F condition and n =1200-1400 cells tracked in the 6FDD condition. (D) Percentage of
Hb9::GFP+ cells out of all viable cells for 6F, 6FDD, or 6FDDRR conditions measured by flow cytometry at 8 dpi. n = 6 independent
conversions per condition. Mean +/- s.e.m. One-way ANOVA. (E) Percentage of Hb9::GFP+ cells with neuronal morphology out of
total Hb9::GFP+ cells for 6F, 6FDD, or 6FDDRR conditions at 17 dpi. n = 9 independent conversions per condition. Mean +/- s.e.m.
One-way ANOVA. Significance summary: ∗∗∗∗p ≤ 0.0001.

To identify transcriptional patterns that differentiate successful from unsuccessful
reprogramming, we collected cells at 14 dpi and flow sorted based on Hb9::GFP+ into three
populations: No, Low, and Bright Hb9::GFP, with Bright Hb9::GFP again being defined as cells
in the top 50% of Hb9::GFP intensity in the 6F condition. With a panel of fibroblast and neuronal
markers, we used qRT-PCR to evaluate gene expression across the populations. We observed
three primary clusters of gene expression. Cells lacking Hb9::GFP (No, top, grey) expressed
high levels of a cluster enriched with fibroblast genes (Cluster 1, left, gray) compared to Low
and Bright Hb9::GFP populations (Fig. 3.9A). Cluster 3 (left, bright green) genes were enriched
with transgenes and neuronal markers and highly expressed in the Bright Hb9::GFP population
57

(Bright, top, bright green) (Fig. 3.9A). Hb9::GFP Bright cells were more neuronal and showed
sustained transgene expression compared to the other two populations (Fig. 3.9A, B). These
results suggest that the ability to sustain the expression of the exogenous reprogramming
transcription factors is critical for reaching the Bright Hb9::GFP+ iMN state. We have previously
verified that functional iMNs retain transgene expression. However, shutting off transgene
expression after iMN formation by using doxycyline-inducible reprogramming vectors does not
overtly affect iMN viability or function (Son et al., 2011).
To identify the transcriptional state associated with the Hb9::GFP+ intermediates, we
isolated individual Hb9::GFP+ iMNs with a neuronal morphology and Hb9::GFP+ intermediates
with fibroblast morphologies (Fig. 3.8A) and measured gene expression via qRT-PCR. iMNs
(Fig. 3.8A, bottom) displayed increased expression of neuronal markers relative to Hb9::GFP+
fibroblast-like intermediates (Fig. 3.9C). Hb9::GFP+ fibroblasts did not show substantially more
fibroblast gene expression than Hb9::GFP+ neurons, suggesting that activating neuronal gene
expression rather than suppressing fibroblast expression was the limiting step in the Hb9::GFP+
intermediate stage (Fig. 3.9C). In particular, we observed that high expression of Isl1, both
transgenic and endogenous, differentiated neuronal from fibroblast morphologies (Fig. 3.9D).
58

Fig 3.9. Activation of neuronal genes differentiates successful from unsuccessful reprogramming. (A) Relative gene
expression of cells collected at 14 dpi sorted based on No, Low, or Bright Hb9::GFP expression. Bright Hb9::GFP was defined as
cells in the top 50% of Hb9::GFP in the 6F condition. Gene expression was calculated based on qRT-PCR data. The expression
level that was highest amongst the three conditions was set to 1 and used to normalize levels for the other two conditions. n = 2
independent experiments for each condition. (B) Relative expression of viral Isl1 (vIsl1), viral Lhx3 (vLhx3), viral Ngn2 (vNgn2), or
endogenous Isl1 of cells collected at 14 dpi sorted based on no, low, or bright Hb9::GFP expression. n = 3 independent
transductions per condition. Mean +/- s.e.m. One-way ANOVA within each dpi. (C) Heatmap of relative expression for single cells
with either fibroblast (top gray, n=16) or neuronal (top green, n=39) morphology for qPCR assays for fibroblast (side gray) or
neuronal (side green) genes. Cells were picked at 14 dpi. (D) Relative expression for single cells with either fibroblast (n=16) or
neuronal (n=39) morphology for qPCR assays for endogenous Ngn2, Isl1, viral Isl1 (vIsl1), and endogenous Lhx3. Median +/-
interquartile range. Mann-Whitney test. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001.

To examine how transgene expression is affected during reprogramming in different
populations and conditions, we constructed a fusion construct of Isl1 with GFP. This Isl1-GFP
fusion was insufficient to replace Isl1 in reprogramming, suggesting the fusion impacted the
function of Isl1 (Fig. 3.10A). In reprogramming, we observed that Isl1-GFP intensity dropped in
hyperproliferative cells in both 6F and 6FDDRR conditions (Fig. 3.10B). DDRR doubled the
percentage of cells with detectable Isl1-GFP expression in hyperproliferative cells, suggesting
that DDRR could sustain transgene activation that was diminished in hyperproliferative cells
(Fig. 3.10C). To evaluate how multiple virus expression varied in 6F and 6FDDRR, we used
59

YFP and DsRed labeled viruses (Fig 3.10D, E). While individual viruses showed high
expression efficiency in 6F and 6FDDRR after single virus infections, with 80-90% of cells
showing detectable fluorescence by imaging (Fig. 3.10D), the percentage of cells exhibiting
detectable expression of both fluorescent proteins upon double infection was significantly lower
in 6F than 6FDDRR (Fig. 3.10E).

Figure 3.10. DDRR sustains transgene expression in hyperproliferating cells during reprogramming. (A) Yield of iMNs in 6F
conditions or with 5F (i.e. Ascl1, Brn2, Myt1l, Ngn2, Lhx3) +Isl1-GFP at 14 dpi. n = 6-7 independent conversions per condition.
Mean +/- s.e.m. Unpaired t-test. (B) Relative Isl1-GFP intensity in all viable cells (All Cells) and hyperproliferative cells (HyperP
Cells) infected with Isl1-GFP and 6F or 6FDDRR measured by flow cytometry at 4 dpi. n = 4-6 independent transductions per
condition. Mean +/- s.e.m. One-way ANOVA. (C) Percentage of Isl1-GFP+ cells in 6F and 6FDDRR in all viable cells (All Cells) and
hyperproliferative cells (HyperP Cells) measured by flow cytometry at 4 dpi. Isl1-GFP+ determined by expression exceeding FITC
values for untransfected cells. n = 6 independent transductions per condition. Mean +/- s.e.m. One-way ANOVA. (D) Percentage of
6F or 6FDDRR cells expressing a fluErescent protein. Cells were infected with a single fluorescent protein (e.g. individual viruses
YFP or RFP) and measured via flow cytometry at 4 dpi. n = 3 independent transductions per condition. Mean +/- s.e.m. One-way
ANOVA. (e) Percentage of 6F or 6FDDRR cells infected with both YFP and RFP and expressing either fluorescent protein alone or
both. Cells were measured via flow cytometry at 4 dpi. n = 3 independent transductions per condition. Mean +/- s.e.m. One-way
ANOVA. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

To determine if the increased percentage of 6FDDRR cells producing detectable levels
of both transgenic proteins was due to an increased number of transgene integrations in
6FDDRR cells or an increased ability of 6FDDRR cells to maintain high expression levels of
transgenic integrations, we tested the number of transgenic integrations in DDRR vs. non-
DDRR conditions. To measure this in an iMN reprogramming-relevant context but eliminate the
60

complexity of having 6 individual transcription factors, we constructed a polycistronic cassette of
the motor neuron reprogramming factors Ngn2, Isl1, and Lhx3 (NIL). These factors had been
previously demonstrated to reprogram ES cells to motor neurons (Mazzoni et al, 2013). This
system allowed us to control for transcription factor stoichiometry and ensure uniform delivery of
the transcription factors to the cells. Utilizing the polycistronic cassette to ensure that cells
infected with one virus had all three transcription factors, we observed that NIL is sufficient to
mediate reprogramming and DDRR increased HHCs and reprogramming in this system (Fig
3.11A, B). Therefore, the trends observed with the 6F are preserved with a single multicistronic
cassette indicating that the HHC population and reprogramming rate are not simple functions of
multiple different viruses. At 4 dpi, we isolated genomic DNA from NIL and NIL+DDRR
conditions and used qPCR to determine the number of integrations of the polycistronic set
compared to a native genomic region. We found that NIL+DDRR did not have more integrations
of the NIL cassette than NIL cells without DDRR (Fig. 3.11C). Similarly, NIL+DDRR cells
transduced with Isl1-GFP from the previous experiments did not contain more Isl1-GFP
transgenes than NIL cells (Fig. 3.11C). Thus, DDRR does not enable higher transgene
expression by increasing the number of transgene integrations. Instead, DDRR enables higher
levels of transgene expression in hyperproliferative cells, which leads to efficient activation of
Hb9::GFP and transition of partially-reprogrammed intermediates to the neuronal state. Taken
together these data indicate that transgene expression levels in proliferating cells, rather than
retroviral infection rates, limit reprogramming. These results further support a model of tradeoffs
between transcription and proliferation that must be delicately balanced.

61

Fig 3.11. DDRR does not promote reprograming by increasing viral integrations. (A) Percent HHCs measured at 4 dpi with the
multicistronic NIL virus (NIL) and NIL+DDRR. Hyperproliferating cells were defined as cells showing a two-fold increase in division
rate (i.e. an eight-fold decrease in CFSE intensity) compared to the average of the control population, which was comprised of NIL-
infected MEFs. n = 3 independent transductions per condition. Mean +/- s.e.m. Unpaired t-test. (B) Yield of iMNs in NIL and
NIL+DDRR conditions at 14 dpi. n = 4-5 independent conversions per condition. Mean +/- s.e.m. Unpaired t-test. (C) Relative
integrations of Isl1-GFP and NIL viruses in NIL and NIL+DDRR conditions in cells collected at 4 dpi. Relative integrations
determined by qPCR of genomic DNA. Delta Ct of transgene calculated by difference of Ct between transgene and endogenous
genomic region. Relative integrations calculated by normalizing to NIL condition n = 3 independent transductions per condition.
Mean +/- s.e.m. One-way ANOVA. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Topoisomerase expression enables simultaneous hypertranscription, hyperproliferation
in HHCs
To identify the mechanisms that enable combined hypertranscription and
hyperproliferation in the presence of DDRR, we transcriptionally profiled single cells at different
time points in iMN conversion (Fig. 3.12A). To focus our analysis on cells that were on a
successful reprogramming trajectory, we collected hyperproliferative cells at 4 dpi (CFSE-Low),
and Hb9::GFP+ cells at 8 dpi and 14 dpi (Fig. 3.12A).
To determine if cells in 6F and 6FDDRR conditions take similar or distinct trajectories
from the fibroblast to the iMN state, we profiled cells from both conditions (Fig. 3.12A, B).
Analysis of tSNE clustering showed that 6F and 6FDDRR iMNs were similar to each other
relative to MEFs and reprogramming cells, suggesting that cells could take similar trajectories to
reach the iMN state in either condition (Fig. 3.12B). Analysis of tSNE clustering indicated that
converting cells at 4 dpi and 8 dpi mapped between fibroblasts and fully-converted iMNs (Fig.
3.12B). As expected, cells at 4 dpi clustered closer to fibroblasts and Hb9::GFP+ cells at 8 dpi
clustered closer to iMNs (Fig. 3.12B). At 4 dpi, cells with 6F alone mapped to similar locations
62

as cells in 6FDDRR conditions (Fig. 3.12B). Pseudotime analysis indicated that at 8 dpi, more
cells were proximal to the iMN state in the 6FDDRR condition than in the 6F condition (Fig.
3.12C, note that the color scheme is consistent amongst Fig. 3.12C, E-H, and this scheme is
distinct from that used in Fig. 3.12B). Together, these results suggest that cells traverse through
a conserved trajectory during lineage conversion regardless of condition, but DDRR increases
the speed and efficiency of reprogramming.

63

Fig 3.12. DDRR accelerates reprogramming cells across a conserved trajectory. (A) Schematic of populations collected across
conversion and profiled via single-cell RNAseq. Individual libraries were prepared for MEFs (1357 cells), hyperproliferating cells
(CFSE-Low) for 6F (1174 cells) and 6FDDRR (1189 cells) collected at 4 dpi (6F 4dpi, 6FDDRR 4dpi), Hb9::GFP+ cells for 6F (259
cells) and 6FDDRR (406 cells) at 8 dpi (6F 8dpi, 6FDDRR 8dpi), 6F iMNs (1863 cells) and 6FDDRR iMNs (2869 cells) at 14 dpi
(iMNs). (B) tSNE projection of all cells mapped during reprogramming colored by condition. (C) Distribution of pseudotime across
cells in each condition. (D) Relative UMI distribution across cells. (E) Clustering of three cellular states across the tSNE projection.
(F) Relative expression of Col1a1, Mki67, Top2a, Top1, and Map2 over pseudotime. Colors correspond to states identified in (E).
(G) Violin plot of UMI (top, unique molecular identifiers) and relative Mki67 expression (bottom) for clusters identified in (E). (H)
Violin plot of relative expression of Top1 (top) and Top2a (bottom) for clusters identified in (E). (I) Reads from Top1 and Top2a in 6F
and 6FDDRR quantified by cell number normalized (CNN) RNAseq at 4 dpi. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤
0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001.

64

We next examined the different single cell states to identify the transcriptional programs
that enable combined hypertranscription and hyperproliferation (Fig 3.12D-H). As expected,
fibroblasts expressed the highest levels of collagen genes such as Col1a1, while converting
cells decreased collagen gene expression during transit to iMNs (Fig. 3.12F). Map2, a marker of
post-mitotic neurons, increased over pseudotime (Fig. 3.12F). Converting cells grouped within
two states. To establish clusters of cells, we utilized the Monocle 2 pseudotemporal ordering
algorithm to cluster cells based on differentially expressed genes and aligned cells along a
reprogramming trajectory. Most State 1 (red) cells remained close to the starting fibroblasts with
high expression Col1a1, although some cells from the 6F iMN population clustered into State 1
most likely based on sustained Col1a1 expression (Fig. 3.12E). In contrast, State 2 (blue)
contained cells with a proliferative signature including high expression of Mki67 (Fig. 3.12F,G).
We note that although Mki67 is higher in State 2 than in State 1, we observed that about 80% of
MEFs were Ki67+, indicating that they were proliferating (Fig. 3.2C). Therefore, the low
apparent level of Mki67 in State 1 is due to the limited sensitivity of single cell RNA sequencing
and indicates that Mki67 transcription increases in State 2 above levels observed in proliferating
MEFs. Additionally, State 2 was enriched in unique molecular identifiers (UMIs), a proxy to total
mRNAs (Fig. 3.12D, G), signifying a putative HHC population.
In the State 2, we identified increased expression of two topoisomerases (Fig. 3.12H).
Top1 expression increased at early stages and was sustained throughout reprogramming, while
Top2a showed transient expression, peaking as cells transitioned from fibroblasts (high Col1a1,
low Map2) to iMNs (low Col1a1, high Map2) (Fig. 3.12F). Bulk RNA-seq analysis comparing 6F
and 6FDDRR conditions at 4 dpi confirmed that the addition of DDRR significantly increased
levels of Top1 and Top2a (Fig. 3.12I). Topoisomerases decatenate DNA and reduce DNA
torsion induced by transcription and replication as well as resolve collisions between
transcription and replication machinery (Keszthelyi et al., 2016; Tuduri et al., 2009). Therefore,
65

we hypothesized that increased topoisomerase expression induced by DDRR may allow HHCs
to sustain high rates of replication and transcription.

Fig 3.13. DDRR upregulates topoisomerases to enable combined hypertranscription and hyperproliferation. (A) Percentage
of mitotic anaphase-telophase cells with a chromatin bridge at 4 dpi for 6FDDRR conditions treated with Scrambled, Top1, or Top2a
shRNAs. n = 3 independent conversions per condition, n=50-75 cells per condition. Significance determined using a Chi-square test
to compare the frequency in encountering a mitotic cell with a chromatin bridge between conditions. Percentage +/- 95% confidence
interval. (B) Percentage of HHCs in 6FDDRR conditions treated with Scrambled, Top1, or Top2a shRNAs measured at 4 dpi as
described previously. n = 4-6 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (C) Percentage of HHCs in
6FDDRR conditions treated for 18 hrs with camptothecin (Cpt) or doxorubicin (Doxo) prior to 4 dpi compared to DMSO control
measured as described previously. n = 4-5 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. Significance
summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

To examine the necessity of each topoisomerase in supporting the HHC population and
reprogramming, we introduced shRNAs targeting either Top1 or Top2a. Reduction of Top1 or
Top2a significantly increased genomic stress and reduced the population of HHCs in 6FDDRR
conditions (Fig. 3.13 A, B). In addition, transient inhibition of TOP1 or TOP2A by camptothecin
or doxorubicin treatment, respectively, decreased the percentage of HHCs under 6FDDRR
conditions (Fig. 3.13C). As would be expected, inhibition of TOP2A by doxorubicin treatment
significantly reduced the population cells displaying active DNA synthesis as determined by
having reduced EdU incorporation compared to DMSO treated controls, and similar EdU
incorporation compared to irradiated MEFs. Consistent with HHCs comprising the majority of
66

the reprogramming-competent cell population, shRNA knockdown of either topoisomerase or
inhibition of TOP1 or TOP2A activity by camptothecin or doxorubicin, respectively, resulted in a
significant drop in iMN yield in the 6FDDRR condition (Fig. 3.14A, B). To test the sufficiency of
the topoisomerases to increase reprogramming, we overexpressed Top1 and mCherry-tagged
Top2a, or mCherry alone during conversion. Overexpression of Top2a did not increase
conversion (Fig. 3.14C, D). Given that Top2a is not expressed in post-mitotic cells and is only
transiently induced during 6FDDRR reprogramming (Fig. 3.13F), constitutive overexpression
may prohibit the formation of post-mitotic cells. Consistent with TOP1 playing a key role in
reprogramming, the addition of Top1 significantly increased iMN conversion (Fig. 3.14E, F).
These results indicate that the upregulation of topoisomerases by DDRR is a critical mechanism
by which these conditions balance replication and transcription to promote the HHC state and
enable highly efficient reprogramming.

67

Figure 3.14. DDRR-mediated topoisomerase expression drives robust reprogramming. (A) Yield of iMNs in 6DDDRR
conditions treated with Scrambled, Top1, or Top2a shRNAs quantified as described previously at 14 dpi. n = 7-9 independent
conversions per condition. Mean +/- s.e.m. One-way ANOVA. (B) Yield of iMNs in 6FDDRR conditions treated for 18 hrs with
camptothecin (Cpt) or doxorubicin (Doxo) prior to 4 dpi compared to DMSO control quantified as described previously at 14 dpi. n =
3-4 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (C) Effect of mCherry, Top2a-T2A-mCherry, or
Top2a-T2A-mCherry +Top1 overexpression on percentage of Hb9::GFP+ iMNs at 14 dpi for 6FDD+RepSox condition. n = 3-4
independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (D) Effect of mCherry, Top2a-T2A-mCherry, or Top2a-
T2A-mCherry +Top1 overexpression on percentage of Hb9::GFP+ and mCherry+ double positive iMNs at 14 dpi for 6FDD+RepSox
condition. n = 3-4 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (E) Effect of Top1 overexpression on
conversion yield of 6FDD with and without RepSox at 14 dpi. n = 7-14 independent conversions per condition. Median +/-
interquartile range. Kruskal-Wallis test. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤
0.0001.

DDRR and topoisomerase expression reduces negative DNA supercoiling and R-loop
formation, and sustain transcription in S-phase.
To determine how topoisomerases help to balance replication and transcription to
promote the HHC state, we first examined levels of DNA supercoiling in the genome during
reprogramming. Transcription and DNA replication increases levels of positive and negative
supercoiling in the genome (Ma and Wang 2016). Negative supercoiling promotes R-loop
formation, which in turn can induce DNA replication fork stalling (Manzo et al., 2018; Gan et al.,
2011). To investigate if reprogramming perturbed the normal supercoiling state, we incubated
cells with trimethylpsoralen, which preferentially intercalates into negatively supercoiled DNA
68

(i.e. underwound) and can therefore quantify the amount of negative supercoiling in the genome
(Teves et al, 2014; Naughton et al, 2013), Because DNA synthesis can significantly influence
DNA supercoiling levels (Kurth et al 2013) and our Control-Puro, 6F, and 6FDDRR cultures
contained differing amounts of proliferative cells, we normalized DNA synthesis in these
conditions before detecting supercoiling by treating cells with 1 µM aphidicolin to inhibit of DNA
polymerases.
We first used biotinylated trimethylpsoralen to enable streptavidin-conjugated dye-based
quantification of trimethylpsoralen intercalation in the nuclei of cells at 4 dpi (Fig. 3.15A)
(Naughton et al, 2013; Bermúdez et al, 2010). Naughton et al. previously used this approach to
measure negative supercoiling in cells (Naughton et al, 2013). When we examined
reprogramming cultures at 4 dpi, we observed increased biotinylated trimethylpsoralen
intercalation in 6F MEFs compared to Control-Puro MEFs, indicating that transcription factor
overexpression increased overall levels of negative supercoiling in the genome (Fig. 3.15A, B).
A central function of topoisomerases is to relax supercoiled DNA to mitigate mechanical
stress across the genome (Vos et al., 2011). Consistent with this known role of topoisomerases,
the addition of DDRR to 6F normalized trimethylpsoralen incorporation, indicating that DDRR
reduced negative supercoiling (Fig. 3.15A, B). shRNA knockdown of either topoisomerase in
6FDDRR conditions blocked the reduction in trimethylpsoralen intercalation (Fig. 3.15C),
indicating that topoisomerase activity is required for DDRR’s ability to limit the accumulation of
negative supercoils.

69

Figure 3.15. Topoisomerases enable DDRR-mediated reduction of negative supercoiling. (A) Representative images of
psoralen incorporation in Control-Puro, 6F, and 6FDDRR conditions at 4 dpi. Biotinylated psoralen incorporation into DNA visualized
by conjugation to streptadvidin-Alexa Fluor 594 and by Hoecsht. Scale bars represent 10 µm. Dotted white lines outline the nucleus.
(B) Mean intensity of biotinylated psoralen conjugated streptadvidin-Alexa Fluor 594 at 4 dpi in Control-Puro, 6F, and 6FDDRR
conditions treated with 1 µM aphidicolin for 2 hours prior to collection at 4 dpi. Mean intensity determined by total intensity in nuclear
area as determined by Hoecsht and normalized to total nuclear area. n=42-130 cells from 3 independent conversions per condition.
Median +/- interquartile range. Kruskal-Wallis test. (C) Mean intensity of biotinylated psoralen conjugated streptadvidin-Alexa Fluor
594 at 4 dpi in 6FDDRR+Scrambled shRNA, shTop1, and shTop2a shRNA conditions at 4 dpi. n = 99-162 cells from 3 independent
conversions per condition. Median +/- interquartile range. Kruskal-Wallis test.

Transcription bubbles induce negative supercoiling upstream of the transcription start
site (Ma and Wang 2016; Ma et al, 2013). Previous studies have quantified negative
supercoiling in these specific genomic regions by using trimethylpsoralen cross-linking to protect
negatively-supercoiled DNA regions against digestion with purified exonuclease I, enabling their
amplification and quantification by qPCR (Teves and Henikoff 2014). Thus, to examine negative
supercoiling levels with greater genomic resolution, we employed this technique. When we used
this approach to quantify negative supercoiling 500 base pairs upstream of the transcription
start sites of three genes that showed similar levels of expression between 6F and 6FDDRR
70

cells (Gapdh, Actb, Sod1) at 4 dpi, we found that 6FDDRR cells had significantly less negative
supercoiling at all three promoters than 6F cells (Fig. 3.16A). Taken together, our results show
that transduction of the six reprogramming transcription factors increases negative DNA
supercoiling and that DDRR reduces this supercoiling in a topoisomerase-dependent manner.

Figure 3.16. DDRR rescues transcription factor-induced negative supercoiling, reduces DNA torsion. (A) Relative amount of
DNA protected by exonuclease digestion in regions 500bp upstream of transcription start sites for listed genes in 6F or 6FDDRR
cells at 4 dpi. Relative DNA protected by exonuclease digestion determined by psoralen intercalation in exonuclease-digested
compared to non-exonuclease-digested control for each sample and measured with TMP-qPCR. n = 4 independent transductions
per condition per gene. Mean +/- s.e.m. Unpaired t-test. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01.

R-loops are RNA-DNA hybrid structures that form between genomic DNA and nascent
transcripts (Thomas et al, 1976). Negatively supercoiled DNA and high transcription rates both
promote the formation of R-loops (Aguilera and García-Muse 2012), which in turn impair DNA
replication and reduce genomic stability (Gan et al., 2011). Using an R-loop-specific antibody
(S9.6) (Boguslawski et al, 1986), we employed both dot blot analysis of cell lysates (Fig. 3.17A,
B) and immunofluorescence analysis of individual cells (Fig. 3.17C-E) to probe cells for R-loop
formation at 4 dpi in Control-Puro, 6F, and 6FDDRR conditions. We first verified that the S9.6
antibody detected R-loops by treating cultures with RNAse H to degrade R-loops. As expected,
RNAse H treatment reduced S9.6 signal intensity in the dot blot assay (Fig. 3.17A, B) and in
non-nucleolar nuclear regions as measured by immunofluorescence (Fig. 3.17C).
Consistent with our findings that transduction with the six reprogramming factors
increases transcription and negative supercoiling, 6F cells showed increased R-loop formation
compared to Control-Puro cells by immunofluorescence and dot blot (Fig. 3.17A, B, D, E).
71

Importantly, DDRR significantly reduced R-loop formation compared to 6F (Fig. 3.17A, B, D, E).
In addition, shRNA-mediated suppression of either Top1 or Top2a increased R-loop formation in
DDRR conditions (Fig. 3.17F). These results indicate that 6F increases R-loop formation
through elevated transcription and negative supercoiling levels and DDRR reduces R-loops in a
topoisomerase-dependent manner.

Figure 3.17. DDRR reduces formation of R-loops via topoisomerases. (A) Representative images of dot blot of S9.6 R-loop and
ssDNA intensities for Control-Puro, 6F, and 6FDDRR at 4 dpi. n = 6 independent transductions per condition. (B) Relative R-loop
intensity quantified and normalized to ssDNA for Control-Puro, 6F, and 6FDDRR at 4 dpi. n = 6 independent transductions per
condition. Mean +/- s.e.m. One-way ANOVA. (C) R-loop intensity per area excluding nucleoli in 6F MEFs treated with buffer or
RNAse H. n = 110-119 cells from 3 independent conversions per condition. Median +/- interquartile range. Mann-Whitney test. (D)
Representative images of Control-Puro, 6F, or 6FDDRR conditions immunostained for R-loops (S9.6) at 4 dpi. Scale bars represent
10 µm. Dotted white lines outline the nucleus. (E) R-loop intensity per area at 4 dpi in Control-Puro, 6F, and 6FDDRR. R-loop
intensity per area determined by S9.6 intensity normalized to total nuclear area. n=101-158 cells from 3 independent conversions
per condition. Median +/- interquartile range. Kruskal-Wallis test. (F) R-loop intensity per area at 4 dpi in 6FDDRR+Scrambled
shRNA, shTop1-A, and shTop2a-A shRNAs measured as described previously. n=119-135 cells from 3 independent conversions
per condition. Median +/- interquartile range. Kruskal-Wallis test. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤
0.001, and ∗∗∗∗p ≤ 0.0001.

To determine if the increase in DNA supercoiling and R-loop formation after 6F
transduction affects DNA replication fork processivity, we performed a DNA fiber labeling assay.
The DNA fiber labeling assay utilizes pulse-labeling of IdU for 20 minutes followed by a 30
72

minute labeling period with CldU. Patterns of IdU and CldU labeling mark different replicative
species including progressing forks (fibers containing IdU and CldU labeling), stalled forks
(fibers containing only IdU labeling), and new origins (fibers containing only CldU labeling)
(Kotsandis et al, 2016; Nieminuszcy et al, 2016) (Fig. 3.18A). Examining over 4000 fibers per
condition at 4 dpi, we observed a higher percentage of stalled replication forks in 6F MEFs
compared to Control-Puro MEFs at 4 dpi (Fig. 3.18B, D top). 6FDDRR MEFs possessed a
significantly lower percentage of stalled replication forks than 6F MEFs (Fig. 3.18B, D, top).
Additionally, 6FDDRR MEFs initiated a higher rate of new replication origins compared to 6F
cells (Fig 3.18C, Dbottom). Together, these data indicate that the forced expression of the iMN
reprogramming factors impairs cell proliferation at least in part by increasing DNA replication
fork stalling, and that DDRR rescues replication fork stalling caused by 6F overexpression. In
addition, DDRR induces the formation of more DNA replication origins, which can also facilitate
hyperproliferation. Given that DNA replication fork stalling is linked to genome instability
(Branzei and Foiani 2010), increased fork stalling may directly contribute to the increased
micronuclei and chromatin bridges we observed in 6F MEFs. Together, these results indicate
that the forced expression of reprogramming transcription factors induce genomic stress by
increasing negative DNA supercoiling, R-loop formation, and DNA replication fork stalling.
DDRR rescues these sources of genomic stress in a topoisomerase-dependent manner.

73

Figure 3.18. DDRR restores replication fork processivity. (A) DNA fiber labeling scheme to identify progressing replication forks
(red-green), stalled forks (red only), and new origins (green only). Cells are incubated with IdU for 20 minutes followed by CldU for
30 minutes. IdU and CldU labels are probed by specific antibodies and fluorescently labelled secondary antibodies. (B) Relative
number of stalled replication forks in Control-Puro, 6F, and 6FDDRR conditions at 4 dpi. Stalled replication forks were quantified and
normalized to all replicative fiber species to generate the percentage of stalled replication forks. n = 1000 fibers per condition from 4
independent transductions. Percentage +/- 95% confidence interval. Fisher’s exact test. (C) Relative number of new origins in
Control-Puro, 6F, and 6FDDRR conditions at 4 dpi. New origins were quantified and normalized to all replicative fiber species to
generate the percentage of stalled replication forks. n = 1000 fibers per condition from 4 independent transductions. Percentage +/-
95% confidence interval. Fisher’s exact test. (D) Representative images of stalled/terminated replication forks and new origins in
DNA fiber labeling assay for 6F and 6FDDRR conditions at 4 dpi. Scale bar represents 10 µM. Significance summary: p > 0.05 (ns),
∗p ≤ 0.05, ∗∗p ≤ 0.01.

To refine the effects of the topoisomerases on transcription and DNA replication during
reprogramming, we examined how transcription was impacted in S-phase cells. Studies of
nuclear reprogramming through cellular fusion identified DNA synthesis as a vital determinant of
somatic cell reprogramming (Tsubouchi et al, 2013). During S-phase, replication forks stall as
they encounter replication blocks (e.g. secondary structures, R-loops, colliding RNA and DNA
polymerases, DNA supercoils) (Branzei et al, 2010). In normal cell cycle, global transcription
rates decrease during S-phase. The induction of replication stress through aberrant transcription
results in genome instability (Kotsandis et al, 2016). Given that TOP1 prevents genome
instability by preventing interference between replication and transcription (Tuduri et al, 2009),
we hypothesized that TOP1 may support higher rates of transcription during S-phase by
mitigating conflicts between transcription and replication machinery.
To identify S-phase cells at 4 dpi, we utilized an EdU-click method to label newly
synthesized DNA. Phosphorylation of serine 2 within the RNA polymerase II CTD
74

(RNAPIISer2p) marks the active, elongating form of RNA polymerase II (Phatnani and
Greenleaf 2006). To measure transcriptional activity in S-phase cells, we examined
RNAPIISer2p levels via immunolabeling of EdU-clicked cells and analyzed both fluorescent
labels by flow cytometry (Fig. 3.19A). We defined high RNAPIISer2p as defined as the top
quartile of RNAPIISer2p intensity in Control-Puro infected cells in S-phase cells. We observed
that RNAPIISer2p intensity varied by condition in S-phase cells (Fig. 3.19B). We observed that
S-phase cells with high RNAPIISer2p were limited in Control-Puro and 6F conditions (~ 5%)
whereas 6FDDRR sustained a 3-fold larger percentage of the population (Fig. 3.19B).
Additionally, synthesis rates of both DNA and RNA in S-phase cells, as measured by EdU
intensity and active RNAPII intensity, respectively, were higher in 6FDDRR compared to 6F
(Fig. 3.19C-F). Knockdown of Top1, but not Top2a reduced the percentage of S-phase cells
with high RNAPIISer2p (Fig. 3.19G, H). These data suggest that TOP1 is responsible for
maintaining high transcription rates via active RNAPII in rapidly proliferating cells. Thus, DDRR
enables hypertranscription in hyperproliferating cells at least in part by increasing RNA
polymerase II activity through TOP1 activation.
Together, our data suggest a model in which TOP1, supported by TOP2A, enables high
rates of RNA and DNA synthesis by removing barriers to synthesis including DNA supercoiling,
R-loops, and DNA replication fork stalling.
75

Figure 3.19. DDRR sustains high levels of transcription in actively synthesizing cells via topoisomerases. (A)
Representative dot plot of EdU and active RNA Polymerase II intensity at 4 dpi for Control-Puro (grey), 6F (green), and 6FDDRR
(red). Gating to demark S-phase cells with high active RNAPII (RNAPII Ser2p). S-phase determined by intensity above EdU
incorporation in non-proliferative, irradiated MEFs (Fig. 3.13E). High RNAPIISer2p as defined as the top quartile of RNAPIISer2p
intensity in Control-Puro infected cells in S-phase cells. (B) Percentage of Control-Puro, 6F, and 6FDDRR cells in S-phase with high
RNAPII activity from area gated in (A) measured via flow cytometry at 4 dpi. n = 4 independent conversions per condition. Mean +/-
s.e.m. One-way ANOVA. (C) Relative DNA synthesis rate of S-phase cells in Control-Puro, 6F, and 6FDDRR at 4 dpi. Relative DNA
synthesis rate determined by EdU intensity of S-phase population and normalized to EdU intensity of S-phase population in Control-
Puro condition. n = 4 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (D) Relative active RNAPII of S-
phase cells in Control-Puro, 6F, and 6FDDRR at 4 dpi. Relative active RNAPII rate in S-phase cells determined by intensity of
RNAPII Ser2p in S-phase population and normalized to intensity of RNAPII Ser2p in S-phase population in Control-Puro condition. n
= 4 independent conversions per condition. Mean +/- s.e.m. One-way ANOVA. (E) Percentage of cells in S-phase for Control-Puro,
6F, or 6FDDRR conditions measured at 4 dpi via EdU incorporation using flow cytometry. n = 3-4 independent transductions per
condition. Mean +/- s.e.m. One-way ANOVA. (F) Fraction of S-phase cells with high RNAPII activity measured via RNAPII Ser2p
intensity at 4 dpi via flow cytometry. S-phase determined as previously described. High RNAPIISer2p defined as the top quartile of
RNAPIISer2p intensity in Control-Puro infected cells in S-phase cells. n = 4 independent transductions per condition. Mean +/-
s.e.m. One-way ANOVA. (G) Representative dot plot of EdU and active RNA Polymerase II intensity at 4 dpi for
6FDDRR+Scrambled shRNA (grey), 6FDDRR+shTop2a (blue), and 6FDDRR+shTop1 (red) shRNAs. Gating to demark S-phase
cells with high active RNAPII (RNAPII Ser2p). (H) Percentage of cells in S-phase with high RNAPII activity from area gated in (G) at
4 dpi for 6FDDRR+Scrambled shRNA, 6FDDRR+shTop1, and 6FDDRR+shTop2a shRNAs. n = 4 independent conversions per
condition. Mean +/- s.e.m. One-way ANOVA. Significance summary: p > 0.05 (ns), ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤
0.0001.
76

Discussion
Although direct lineage conversion enables access to increasing numbers of somatic cell
types for translational studies, reprogramming remains a rare event. Studies investigating
universal strategies for overcoming barriers to lineage conversion into somatic cell states have
been limited. Our studies into the systems-level features of reprogramming have identified
combined hypertranscription and hyperproliferation as a central driver of reprogramming that is
able to overcome epigenetic barriers to lineage conversion across multiple species and somatic
cell states.
Our data show that combined hypertranscription and hyperproliferation is rare because
transcription and proliferation antagonize each other during reprogramming. We show that the
forced expression of the reprogramming transcription factors increases genomic stress in the
form of DNA torsion (supercoiling), R-loops, and reduced processivity of DNA replication forks.
Because of this, few cells possess the processing capacity to mediate high rates of both
processes. Consequently, reprogramming remains restricted to rare, privileged cells with high
transcriptional and proliferative capacity. These privileged cells contribute to the majority of
reprogramming events and reprogram at near-deterministic rates. By introducing chemical and
genetic perturbations that mitigate antagonism, we expand capacity for high rates of coincident
transcription and proliferation. By extending the properties of HHCs beyond the normally
resident population of HHCs, we boost the reprogramming rate and extend conversion to
otherwise unreprogrammable cells.
Through single-cell RNA-seq, we identified topoisomerase expression as a key
parameter modulating a cell’s ability to sustain combined hypertranscription and
hyperproliferation and undergo reprogramming. Topoisomerases mediate collisions between
transcriptional and DNA-replicative machinery as well as resolve DNA torsion introduced by
both processes. Our data indeed show that DDRR resolves DNA torsion, R-loops, and DNA
77

replication fork stalling induced by the forced expression of the reprogramming transcription
factors. In addition, we show that this DDRR effect is dependent on TOP1 and TOP2A. Our data
suggest a model in which limits in reprogramming arise from biophysical antagonism between
transcription and replication rates that results from topoisomerase-related mechanisms such as
torsional strain. When replication and transcription exceed the cell's capacity to resolve
topological tangles and DNA breaks, both replication and transcription stall, retarding
reprogramming processes. In the absence of perturbations that enable hyperproliferation and
sustain hypertranscription, few cells possess the cellular machinery to balance the dynamic
demands of rapid proliferation and hypertranscription. Increasing expression of topoisomerases
expands the cell’s ability to mediate conflict between these two processes, enabling robust
cellular reprogramming. Our data suggest that TOP1 principally promotes transcription and
TOP2A promotes replication. Supporting the competing, dynamic demands of transcription and
replication requires balanced expression of each topoisomerase (Fig. 3.20).
78

Fig. 3 20. Model of topoisomerase-mediated reprogramming through hypertranscribing, hyperproliferating cells.
Introduction of the reprogramming factors (TFs) to fibroblasts induces transcription and reduces cell cycle rate for most cells. For
cells that continue to rapidly cycle, few sustain high rates of transcription. Increased expression of topoisomerases, Top1 and
Top2a, which resolve supercoils and mediate conflicts in transcription and replication machinery, support the rare population of
hypertranscribing, hyperproliferating (HHCs). HHCs reprogram at near-deterministic rates to generate functionally mature cell type
such as induced motor neurons (iMNs).

79

Our study suggests that the biophysical properties of cells provide a formidable barrier to
cellular transitions and inhibit maturation of in vitro derived cell types. Transcription factor
overexpression induces high rates of transcription but with inherent antagonism to cellular
proliferation. This is because transduction with the reprogramming factors increases DNA
supercoiling, which leads to R-loop formation and DNA replication fork stalling. Importantly, we
show that DDRR rescues these genomic stresses by increasing the expression of Top1 and
Top2a, topoisomerases that relieve supercoiling and thereby reduce R-loops and restore
processive DNA replication. We demonstrate that by increasing the cell’s capacity to balance
tradeoffs during conversion we surmount maturity barriers.
Previous work identifying key cellular features enabling cellular reprogramming
examined the role of proliferation and found that fast-cycling, or hyperproliferating cells
preferentially reprogrammed into induced pluripotent stem cells (iPSCs) (Guo et al., 2014a).
However, because iPSCs are themselves fast-cycling, these studies could not determine if
hyperproliferation is generally required for transcription factor-mediated reprogramming or if it is
specifically required in iPSC generation because it aligns the cell division rate of the starting
somatic cells with that of pluripotent stem cells. Here, we show that hyperproliferation stimulates
reprogramming into multiple, post-mitotic lineages. It is perhaps surprising that
hyperproliferation stimulates reprogramming into post-mitotic lineages. However, previous work
in E. coli indicates that rapid cell cycle promotes state switching (Jaruszewicz et al., 2014).
Therefore, rapid proliferation may be a general motif to facilitate cell fate transitions, while
slower proliferation favors stable maintenance of cellular identity. Proliferation may facilitate the
transcriptional shift away from the fibroblast identity. For example, macrophages have been
shown to stabilize commitment to the myeloid lineage by lengthening their cell cycle as they exit
the progenitor phase, leading to the accumulation of highly stable PU.1 (Kueh et al., 2013).
Differences in cell cycle rate most significantly impact the concentrations of highly stable
80

molecular species. In the context of conversion, rapid replication may facilitate dilution of highly
stable mRNAs and proteins (e.g. collagens) that may limit full adoption of an alternative identity.
An important advance of our work is that hyperproliferation alone is not sufficient to
support privileged reprogramming. We find that coincident ability to maintain hypertranscription
is also required. Although previous studies suggested that p53 inhibition increased iPSC
reprogramming by preventing replicative senescence (Tapia and Scholer, 2010; Utikal et al.,
2009), our results suggest this alone is not sufficient to allow efficient reprogramming and the
full mechanism by which p53 inhibition promotes conversion is more complicated. p53 inhibition
helps to increase topoisomerase activity, mitigate antagonism between transcription and cell
proliferation, and support the expansion of HHCs.
While we have considered the synthetic transition of fibroblasts to motor neurons and
other post-mitotic cells, our findings raise questions about the transition of healthy cells to
pathological states such as cancer. The genetic mutants we have employed to promote
reprogramming are known to contribute to oncogenesis. The mechanisms that we have
uncovered, such as replication prior to differentiation and hypertranscription, are recognized
motifs in development (Percharde et al., 2017a; Percharde et al., 2017b). The overlap in
developmental and oncogenic processes suggest central mechanisms for promoting transitions
of cellular identity. Observing these stereotypical patterns of transitions in the synthetic context
of reprogramming strengthens the hypothesis that healthy cells co-opt developmental processes
in transit to their pathological state. Our data suggest that topological stress is a primary barrier
to cellular reprogramming. In the context of cancer, our data would suggest that topological
tangles act as genomic stabilizers of cellular identity, buffering cells against pathological
transitions. Cancerous cells commonly express high levels of topoisomerases and
topoisomerase inhibitors represent some of the front-line chemotherapy agents. The model of
cellular conversion may provide a useful system to screen molecules that effectively block
81

pathological transitions while preserving cells that maintain non-pathogenic states. Small
molecules and cocktails that block the development of the HHC state may illuminate new
therapeutic agents for targeting treatment to highly pathogenic cell states.

82

Chapter 4

Conclusion

83

1.1 Summary of Findings
Cellular reprogramming technologies have fundamentally reshaped our understanding of
basic biology. The biomedical research advancements made using patient-derived pluripotent
and transdifferentiated cells have accelerated treatments for patients with diseases like
amyotrophic lateral sclerosis (ALS) (McNeish et al, 2015; Wainger et al, 2014), led to discovery
of unknown disease pathologies (Shi et al, 2018), and even autologous cell therapies, as has
been used for treating blood cancers like leukemia and lymphoma (Copelan 2006). Despite
these achievements, there remained a significant gap in our understanding of the general
molecular and cellular changes required for fate transition that intertwines with how cells
maintain stable identities. Further, bulk characterization offered low resolution of the differences
between reprogrammed cells and bona fide counterparts. Using induced motor neuron
reprogramming as our testbed system, we discovered that the processes of proliferation and
transcription antagonize one another, leading to genomic stress and inefficient reprogramming.
Addition of a genetic and small molecule combination relieves this genomic stress and expands
a privileged population of reprogrammable cells through upregulation of topoisomerases. This
expansion correlates with increased reprogramming efficiency and, importantly, better
acquisition of motor neuron identity that will significantly improve in vitro disease modeling.
Extensive work had been done to elucidate barriers to iPSC reprogramming, but
progress remained limited in identifying ways to enhance direct lineage reprogramming. We
identified key behaviors and phenotypes of cells successfully able to undergo
transdifferentiation from fibroblast to induced motor neuron, behaviors that, when harnessed,
significantly improve the quantity and quality of the cells created. We discovered that the cellular
processes of proliferation and transcription are inherently antagonistic, and that most cells are
only capable of sustaining high levels of one but not the other. During reprogramming, the
transcription factors drive a state of hypertranscription while severely retarding the proliferative
abilities of the starting cells. This antagonism leads to genomic stress and manifests as R-loops,
84

torsion, and stalled replication processivity. We identified a cocktail of factors and small
molecules (DDRR) that endow a large proportion of cells to mitigate this antagonism and
sustain high levels of transcription and proliferation, termed hypertranscribing, hyperproliferating
cells (HHCs). These cells contribute to the majority of reprogramming events, at near-
deterministic rates.
Through single cell and bulk RNA sequencing, we discovered that upregulation of
topoisomerases enables reprogramming cells to sustain the high levels of transcription and
proliferation characteristic of the HHC state. Topoisomerases modulate DNA topology by
relieving torsion that occurs as a result of replication and transcription, as well as participate in
successful DNA decatenation. DDRR elevates topoisomerase expression to allow successful
resolution of genomic conflicts induced by the reprogramming factors, activities specifically
mediated by Top1 and Top2a. Furthermore, we find DDRR enhances molecular, physical, and
functional properties of the induced motor neurons. Additionally, the DDRR affect is a general
phenomenon applicable to a range of reprogramming strategies and starting cells.
Our findings support the idea that cells are inherently faced with conflict between
transcription and proliferation, especially during reprogramming. In mediating these conflicts
through elevated topoisomerases, we observe resolution of genomic stress, torsion, and stalled
DNA replication processivity in reprogramming cells. We expand the reprogrammable HHC
population through addition of DDRR, and significantly increase the quantitative and qualitative
features of the induced motor neurons. Importantly, this phenomenon is applicable across a
variety of reprogramming approaches, a key advancement in our understanding of general
roadblocks to lineage conversion.

1.2 Future Directions and Work
I. DDRR – alternative, transient approaches
85

a. Trp53 inhibition
The DDRR cocktail has proven effects on enhanced reprograming to a variety of somatic
cells, evidenced in this thesis. However, the DDRR cocktail introduces Trp53 and hRas mutants
via integrating viruses. Integrating oncogenes may confound the results of in vitro disease
modeling and preclude in vivo cell transplantation and therapy. Previous studies of iPSCs
generated with mutant or silenced Trp53 reveal genetic abnormalities and instabilities. Such
analyses have not yet been made in our direct reprogramming approach, though it will be
worthwhile to assess the integrity and potential adverse effects of inactive Trp53.
Transient methods of Trp53 inhibition may provide suitable alternatives to elicit p53DD-
like responses. Pifithrin is a well-characterized small molecule inhibitor that suppresses p53-
mediated activation of p21, Cyclin G, and Mdm2 (Komarov et al., 1999). Our group has shown
that Trp53 inhibition using this small molecule improves reprogramming from fibroblasts to
iMNs. We have also demonstrated that a doxycycline-inducible system regulating p53DD
enhances conversion when induced early in reprogramming, though at lower rates than that of
constitutive p53DD. Transient inactivation using an siRNA-mediated approach represents
another option. However, the variable results achieved via transient Trp53-inhibition suggest
there are likely other non-canonical binding and activities of p53DD that require further
exploration to better understand its molecular behaviors. Methods circumventing Trp53
inhibition altogether may be most favorable. As many diseases are characterized by either
protective or detrimental roles for Trp53, for instance in FUS and TDP-43-related ALS
pathologies (Vogt et al, 2018; De Santis et al, 2017; Turnquist et al, 2016), normal Trp53
expression and activity may be required to infer meaningful conclusions of in vitro phenotypes.

b. Mutant RAS
RAS genes were first identified as potent, transformative oncogenes in the 1980s, and,
similar to Trp53, mutants have since been associated with nearly every major cancer (Hobbs et
86

al, 2016; Cox and Der 2010; Malumbres and Barbacid 2003). Interestingly, roles for the RAS
family in cellular reprogramming contexts have only recently been explored. Using single cell
RNA sequencing, Kim and colleagues from the Wold lab discovered activation of RAS signaling
to be an early molecular event in iPSC reprogramming (Kim et al, 2015). Additionally, Ferreirós
and colleagues discovered that combination of the Yamanaka factors and early overexpression
of HrasV12 or other Ras mutants significantly increased the number of iPSC colonies generated
(Ferreirós et al, 2019). Interestingly, they report reduced expression of the starting fibroblast
“identity genes” and increased expression of pluripotency genes upon inclusion of HrasV12
early in their reprogramming system. They also report HrasV12 to have non-autonomous effects
on converting cultures, as co-culture with HrasV12-transduced MEFS or conditioned media also
yielded more iPSC colonies generated (Ferreirós et al, 2019).
Together, these data provide interesting mechanistic insights of the roles RAS plays in
fate transition during reprogramming, and strengthen the idea that there are conserved
roadblocks to both iPSC and somatic cell reprogramming. It is interesting to note that, like our
study presented here, RAS overexpression elicits molecular changes in cell identity that
promote iPSC reprogramming, for instance reduced expression of starting cell genes, though
the mechanisms behind such transcriptional changes remain elusive. It should also be noted
that these studies did not analyze long-term affects of mutant RAS on iPSC integrity and
stability, as it is likely they may suffer in one or both areas.
There are also reported roles for RAS in normal motor neuron biology, as many growth
factors initiate pathway signaling cascades that promote survival and homeostasis (Dervishi et
al, 2018). Therefore, the development of methods that avoid constitutive expression of mutant
RAS will facilitate the use of reprogrammed cells for applications in cell replacement therapy
and disease modeling. Alternative approaches such as doxycycline-induction to transiently
activate mutant RAS may still prove problematic due to the issues associated with genomic
insertion. Delivery of siRNAs, mRNAs or proteins represent transient, non-integrating options
87

that may enhance reprogramming without inducing an oncogenic footprint in the genome. Unlike
p53, there are currently no small molecule targets to activate RAS specifically, though it is
possible that targeting multiple components of its up- and downstream signaling pathways could
elicit similar response.

II. The HHC state, cell cycle/proliferation, and genome integrity
a. Dynamics and existence of the HHC state
Our work has uncovered that antagonism between the cellular processes of proliferation
and transcription serve as roadblocks to robust direct reprogramming. A privileged group of
HHCs in the iMN reprogramming population displays remarkable cellular processivity. Their
ability to catalyze the demands of simultaneous transcription and proliferation, sustaining high
rates of both, enables near-deterministic reprogramming of this population. However, though we
observe increased conversion to other cell types using the DDRR cocktail, presence of an HHC-
like population has yet to be analyzed in these other somatic reprogramming strategies (eg.
iDAs and iNs). Additionally, while inclusion of p53DD significantly enhanced human
reprogramming, DDRR did not, suggesting there are distinct features of the different species
systems. Expanding the HHC flow cytometry analysis to additional reprogramming schemes will
illuminate features of these different systems, cell types, and reprogramming cocktails.
Furthermore, HHC profiling may serve as a predictive tool to infer reprogramming outcomes in
new protocols and cellular contexts.
The interplay between proliferation and transcription driving the HHC state during
reprogramming is complex and there remain fundamental questions about when and what their
specific role(s) are in promoting these fate transitions. When do cells transition through this
state during reprogramming? How long do cells maintain this hypertranscribing,
hyperproliferating phenotype? Do cells require cell cycle re-entry prior to hypertranscribing, or is
hypertranscription induced first? While our observations of cell state are static, it is likely these
88

processes occur in a dynamic yet temporally ordered fashion. Our current methodology doesn’t
permit real-time tracking of transcriptional changes during reprogramming as we are able to with
proliferation history. Improvement of the technology, for instance, development of a reporter for
transcriptional activity or transcriptional state without cell fixation, would expand our ability to
dynamically track these processes using flow cytometry and time-lapse microscopy.
Additionally, single cell RNA sequencing of cells at more timepoints in the reprogramming
trajectory will lend deeper insight into the transcriptional states of converting cells in their
transitions.
A final intriguing use of this flow methodology would be metabolic analysis of the
reprogramming cells. Both Trp53 and RAS have roles in maintaining cellular metabolic
homeostasis (Hanahan and Weinberg 2011; Jones and Thompson 2009). Because our DDRR
cocktail introduces mutations to both, it is very likely that cellular metabolism is altered during
reprogramming. Utilizing our HHC methodology in combination with established assays to
measure mitochondrial biogenesis or activity (e.g. Abcam’s TMRE-Mitochondrial Membrane
Potential or MitoBiogenesis kits), we could assess metabolic function of reprogramming cells in
conjunction with their proliferation history and transcriptional state. Furthermore, because many
of these assays do not require cell fixation, cells could be collected post-FACS and re-plated for
conversion or analyzed in other downstream assays. Employing this flow cytometry approach
allows assessment of multiple components contributing to reprogramming in real time and
allows deeper investigation of the processes important for robust conversion.

b. Cell cycle and proliferation
The importance of cell cycle and proliferation in somatic cell reprogramming will
undoubtedly be surprising to many. Other groups have previously reported absence or little to
no proliferation in conversion to neuronal cells (Son et al., 2011; Vierbuchen et al., 2010). Our
findings demonstrate that cells rapidly transition through a dividing stage that accelerates and
89

improves conversion. This may be due to several reasons. DNA synthesis was reported to be
critical for cell fusion-mediated iPSC reprogramming (Tsubouchi et al., 2013). DNA synthesis is
an opportunity for epigenetic and DNA methylation remodeling, which are required for fate
transition. During DNA synthesis, the replication fork clears histones and transcription factors
from their binding sites (Yan et al, 2013). Via division-mediated dilution, cell cycle and
proliferation removes factors associated with the starting cell gene regulatory network (GRN).
Previous studies of directly converted cells (e.g. CellNet) revealed aberrant sustained
expression of the starting GRN in reprogrammed cells. Cell cycle re-entry and transition through
synthesis may accelerate reprogramming by facilitating the removal of genes associated with
the starting GRN.

c. Genome integrity
Finally, assessment of genomic, epigenetic, and functional integrity of reprogrammed
cells compared to bona fide counterparts is required for faithful in vitro modeling applications.
Previous studies observed unstable and genomically abnormal iPSC colonies generated under
Trp53 inhibition (Kawamura et al., 2009; Utikal et al., 2009). While iMNs themselves are not a
proliferative cell type, they may still be vulnerable to Trp53 inhibition- and proliferation-induced
genetic abnormalities incurred by silenced Trp53 during the early stages of reprogramming. We
observe the presence of stalled replication forks and reduction of cell cycle upon transduction of
the reprogramming factors, indicative of reprogramming cells’ response to stress and DNA
damage (Zegerman and Diffley 2009). Both key safeguards to genomic stability and integrity,
these phenomena serve as potent inhibitors of robust reprogramming. While DDRR removes
signs of genomic stress, it will be critical to perform karyotyping, immunocytochemistry for DNA
damage (e.g. γ-H2Ax), and other such assays on fully-reprogrammed DDRR cells to assess
their limitations for in vitro and/or in vivo modeling and cell therapy.
90

To identify roadblocks and understand the molecular principles of reprogramming, I have
identified a combination of genetic and chemical factors capable of eliciting a potent,
reproducible response in direct reprogramming. These conditions served as tools with which to
understand problems hindering reprogramming and driving transformation. This cocktail has
enabled us to uncover previously unreported blocks to reprogramming because of the scale at
which we can generate reprogramming and reprogrammed cells. However, future work will be
required to induce these processes without inclusion of oncogenes. Trp53 inhibition and
inclusion of mutant hRas in the DDRR cocktail may limit accurate in vitro modeling and cell
therapy. Given our discovered roles for cell cycle/proliferation and transcription in direct
reprogramming, generating an HHC-like, reprogramming permissive state in transient and non-
integrating ways represents an important objective. Screens of small molecules, compounds,
and/or mitogens that stimulate cell cycle and transcription either individually or in combination
may yield equally robust reprogramming results with reduced concerns of genomic integrity and
stability.

III. Other applications for DDRR and HHC-mediated systems
Beyond its use as a tool to understand basic principles and processes driving cellular
reprogramming, the DDRR cocktail and its effects on the converted cells hold much promise for
disease modeling. Our study reveals that addition of DDRR drives expression of a neuronal
gene regulatory network while reducing that of the fibroblast program, thus enabling more
complete morphological remodeling and acquisition of neuron function compared to non-DDRR
cells. However, we have not yet assessed these cells for disease relevance in vitro.
Measurement of survival and excitotoxicity under ALS-like disease conditions (e.g. excess
glutamate) are two metrics with which to analyze these cells. We expect patient iMNs to
demonstrate greater susceptibility to insults and display exacerbated disease phenotypes
compared to controls. For example patient-derived iMNs would be expected to display
91

significantly reduced survival, if DDRR-iMNs are indeed more similar to bona fide motor
neurons. While p53 expression is quite similar between DDRR-iMNs and primary embryonic
motor neurons, re-introduction of wildtype Trp53 might be essential to draw disease-relevant
conclusions, as Trp53-dependent apoptosis is identified as one cause of motor neuron death in
ALS (Vogt et al, 2018; Ranganathan and Bowser 2010).
Additionally, the HHC state that we identified using DDRR may exist “naturally” in the
development and progression of malignancies like cancer. Many cancers are characterized by
mutations in Trp53 and RAS, display uncontrolled cell cycle, and aberrant gene expression (Lee
and Young 2013; Hanahan and Weinberg 2011). Thus, it is possible that an HHC-like transition
state might be identified using our flow cytometry methodology. Interesting application of this
technique would first be to screen tumors from various tissues (e.g. breast, colon, blood),
assessing if an HHC-like state exists in a(ny) particular tumor type. It would be an equally
interesting assessment if cells derived from these different tumors exhibit differences in cell
cycle, transcription, or both properties as measured with this assay.
Furthermore, our flow cytometry methodology could be harnessed to play predictive
roles in cancer progression and outcomes for patients. Over the course of a person’s disease,
patients undergo routine physician visits, check-ups, and blood work. If such blood samples
could be collected, flow sorted, and assessed for cell cycle and/or transcriptional changes since
previous visits, it is possible this technique could be employed as a monitor of disease treatment
and outcomes, and possibly a predictor of metastasis. Metastasis is a multi-step cascade
resulting from genetic and epigenetic changes, response to changing environmental stimuli, and
massive transcriptional changes (Seyfried and Huysentruyt 2013; Hanahan and Weinberg
2011). This cellular event leads to most patient deaths (Seyfried and Huysentruyt 2013). If cells
prior to, during, or after metastasis could be analyzed using our “HHC-like” flow cytometry
assay, and potentially collected for further molecular characterization after, greater
understanding of pathogenic cancer progression might be gained.
92

Beyond employing this flow cytometry assay for prediction of disease progression,
another intriguing idea would be its use in predicting patient response to both established and
novel treatment regimens. Topoisomerases I and II-α are potent anti-cancer targets in many
malignancies including breast, esophageal, oral, bladder, and prostate cancers (Delgado et al,
2018; Chen et al, 2015). Most drugs are inhibitors or poisons that trap a covalent
topoisomerase-drug-DNA complex that is toxic to cells (Delgado et al, 2018). We observe
changes in the HHC profile of reprogramming cells under conditions of topoisomerase inhibition
(both genetic and small molecule manipulation). Similarly, we might also expect that patient
cancer cells treated with such topoisomerase poisons might exhibit changed behaviors in their
HHC profile. Unfortunately, as is the outcome for many patients, treatment with topoisomerase
poisons often leads to secondary cancer and/or heart damage (Delgado et al, 2018). Thus, this
methodology could potentially predict both susceptibility and, eventual, resistance to these types
of treatments. Further, predicted response to alternate or novel treatments could also be
assessed with this methodology.
In this study, we identify a privileged group of the reprogramming population with
remarkable processing capabilities. These cells are endowed with the ability to mediate
simultaneous high levels of transcription and proliferation, sustained by high topoisomerase
expression, and they contribute to most successful reprogramming events. Expansion of this
privileged population is likely to promote reprogramming to other lineages. These findings are
generally applicable across reprogramming strategies, highlighting a major advance in our
understanding of molecular underpinnings to reprogramming. Additionally, this novel
methodology may be used to predict reprogramming outcomes and employed in cancer biology
to predict disease progression and treatment prognoses.

93

Appendix I: Materials and Methods

EXPERIMENTAL MODELS AND SUBJECT DETAILS
Cell Lines and Tissue Culture
HEK293, Plat-E, mouse embryonic fibroblasts, and primary human fibroblasts were cultured in
DMEM supplemented with 10% FBS at 37degC with 5% CO
2
. Mouse tailtip fibroblasts were
cultured in DMEM supplemented with 20% FBS at 37degC with 5% CO
2
. The following are the
sex of primary human fibroblasts used in this study: Foreskin fibroblasts (BJ) – male, adult
fibroblasts (GM05116) – female.

METHOD DETAILS
Isolating fibroblasts and cell culture
Hb9::GFP-transgenic mice (Jackson Laboratories) were mated with C57Bl/6 mice (Jackson
Laboratories) and MEFs were harvested from Hb9::GFP E12.5 – E13.5 embryos under a
dissection microscope (Nikon SMZ 1500). To eliminate contaminating neurons, the head and
spinal cord were removed. The fibroblasts were passaged at least once before being used for
experiments. Neonatal mesenchymal cells were harvested by collagenase I digestion of
hindlimb muscle as described (30). Human adult fibroblasts were obtained from Coriell
(GM05116). Irradiated MEFs were obtained from GlobalStem (Cat. No: GSC-6001G).

Plasmid construction
Retroviral and lentiviral plasmids were constructed by Gateway and Gibson cloning into entry
vectors pDONR221 or pENTR4-DS. Entry clones were recombined into destination vectors via
LR reaction (42) into the pMXS-DEST (retro) FUWO-tetO-DEST (lenti). For complete list of
plasmids see Table S1.

Viral transduction and iMN reprogramming
Retroviral transduction and iMN reprogramming from MEFs was performed as described (7).
Briefly, retroviral transduction was performed using Plat-E retroviral packaging cells (Cell
Biolabs, Inc., RV-101). MEFs were transduced twice with Ascl1, Brn2, Isl1, Lhx3, Myt1l, and
Ngn2 or a polycistronic NIL construct at 48 and 72 hours after transfection. For human iMN
experiments, retroviruses were generated in 293T cells and co-transfected with pLK and
pHDMG packaging plasmids, and NEUROD1 was added to the reprogramming cocktail. In
experiments in which iMN formation was quantified by microscopy or iMNs were functionally
evaluated using electrophysiology, we added mixed glia isolated from P2 ICR mouse pups to
infected fibroblasts 2 days after transduction. It is important to note that in other experimens,
including those in which the number of hyperproliferating or hypertranscribing cells were
analyzed, CFSE labeling was used, FACS sorting was used prior to iMN formation, or FACS
sorting was used to quantify the number of iMNs out of total viable cells at the end of
reprogramming, we omitted glia from the glia media added to the reprogramming cultures in
order to avoid confounding our results. The day after glia media addition, medium was switched
to complete N3 neuronal medium (DMEM/F-12 (VWR) with N2 and B27 (ThermoFisher)
supplements and 1% glutamax (ThermoFisher)). Medium was supplemented with
neurotrophics, GDNF, BDNF, and CNTF (R&D Systems) and FGF-Basic (Peprotech), each at
10 ng/ml. When included for reprogramming, RepSox (Selleck) was added to N3 media at a
final concentration of 7.5uM.

iHC reprogramming
Retroviral transduction of MEFs was performed using Plat-E retroviral packaging cells (Cell
Biolabs, Inc., RV-101). Atoh1::nGFP MEFs were transduced with Atoh1, Brn3c, and Gfi1 at 48
94

and 72 hours post-transfection. Two days after transduction, media was changed to induced
hair cell media (DMEM/F-12 + N2 + B27) supplemented with FGF-Basic (Peprotech) and HB-
EGF (Peprotech) and final concentrations of 2.5 ng/uL and 5 ng/uL, respectively.

iDAN and iN reprogramming
Retroviral transduction of non-transgenic or Tau::GFP MEFs was performed using Plat-E. MEFs
were transduced with Ascl1, Brn2, Myt1l, FoxA2, and Lmx1a for iDAN reprogramming (28).
Alternately, MEFs were transduced with Ascl1, Brn2, and Myt1l for iN reprogramming (43). Two
days after transduction, mixed glia isolated from P2 ICR mouse pups were added to converting
cultures. The next day, complete neuronal N3 media with neurotrophic factors (FGF, GDNF,
CNTF, and BDNF at each at 10 ng/ml) was added to converting cultures.

shRNA-mediated knockdown of Mbd3, Gatad2a, Top1, Top2a
Mbd3, Gatad2a, Chd4, Top1 and Top2a shRNA lentiviral constructs were obtained from Sigma.
Lentiviruses were generated in 293T cells and packaged via co-transfection with pPax2 as well
as VSVG envelope using PEI transfection reagent. Hb9::GFP+ MEFs were co-transduced with
scrambled, Mbd3, Gatad2a, Chd4, Top1, or Top2a shRNAs during the second day of PlatE
transduction with the motor neuron factors.

qRT-PCR quantification of shRNA-mediated Mbd3, Gatad2a, Top1, Top2a knockdown
For experiments measuring knockdown of Mbd3, Gatad2a, Top1 or Top2a, cells were collected
4 days after transduction with motor neuron factors and shRNAs. RNA isolation was performed
using TRIzol LS reagent according to the manufacturer’s instructions (ThermoFisher Cat. No:
10296010). Reverse transcription of purified RNA was performed using random hexamer
primers and New England Biolabs protoscript first strand cDNA synthesis (VWR Cat. No:
101640-908). qPCR was performed using primers for Mbd3, Gatad2a, Top1 and Top2a and
iTaq universal SYBR green (Bio-Rad Cat. No: 1725125). The following primer sequences for
endogenous Mbd3, Gatad2a, Top1 and Top2a genes were used:
Mbd3 (5’ – TCCAGGTCTCAGTGCAGGGA and 5’ – TGACTTCCTGGTGGGCTGCT), Gatad2a
(5’ – AATAACGGGTCCTCACTACAG and 5’ – GTATTCTCGCTGTCGATCCA), Top1 (5’ –
TCTCTAGTCCGCCACGAATTA and 5’ – CATCTCGAAGCCTCTTCAATGG) and Top2a (5’ –
GCTCCTCGAGCCAAATCTGA and 5’ – CTACCTATAAAACTGGCTCCGT).

Quantification of Conversion Yield
All reprogrammed cultures were imaged using either the Biostation CT or Molecular Devices
Image Express and manually quantified using Fiji. Yield of converted cells was calculated as the
number of cells with the proper morphology and marker(s) on the final day of conversion over
the number of cells seeded for conversion. For iMNs, the number of Hb9::GFP+ MEF- or
explant-derived cells with neuronal morphologies was quantified between 14-17 dpi. In the
single cell RNA sequencing experiments in Fig. 4, iMNs were collected at 14 dpi. For iDANS
and iNs, Tau::GFP+ or Map2+ cells with neuronal morphologies were manually quantified
between 17dpi. For iHCs, the number of Atoh1::nGFP+ cells was used to quantify percent iHCs
between 17 dpi. For iMNs, adult human fibroblast-derived Map2+/DsRed or p53DD-T2A-RFP+
double-positive cells with neuronal morphologies were manually quantified between 35 dpi.

Whole cell patch clamp electrophysiology
Whole cell membrane potential and current recordings in voltage- and current-clamp
configurations were made using an EPC9 patch clamp amplifier controlled with PatchMaster
software (HEKA Electronics). Voltage-and current-clamp data was acquired at 50kHz and
20kHz, respectively, with a 2.9kHz low-pass Bessel filter. For experiments, culture media was
exchanged with warm extracellular solution consisting of (in mM): 140 NaCl, 2.8 KCl, 10
95

HEPES, 1 MgCl
2
, 2 CaCl
2
, and 10 glucose, with pH adjusted to 7.3 and osmolarity adjusted to
310 mOsm. Glass patch pipettes were pulled on a Narishige PC-10 puller and polished to 5-7
MΩ resistance. Pipettes were also coated with Sylgard 184 (Dow Corning) to reduce pipette
capacitance. The pipette solution consisted of (in mM): 130 K-gluconate, 2 KCl, 1CaCl
2
, 4
MgATP, 0.3 GTP, 8 phosphocreatine, 10 HEPES, 11 EGTA, adjusted to pH 7.25 and 300
mOsm. Pipettes were sealed to cells in GΩ-resistance whole cell configuration, with access
resistances typically between 10-20 MΩ, and leakage currents less than 100 pA. Capacitance
transients were compensated automatically through software control. For current-voltage (IV)
curves, cells were held in voltage clamp configuration at -70 mV and stepped through
depolarizing voltages from -70 to 100 mV. A P/4 algorithm was used to subtract leakage
currents from the traces. For action potential measurements, cells were held in current clamp
configuration at 0 pA and action potentials were evoked by injecting positive depolarizing
currents for 1 s. SFA ratios were calculated as the time interval between the first two APs
evoked to the time interval between the last two APs evoked (35) using the lowest current
injection that generated APs. Measurements were taken at room temperature (approximately
20-25°C). Data was analyzed and plotted in Igor Pro (WaveMetrics).

CFSE cell labeling to measure cellular proliferation
One day after retroviral infection, fibroblasts were labeled with CellTrace CFSE Cell Proliferation
Kit GFP (Invitrogen, Cat. No: C34554) or Far Red (Invitrogen, Cat. No: C34572) at a final
concentration of 10uM. Briefly, media was removed, CFSE added to the cells, and incubated at
37
o
C for 30 minutes. After incubation cells were washed once with PBS, then replaced with
fresh media. Generally, cells were harvested for FACS sorting 72 hrs following labeling without
addition of glial cultures. Fast cycling cells were determined by examining the distribution from
cells infected with reprogramming factors. During reprogramming, the dimmest 15% of cells in
6F conditions at 4 dpi were used to set the gate for fast-cycling cells. Cells with lower CFSE
intensity were gated as fast-cycling. For all replating experiments, gates were set using the
dimmest 15% of cells in 6F conditions. Generally, we found that the absolute CFSE intensity of
the the fast-cycling cells was 8-fold lower than mean CFSE of the entire population, indicating
three more division over 72 hrs. With a putative average 24 hr cell cycle, cells divide 3 times
over 72 hours, while fast cells divide 6 times or more, suggesting a <12 hr cell cycle.

Cleaved caspase-3, mKi67, RNA PolII immunolabeling for FACS sorting
One day after addition of N3 media (and without addition of glial cultures), cleaved caspase-3
and mKi67 labeling and subsequent FACS sorting for analysis was performed. Cells were
trypsinized with 0.25% Trypsin-EDTA (Genesee Scientific), resuspended, and then spun down.
Cells were then fixed in 4% paraformaldehyde for 15 minutes at room temperature in the dark.
Cells were washed with PBS, pelleted, and permeabilized with 0.5% Triton X-100 for 15 minutes
at room temperature in the dark. After permeabilization, cells were blocked in 3% FBS in PBS
block solution for 30 minutes at room temperature in the dark with rotation. After being spun
down, cells were then incubated in primary antibodies (1:200 dilution in 3% block solution) for
45 minutes at room temperature with rotation. Cells were washed with block solution, spun
down, and then incubated in secondary antibodies (1:200 dilution in 3% block solution) for 30
minutes at room temperature in the dark with rotation. Cells were then washed in block solution,
spun down, and resuspended in 150-200µL PBS containing DAPI (100x) prior to FACS sorting
and analysis. The following primary antibodies were used: rabbit anti-cleaved caspase-3
(abcam Cat No: ab13847), rabbit anti-Ki67 (GeneTex GTX16667 Cat. No: 89351-224) and
rabbit anti-RNA polymerase II CTD repeat YSPTSPS (phospho S2) (abcam Cat No: ab5095).

96

DNA Fiber Assay
One day after addition of N3 media (and without glial cultures), cells were pulse-labeled with IdU
(50µM) and CIdU (100µM final concentration) for 20 and 30 minutes, respectively at 37
o
C. Cells
were washed with PBS, trypsinized with 0.25% Trypsin-EDTA and spun down. Cells were
resuspended in 50µL, put on ice, and resuspended to a concentration of 400 cells/µL in PBS.
Three, 2 µL aliquots of each cell sample was spotted onto silane-coated slides and tilted to
allow the cells to streak across the slide lengthwise. The cell preparations were dried for ~15-20
minutes and then lysed (1M Tris pH 7.4 + 0.5M EDTA + 10% SDS in ddH
2
O). DNA spreads
were air dried for 12 hours at room temperature and then fixed in methanol: acetic acid (3:1) for
2 minutes at room temperature. Slides were dried overnight at room temperature protected from
light and then stored at -20
o
C for at least 24 hours before antibody labeling. The fiber spreads
were treated with 2.5M HCl for 30 minutes and then blocked in 5% BSA for 30 minutes in
“humidified chamber.” Fiber spreads were incubated with mouse α-BrdU (1:500, to detect IdU)
and rat α-BrdU (1:500, to detect CIdU) primary antibodies for 1 hour at room temperature and
then incubated for 15 minutes in stringency buffer (1M Tris pH 7.4 + 5M NaCl + 10% Tween +
10% NP40 in ddH
2
O). Slides were blocked again for 30 minutes and then incubated with rabbit
α-mouse 594 (1:1000) and chicken α-rat 488 (1:750) secondary antibodies for 30 minutes at
room temperature. After washes in 0.1% Tween in PBS, slides were blocked again at room
temperature and then incubated with goat α-rabbit 594 (1:1000) and goat α-chicken 488 (1:750)
tertiary antibodies for 30 minutes at room temperature. After a wash with 0.1% Tween in PBS
followed by PBS washes, glass coverslips were mounted onto the silane slides using Antifade.
The following primary antibodies were used: Monoclonal anti-IdU antibody produced in mouse
(Sigma Cat. No: SAB3701448-100UG) and anti-BrdU antibody [BU1/75 (ICR1)] detects CIdU
(abcam Cat. No: ab6326).

DNA-RNA Hybrid R-loop Staining and RNase Treatment
One day after addition of N3 media (and without addition of glial cultures), cells were fixed in 4%
paraformaldehyde for one hour at 4°C in the dark. Cells were then permeabilized in 0.2% Triton
X-100 in PBS for one hour in the dark. Coverslips were then split into two and 1 half was used
for RNase H treatment. Briefly, coverslip halves were treated with 250µL of 1X buffer + 2µL
RNase H at 37°C for 36 hours prior to proceeding with antibody labeling. Then, all coverslips
were incubated in 2% BSA in PBS block solution for 1 hour at room temperature. Cells were
then incubated in primary antibodies (1:1000 nucleolin to label nucleoli + 1:200 S9.6 to label
DNA-RNA R-loops in 2% block solution) for 1 hour at room temperature followed by two PBS
washes. Then, cells were stained with secondary antibodies (1:500 dilution in 2% block solution)
for 2 hours at room temperature in the dark. After two PBS washes, cells were stained with
Hoescht (1:1000) for 10 minutes at room temperature in the dark, washed again, and mounted
onto glass slides using ImmuMount (ThermoFisher). The following primary antibodies were
used: DNA-RNA R-loop S9.6 antibody (Kerafast Cat. No: ENH001), and nucleolin (abcam Cat.
No: ab22758).

Dot Blot for R-loop Analysis
For each sample, genomic DNA was purified from one well of a 6-well dish using the DNeasy
Kit from Qiagen. Samples were eluted using 150 uls of elution buffer. Samples were then
ethanol precipitated and resuspended in 7-10 uls of water. 1 ul of each sample was spotted onto
a positively charged nylon membrane (GE Healtcare) and dried for 10 minutes before cross-
linking by exposure to 254 nm light for 3 minutes. Membranes were then blocked with 5%
milk/TBST (20 mM Tris-Hcl, 150 mM NaCl, 0.05% Tween 20, pH 7.5) for 1 hr at room
temperature. If RNAse H treatment was performed, the membrane was incubated in 11 mls of
1x RNAse H buffer with 44 uls of RNAse H (New England Biolabs, Cat. No: M0297L) at 37
degrees Celsius for 36 hours. Membranes were then washed twice with 5% milk/TBST. S9.6
97

(1:1000, Kerafast Cat. No: ENH001) or single-stranded DNA (1:10,000, Millipore Cat. No:
MAB3868) antibodies were added in 1% BSA/TBST and incubated at 4 degrees Celsius
overnight. For DNA that was going to be probed with the single-stranded DNA antibody,
samples were heat denatured at 95 degrees Celsius for 10 minutes and snap-cooled on ice for
2 minutes prior to spotting on the membrane. Membranes were then washed twice with TBST
and probed with an anti-mouse horseradish peroxidase-linked antibody (1:5,000, Cell Signaling
Cat. No: 7076S) for one hour at room temperature. Membranes were exposed using either the
Amersham ECL Western Blotting Detection Kit (GE Healthcare, Cat. No: RPN2108) or the
SuperSignal West Femto Maximum Sensitivity Substrate (Thermofisher Scientific Cat. No:
34577).

EU Incorporation for FACS Sorting
One day after addition of N3 media (and without addition of glial cultures), EU incorporation
assays were performed according to manufacturer’s instructions modified for FACS sorting
(Invitrogen, Cat. No: C10330). Cells were incubated with 1mM EU for 1 hour at 37
o
C, washed
once with PBS, dissociated with 0.25% Trypsin-EDTA (Genesee Scientific), resuspended, and
then spun down. Cells were fixed with 3.7% PFA for 15 minutes at room temperature in the
dark. Cells were then washed with PBS, pelleted, and then permeabilized with 0.5% Triton X-
100 for 15 minutes at room temperature in the dark. After permeabilization, Click-iT reaction mix
was added to each sample proceeded by incubation for 30 minutes at room temperature with
rotation in the dark. Cells were then washed with Click-iT Reaction Rinse Buffer (Component F),
pelleted, washed once with PBS, and then pelleted again. Cells were resuspended in N3
neuronal media containing DAPI (100x) and then FACS sorted.

EdU Incorporation for FACS Sorting
One day after addition of N3 media (and always omitting glia), EdU incorporation assays were
performed according to manufacturer’s instructions (Invitrogen, Cat. No: C10424). Cells were
incubated with 10uM EdU for 1 hour at 37
o
C, washed once with PBS, dissociated with 0.25%
Trypsin-EDTA (Genesee Scientific), resuspended, and spun down. Cells were fixed in 100uL
Click-iT fixative (Component D) and incubated for 15 minutes at room temperature in the dark.
Cells were washed with 1% BSA in PBS, pelleted, and resuspended in 100uL of 1X Click-iT
saponin-based permeabilization and wash reagent (Component E) for 15 minutes at room
temperature in the dark. Cells were then incubated with Click-iT reaction cocktail for 30 minutes
at room temperature in the dark with shaking. Cells were washed with 1X Click-iT saponin
based permeabilization and wash reagent (Component E) and then pelleted. Cells were
resuspended in N3 neuronal media or PBS containing DAPI (100x) for FACS sorting.

Flow cytometry and FACS analysis
Cells were harvested as previously described for each cell type with trypsin processing for
MEFs and 4 dpi samples and DNaseI/Papain (Worthington Biochemical) processing for 8 dpi
and iMN samples. Sorting of cells for analysis or collection was performed on an Aria I or Aria II
(BD). Live single cells were identified by SSC and FSC gating and as DAPI negative. For fixed
cells processed for CFSE-EU assays, cells were identified by SSC and FSC gating and DAPI
staining was used to identify positive stained cells. Non-fluorescent controls were included to
identify fluorescent populations. For multiple fluorophore experiments, single-labeled cell
populations were included to allow proper compensation (e.g. EU-only, EdU-only, CFSE-only
controls, primary antibody-only controls, non-labeled cells for CFSE-EU/EdU assays). Sample
compensation was performed prior to other analyses. For all CFSE-EU assays, fast cycling cells
were determined by gating the dimmest 15% of cells in 6F conditions at 4 dpi. Cells with lower
CFSE intensity were gated as fast-cycling. From the fast-cycling population of cells,
hypertranscribing cells were identified as the top 50% of the 6F only conditions.
98

Alpha-amanitin treatment for FACS and conversion
Converting cultures were treated with complete N3 media supplemented with water control or α-
amanitin (1 µg/mL) at 3 dpi and transcription rate was measured by flow cytometry at 4 dpi
using EU incorporation. For iMN conversion, cultures were treated complete N3 media
supplemented with water control or α-amanitin (1 µg/mL) from 3-7 dpi, at which point cultures
were maintained in complete N3 without water control or α-amanitin until 14-17 dpi.

Aphidicolin, camptothecin, doxorubicin treatment for FACS and conversion
Converting cultures were treated with complete N3 media supplemented with DMSO control,
aphidicolin (1 µM) or doxorubicin (0.25 µM) at 3 dpi for 18 hours. Transcription rate was
measured by flow cytometry using EU incorporation or DNA synthesis rate was measured by
flow cytometry using EdU incorporation at 4 dpi. For iMN conversion, cultures were treated with
complete N3 media supplemented with DMSO control, aphidicolin (1 µM), camptothecin (1 µM),
or doxorubicin (0.25µM) at 3 dpi for 18 hours, at which point cultures were maintained in
complete N3 media without DMSO or small molecules until 14-17 dpi.

Quantification of anaphase-telophase chromatin bridges, micronuclei
For quantification of anaphase-telophase micronuclei or bridges, converting cultures grown on
plastic coverslips were fixed with 4% paraformaldehyde at 2 or 4 dpi, respectively. Cells were
then stained with DAPI (1:1000) for 10 minutes at room temperature in the dark. After mounting
onto glass slides using ImmuMount (Thermo Scientific), cells were acquired on the Zeiss LSM
800 confocal microscope using a 40X objective. Anaphase-telophases with chromatin bridges or
micronuclei were identified based on their DAPI profile as has been previously reported (Slattery
et al 2012; Broderick, et al 2015; Dykhuizen, et al 2013; Kotsandis, et al 2016). Anaphase-
telophase cells with one or more non-integrated DNA fragments were determined as having
micronuclei. Anaphase-telophase cells with one or more DNA strands between the
separating/separated daughter cells were determined as having a bridge. The number of
anaphase-telophase mitotic cells with chromatin bridges or micronuclei over all anaphase-
telophases was recorded.

Quantification of multipolar neurons
Converted iMN cultures were imaged using the Molecular Devices Image Express between
DPIs 14-17 and manually quantified using Fiji. Cells expressing the proper marker(s), neuronal
morphology, and at least 3 or more neurite processes were included in the quantification of
percent multipolar neurons.

RNA Polymerase II + CFSE + EdU labeling for FACS analysis
For CFSE labeling, one day after retroviral infection, fibroblasts were labeled with CellTrace
CFSE Cell Proliferation Kit Far Red (Invitrogen, Cat. No: C34572) at a final concentration of
10µM as described above. For EdU labeling, cells were then incubated with EdU one day after
addition of N3 media (without addition of glial cultures) also as described above. After a 30
minute incubation with Click-iT reaction mixture (using Alexa Fluor 594) followed by the wash
with 1X Click-iT saponin based permeabilization and wash reagent (Component E), cells were
then incubated in 3% FBS in PBS block solution for 30 minutes at room temperature with
shaking. After spinning down and resuspending, cells were then incubated with primary
antibody (1:200 dilution in 3% block solution) for 45 minutes at room temperature with rotation.
Cells were washed with block solution, spun down, and incubated in secondary antibody (1:200
dilution of Alexa Fluor 488 in 3% block solution) for 30 minutes at room temperature in the dark
with rotation. Cells were then washed in block solution, spun down, and resuspended in 150-
200µL PBS containing DAPI (100x) prior to FACS sorting and analysis. The following primary
99

antibody was used: rabbit anti-RNA polymerase II CTD repeat YSPTSPS (phospho S2) (abcam
Cat No: ab5095). The following secondary antibody was used: donkey anti-rabbit IgG highly
cross-adsorbed secondary antibody, Alexa Fluor 488 (Thermo Fisher Cat. No: A-21206).

Genomic analysis of viral integrations
To analyze integration of viral constructs into cells during reprogramming, we collected three
replicates of 40,000 cells at 4 dpi by trypsinization. To gather cells based on Isl1-GFP
expression, populations were collected via FACS for high and low Isl1-GFP as well as gated for
CFSE intensity (e.g. CFSE-Low for hyperproliferative populations, CFSE-High for slowly dividing
cells). Following isolation, cells were pelleted and responded in Direct Lysis Buffer (Viagen) with
1 mg/mL Proteinase K (Viagen) and processed per manufacturer’s instructions. Briefly, cell
solutions were incubated at 55
o
C for 45 minutes, followed by 85
o
C for 1 hour to inactivate
Proteinase K. Cell extracts were diluted 1:3 in water. Relative number of integrations were
analyzed by qPCR with iTaq Universal SYBR Green Supermix (Biorad) and primers specific for
the native MALAT1 genomic region (primers: MALAT1-FWD: GGTTTCTCTCTCCCCTCCCT,
MALAT1-REV: TTCGCATACGTGTGTCTGCT), Isl1-GFP transgene (primers: Isl1-GFP-FWD:
AACAGCATGGTAGCCAGTCC, Isl1-GFP-REV: GCTGAACTTGTGGCCGTTTA ), and Ngn2-
F2A-Isl1 transgene (primers: Ngn2-F2A-Isl1 -FWD: GAGAAGCATCGTTATGCGCC, Ngn2-F2A-
Isl1 -REV: TCCCATTGGACCTGGATTGC). Relative integrations were determined by
calculating each samples delta C
T
for the transgenes relative to the native MALAT1 region and
calculating 2 raised to the negative delta C
T
.

Single cell qPCR
Single iMNs of different morphologies were identified and isolated using an inverted microscope
equipped with micromanipulator and micropipette. Cells were collected directly into 5 uL of
CellsDirect 2XBuffer (Cells Direct One-step qRT-PCR kit, Thermo). Cells were processed using
the manufacturer’s protocol for reverse transcription (RT) and specific target amplification
(STA). cDNA was synthesized and pre-amplified from single-cell lysate. Single-cell qPCR was
performed using the Fluidigm BioMark HD system on amplified cDNA templates, with primer
and SsoFast EvaGreen supermix. Primers were validated in-house to yield efficient PCR
amplification. A matrix of C
T
s and quality metrics was generated and extracted for each cell.
Cells and genes were excluded for low-quality scores. In all, expression across 17 genes for
fibroblast and neuronal markers was performed for 25 fibroblast-like cells and 36 neuronal cells.
A profile of expression was generated for each cell using delta C
T
s normalized across total
expression of the panel of genes. A heatmap was generated to visualize the profile of
expression across the different gene sets and morphologies.

Single cell RNA-sequencing
Cells were harvested at different points in conversion. Specific populations were identified and
collected via FACS and all cells were sorted to obtain viable single cell suspensions. Fast-
cycling cells were identified by low CFSE intensity at 4 dpi. Hb9::GFP+ at 8 dpi and 14 dpi cells
were identified relative to Hb9::GFP negative control. Cell suspensions were loaded into a chip
and processed with the Chromium Single Cell Controller (10x Genomics). To generate single-
cell gel beads in emulsion (GEMs), individual populations were assigned individual libraries
using Single Cell 3′ Library and Gel Bead Kit V2 (10x Genomics, 120237). For each population,
the target population size was between 1000-1500 cells. Cell suspensions were calibrated to
capture the target number of cells. Fewer cells were captured at 8 dpi due to limited Hb9::GFP+
cells in 6F condition. RNA from lysed cells was barcoded through reverse transcription in
individual GEMs. Barcoded cDNAs were pooled and cleanup by using DynaBeads® MyOne
Silane Beads (Invitrogen, 37002D). Single-cell RNA-seq libraries were prepared using Single
Cell 3′ Library Gel Bead Kit V2 (10x Genomics, 120237). Sequencing was performed with using
100

multiple NextSeq 500/550 High Output Kit v2 on an Illumina NextSeq with pair end 150bp
(PE150). On average, sequencing generated 100-200K reads per cell on average over the
libraries.

Single Cell RNA-seq Analysis
Cluster analysis via Seurat
Analysis of embryonic motor neurons and induced motor neurons from various conditions was
performed using Seurat 2.2. Following alignment and processing in CellRanger, variable genes
were identified using FindVariableGenes. Clustering was performed using FindClusters based
on the number of PCs identified through the PCElbow plot function. Cluster markers were
identified for each cluster using the FindMarkers function.

Cluster analysis and pseudotemporal ordering via Monocle
We used the cellranger count pipeline (10x Genomics) to align and quantify single cell
expression for each library. Software details available (Table S1). Samples were combined into
a single matrix via the aggr pipeline and normalized by read depth across the libraries.
scRNAseq datasets were imported into Monocle using cellrangerRkit in R to create a
cellDataSet. Data were normalized using estimateSizeFactors. Outliers were removed based on
variance using estimateDispersion to remove 108 outlier cells. Clustering was performed using
10300 genes with high dispersion and mean gene expression >= 0.1 on the first 10 PCs.
Clusters of varying number were examined and clustering via 3 primary clusters was chosen to
capture different populations (e. g. MEFs, converting cells, and iMNs). Pseudotemporal ordering
was performed using identified clusters. Pseudotemporal ordering was rooted in the identified
iMN endpoint. To generate the pseudotime trajectory corresponding to reprogramming time,
pseudotime was reversed to generate trajectory spanning MEFs at t=0 and iMNs at t=30 (end
time). All subsequent pseudotime analyses were performed with the resulting cellDataSet.

Bulk RNA-sequencing and analysis
For cultures at 17 dpi, cells were harvest by DNase/papain (Worthington Chemical) treatment to
dissociate cells. Cells were washed three times in DMEM-F12 media and resuspended in N3
neuronal media for sorting. Cells in replicates of 50K were collected based on gates set to
identify viable, single Hb9::GFP+ cells. Following sorting, cells were spun and resuspended in
100 uL RLT buffer from the RNAeasy micro kit (Qiagen). RNA in RLT and RNA extracted via
RNAeasy kit were sequenced by Amaryllis (Emryville, CA) via single-end sequencing to
generate 30M reads per sample. Additionally, Fastq files for previously acquired data for MEFs,
embryonic motor neurons (embMNs), iPSC-derived MNs, and ES-derived MNs samples in
duplicate were acquired and processed with newly generated data sets. Sequencing reads from
triplicate or more replicates were trimmed and aligned to mm10 reference transcriptome with
STAR aligner 2.5.3a. Gene counts quantified using annotation model (Partek E/M). Differentially
expressed genes were identified using DEseq2 for with genome wide false discovery rate (FDR)
of less than 0.05 and log2 fold change greater than 1. Comparison of MEFs with all MN samples
generated 1186 DEGs. Heatmap analysis of MEFs and iMNs from different conditions was
generated using this DEG set. Direct comparison of iMNs from 6F and DDRR conditions
generated 756 DEGs. Metascape analysis (www.metascape.org) was used to generate GO
terms for up and downregulated genes.

Cell number normalized (CNN) RNA-sequencing and analysis
For cultures at 4 dpi, cells were harvest by trypsin treatment to dissociate cells. Cells were
washed three times in DMEM-F12 media and resuspended in N3 neuronal media for sorting.
Cells in replicates of 50K were collected based on gates set to identify viable, fast-cycling cells
(e.g. CFSE-lo) or each condition (e.g. 6F and DDRR). Following sorting, cells were spun and
101

resuspended in 100 uL RLT buffer from the RNAeasy micro kit (Qiagen). To normalize to a
standard number of cells, ERCC spike-in mix (Thermo-Fisher, 1ul at 1:100 dilution) was added
to 50K cells in RLT. RNA in RLT and RNA extracted via RNAeasy kit, libraries were prepared by
DNAlink (San Diego, CA) using SMARTer Stranded Total RNA-Seq Kit-Pico Input Mammalian
(Clontech) and were sequenced using NextSeq 500 Mid-output 75PE (Illumina) to generate
30M reads per sample. Sequencing reads from triplicate or more replicates were trimmed and
aligned to mm10 reference transcriptome with STAR aligner 2.5.3a. Gene counts quantified
using annotation model (Partek E/M). Samples were aligned to ERCC spike-in reference to
quantify total spike-in reads per sample. Sample reads were normalized by spike-in reads to
generate cell number normalized reads per sample.

Biotinylated-trimethylpsoralen (bTMP) Immunofluorescence
One day after addition of N3 media (and without glial cultures), cells were treated with 1µM
aphidicolin for 1.5-2 hours in N3 media. For control experiments, MEFs were treated with or
without 100 µM bleomycin prior to incubation with psoralen. Then cells were incubated with 0.3
mg/mL EZ-Link Psoralen-PEG3-Biotin (Thermo Cat. No: 29986) for 15 minutes. Cultures were
then exposed to 3kJ m-2 of 365nM light (Fotodyne UV Transilluminator 3-3000 with 15W bulbs)
for 15 minutes at room temperature in the dark followed by 3 washes in PBS. Then cells were
fixed with cold 70% ethanol for 30 minutes at 4°C followed by another 3 washes in PBS. Cells
were then incubated with Alexa Fluor 594 Streptavidin (Thermo Cat. No: S32356) for one hour
at room temperature in the dark, washed with PBS 3 times, and then stained with Hoescht
(1:1000) for 10 minutes at room temperature in the dark. Coverslips were mounted onto glass
slides using ImmuMount and imaged using the Zeiss LSM 800 confocal microscope.

Trimethylpsoralen-qPCR
Cell Harvest and DNA Extraction
One day after addition of N3 media (and without addition of glial cultures), cells were treated
with 1µM aphidicolin for 1.5-2 hours in N3 media. Cells were then trypsinized in 0.25% Trypsin-
EDTA, spun down, and resuspended in complete N3 media + 1µM aphidicolin + 2µg/mL
trimethylpsoralen (Sigma). 500µL of control-puro was removed and saved for the no UV
crosslinking control. Each 1mL of the remaining samples were added to individual wells of a 24-
well plate and then exposed to 3kJ m-2 of 365nM light (Fotodyne UV Transilluminator 3-3000
with 15W bulbs) for 15 minutes at room temperature in the dark. Cells were then re-collected,
spun at 1000 x g for 5 minutes, washed with 1mL PBS, and spun down again. Then cells were
resuspended in 200µL PBS and purified using Qiagen DNeasy Blood and Tissue Kit with
inclusion of an RNase A digestion (Qiagen Cat. No: 69504). Samples were eluted once in
200µL followed by a second elution in 100µL of Buffer AE and eluates were then combined.

Sonication, Quantification, and Exonuclease Digestion
To achieve fragment sized of 100-500bp, each sample was sonicated in a Bioruptor for 30
seconds on/30 seconds off for 45 minutes on High. To ensure the same amount of DNA was
then used for Exonuclease digestion, sample concentrations were quantified with a qPCR
reaction. Briefly, samples were heat denatured at 95°C for 10 minutes, put on ice for 2 minutes,
and then spun down briefly. For the qPCR reaction, 2µL DNA for each sample was used in a
20µL iTaq Universal SYBR Green Supermix (Biorad Cat. No: 1725125) reaction using primers -
500bp upstream of the TSS for Actb. The qPCR results were used to determine the relative
concentrations of each sample, using the least concentrated sample as the reference to adjust
all other sample concentrations to. Samples were brought to a total volume of 280µL after
adjustment for DNA concentration and then heat denatured at 95°C for 10 minutes followed by 2
minute recovery on ice. Then, 240µL of each sample was put into a new tube, saving the
remaining undigested 40µL of DNA at 4°C. The 240µL samples were then heat denatured at
102

95°C for 10 minutes, incubated on ice for 2 minutes, and then briefly spun down. To each 240µL
sample, the following was added: 29µL 10X ExonucleaseI buffer + 1µL ExonucleaseI and
samples were incubated at 37°C for one hour. Samples were then heat denatured at 95°C
again, put on ice for 2 minutes, spun down, and another 10µL of ExonucleaseI was added. After
another 1 hour incubation at 37°C, samples were heated at 95°C for 10 minutes and put on ice
for 2 minutes to stop the exonuclease reaction.

TMP-qPCR
The non-exonuclease digested samples were diluted 1:8 in Milli-Q water to a total volume of
320µL. A qPCR reaction was then performed on both exonuclease digested and non-
exonuclease digested samples with the upstream primers (-500bp from TSS) for several genes.
Inclusion of non-exonuclease digested samples were used to normalize input levels for each
exonuclease treated sample. Each biological sample was run in technical triplicate using 4µL
DNA per well in a 20µL iTaq Universal SYBR Green Supermix reaction using the ViiA 7
Software. The following primers were used for qPCR quantification:

Actb:
5’-GTCTCGGTTACTAGGCCTGC-3’
5’- ATCCACGTGACATCCACACC-3’

Gapdh:
5’-GGTGAGATCAGTGAGGGGAG-3’
5’- CAAGAGGCTAGGGGCTTCC-3’

Sod1:
5’-TCCGCATTTCCAGACACAGT-3’
5’- GAGCGGGGAAAGTCGCTATT-3’

Live imaging
Live imaging was carried out using a Nikon Biostation CT.

QUANTIFICATION AND STATISTICAL ANALYSIS
Sample numbers and experimental repeats are indicated in figure legends. Unless otherwise
stated, data presented as mean + SEM of at least three biological replicates. Significance
determined by one-way ANOVA for multiple comparisons while an unpaired t-test was used
when comparing two data sets. If a data set was non-normally distributed according to the
D'Agostino & Pearson omnibus normality test, Kruskal-Wallis or Mann-Whitney testing was used
for multiple comparisons or when comparing two data sets, respectively. Significance summary:
p > 0.05 (ns),
∗
p ≤ 0.05,
∗∗
p ≤ 0.01,
∗∗∗
p ≤ 0.001, and
∗∗∗∗
p ≤ 0.0001.

DATA AND SOFTWARE AVAILABILITY
RNA sequencing data are available on GEO.

103

Appendix II: References

Abernathy, D.G., Kim, W.K., McCoy, M.J., Lake, A.M., Ouwenga, R., Lee, S.W., Xing, X., Li, D.,
Lee, H.J., Heuckeroth, R.O., Dougherty, J.D., Wang, T., and Yoo, A.S. (2017). MicroRNAs
induce a permissive chromatin environment that enables neuronal subtype-specific
reprogramming of adult human fibroblasts. Cell Stem Cell 21, 332-348.

Aguilera, A., and García-Muse, T. (2012). R Loops: from transcription byproducts to threats to
genome instability. Mol Cell 46, 115-124.

Bermúdez, I., García-Martínez, J., Pérez-Ortín, J.E., and Roca, J. (2010). A method for
genome-wide analysis of DNA helical tension by means of psoralen-DNA photobinding.
Nuclease Acids Res 38, e182.

Biddy, B.A., Kong, W., Kamimoto, K., Guo, C., Waye, S.E., Sun, T., and Morris, S.A. (2018).
Single-cell mapping of lineage and identity in direct reprogramming. Nature 564, 219-224.

Boguslawski, S.J., Smith, D.E., Michalak, M.A., Mickelson, K.E., Yehle, C.O., Patterson, W.L.,
and Carrico, R.J. (1986). Characterization of monoclonal antibody to DNA-RNA and its
application to immunodetection of hybrids. J Immunol Methods 89, 123-130.

Bowman, T., Symonds, H., Gu, L., Yin, C., Oren, M., and Van Dyke, T. (1996). Tissue-specific
inactivation of p53 tumor suppression in the mouse. Genes Dev 10, 826-835.

Branzei, D., and Foiani, M. (2010) Maintaining genome stability at the replication fork. Nat Rev
Mol Cell Biol 11, 208-219.

Briggs, R., and King, T.J. (1952). Transplantation of living nuclei from blastula cells into
enucleated frogs’ eggs. Proc Natl Acad Sci 38, 455-463.

Broderick, R., Nieminuszczy, J., Blackford, A.N., Winczura, A., Niedzwiedz, W. (2015). TOPBP1
recruits TOP2A to ultra-fine anaphase bridges to aid in their resolution. Nat Commun 6, 6572.

Buganim, Y., Faddah, D.A., Cheng, A.W., Itskovich, E., Markoulaki, S., Ganz, K., Klemm, S.L.
van Oudenaarden, A., and Jaenisch, R. (2012). Single-cell expression analyses during cellular
reprogramming reveal an early stochastic and a late hierarchic phase. Cell 150, 1209-1222.

Cahan, P., Li, H., Morris, S.A., Lummertz da Rocha, E., Daley, G.Q., and Collins, J.J. (2014).
CellNet: network biology applied to stem cell engineering. Cell 158, 903-915.

Cambell, K.H., McWhir, J., Ritchie, W.A., and Wilmut, I. (1996). Sheep clones by nuclear
transfer from a cultured cell line). Nature 380, 64-66.

Campbell, K.H.S, Fisher, P., Chen, W.C., Choi, I., Kelly, R.D.W., Lee, J-H., and Xhu, J. (2007)
Somatic cell nuclear transfer: past, present and future perspectives. Theriogenology 68, S214-
S231.

Chambers, S.M., and Studer, L. (2007). Cell fate plug and play: direct reprogramming and
induced pluripotency. Cell 145, 827-830.

104

Chen, T., Sun, Y., Ji, P., Kopetz, S., and Zhang, W. (2015). Topoisomerase IIα in chromosome
instability and personalized cancer therapy. Oncogene 34, 4019-4031.

Copelan, E.A. (2006). Hematopoietic stem-cell transplantation. N Engl J Med 354, 1813-1826.

Corti, S., Faravelli, I., Cardano, M., and Conti, L. (2015). Human pluripotent stem cells as tools
for neurodegenerative and neurodevelopmental disease modeling and drug discovery. Expert
opinion on drug discovery 10, 615-629.

Costa, A., Sanchez-Guardado, L., Juniat, S., Gale, J.E., Daudet, N., and Henrique, D. (2015).
Generation of sensory hair cells by genetic programming with a combination of transcription
factors. Development 142, 1948-1959.

Cox, A.D., and Der, C.J. (2010). Ras history the saga continues. Small GTPases 1, 2-27.

Davis, R.L., Weintraub, H., and Lassar, A.B. (1987). Expression of a single transfected cDNA
converts fibroblasts to myoblasts. Cell 51, 987-1000.

Delgado, J.L., Hsieh, C., Chan, L., and Hiasa, H. (2018). Topoisomerases as anticancer targets.
Biochem J 475, 373-398.

Dervishi, I., Gozutok, O., Murnan, K., Gautam, M., Heller, D., Bigio, E., and Ozdinler, P.E.
(2018). Protein-protein interactions reveal key canonical pathways, upstream regulators,
interactome domains, and novel targets in ALS. Sci Rep 8, 14732.

Di Giorgio, F.P., Carrasco, M.A., Siao, M.C., Maniatis, T., and Eggan, K. (2007). Non-cell
autonomous effect of glia on motor neurons in an embryonic stem cell-based ALS model. Nat
Neurosci 10, 608-614.

dos Santos, R.L., Tosti, L., Radzisheuskaya, A., Caballero, I.M., Kaji, K., Hendrich, B., and
Silva, J.C. (2014). MBD3/NuRD facilitates induction of pluripotency in a context-dependent
manner. Cell Stem Cell 15, 102-110.

Dykhuizen, E.C., Hargreaves, D.C., Miller, E. Cui, K., Korshunov, A., Kool, M., Pfister, S., Cho,
Y., Zhao, K., and Crabtree, G.R. (2013). mSWI/SNF (BAF) complexes facilitate decatenation of
DNA by topoisomerase Iiα. Nature 497, 624 – 627.

Elkon, R., Loayza-Puch, F., Korkmaz, G., Lopes, R., van Breugel, P.C., Bleijerveld, O.B.,
Altelaar, A.F., Wolf, E., Lorenzin, F., Eilers, M., et al. (2015). Myc coordinates transcription and
translation to enhance transformation and suppress invasiveness. EMBO reports 16, 1723-
1736.

Eminli, S., Foudi, A., Stadtfeld, M., Maherali, N., Ahfeldt, T., Mostoslavsky, G., Hock, H., and
Hockedlinger, K. (2009). Differentiation stage determines potential of hematopoietic cells for
reprogramming into induced pluripotent stem cells. Nature Genetics 41, 968-976.

Ferreirós, A., Pedrosa, P., Silva-Álvarez, S., Triana-Martínez, F., Vilas, J.M., Picallos-Rabina,
P., González, P., Gómez, M., Li, H., García-Cabellero, T., González-Barcia, M., Vidal, A., and
Collago, M. (2019). Context-dependent impact of RAS oncogene expression on cellular
reprogramming to pluripotency. Stem Cell Reports 12, 1099-1112.

105

Gan, W., Guan, Z., Liu, J., Gui, T., Shen, K., Manley, J.L., and Li, X. (2011). R-loop-mediated
genomic instability is caused by impairment of replicatinon fork progression. Genes Dev 25,
2041-2056
Guo, S., Zi, X., Schulz, V.P., Cheng, J., Zhong, M., Koochaki, S.H., Megyola, C.M., Pan, X.,
Heydari, K., Weissman, S.M., et al. (2014a). Nonstochastic reprogramming from a privileged
somatic cell state. Cell 156, 649-662.

Guo, Z., Zhang, L., Wu, Z., Chen, Y., Wang, F., and Chen, G. (2014b). In vivo direct
reprogramming of reactive glial cells into functional neurons after brain injury and in an
Alzheimer's disease model. Cell Stem Cell 14, 188-202.

Gurdon, J.B. (1962). The developmental capacity of nuclei taken from the intestinal epithelium
of feeding tadpoles. J Embryol Exp Morphol 10, 622-640.

Hanahan, D., and Weinberg, R.A. (2011). Hallmarks of cancer: the next generation. Cell 144,
646-674.

Hanna, J., Saha, K., and Jaenisch, R. (2010). Somatic cell reprogramming and transitions
between pluripotent states: facts, hypotheses, unresolved issues. Cell 143, 508-252.

Hanna, J., Saha, K., Pando, B., van Zon, J., Lengner, C.J., Creyghton, M.P., van Oudenaarden,
A., and Jaenisch, R. (2009). Direct cell reprogramming is a stochastic process amenable to
acceleration. Nature 462, 595-601.

Heins, N., Malatesta, P., Cecconi, F., Nakafuku, M., Tucker, K.L., Hack, M.A., Chapouton, P.,
Barde, Y.A., and Götz, M. (2002). Glial cells generate neurons: the role of the transcription
factor Pax6. Nat. Neurosci. 5, 308–315.

Ho, R., Sances, S., Gowing, G., Amoroso, M.W., O’Rourke, J.G., Sahabian, A., Wichterle, H.,
Baloh, R.H., Sareen, D., and Svendsen, C.N. (2016). Nat Neurosci 19, 1256-1267.

Hobbs, G.A., Der, C.J., and Rossman, K.L. (2016). RAS isoforms and mutations in cancer at a
glance. J Cell Sci 129, 1287-1292.

Hong, H., Takahashi, K., Ichisaka, T., Aoi, T., Kanagawa, O., Nakagawa, M., Okita, Y., and
Yamanaka, S. (2009). Supression of induced pluripotent stem cell generation by the p53-p21
pathway. Nature 460, 1132-1135.

Hou, P., Li, Y., Zhang, X., Liu, C., Guan, H., Zhao, T., Ye, J., Yang, W., Liu, K., Ge, J., Xu, J.,
Zhang, Q., Zhao, Y., and Deng, H. (2013). Pluripotent stem cells induced from mouse somatic
cells by small-molecule compounds. Science 341, 651-354.

Hu, W., Qiu, B., Guan, W., Wang, Q., Wang, M., Li, W., Gao, L., Shen, L., Huang, Y., Xie, G.,
Zhao, H., Jin, Y., Tang, B., Yu, Y., Zhao, J., and Pei, G. (2015). Direct conversion of normal and
Alzheimer’s disease human fibroblasts into neuronal cells by small molecules. Cell Stem Cell
17, 204-212.

Huangfu, D., Maehr, R., Guo, W., Eijkelenboom, A., Snitow, M., Chen, A.E., and Melton, D.A.
(2008a). Induction of pluripotent stem cells by defined factors is greatly improved by small-
molecule compounds. Nat Biotechnol 26, 795-797.
106

Huangfu, D., Osafune, K., Maehr, R., Guo, W., Eijkelenboom, A., Chen, S., Muhlestein, W., and
Melton, D.A. (2008b). Induction of pluripotent stem cells from primary human fibroblasts with
only Oct4 and Sox2. Nat Biotechnol 26, 1269-1275.

Ichida, J.K., Blanchard, J., Lam, K., Son, E.Y., Chung, J.E., Egli, D., Loh, K.M., Carter, A.C., Di
Giorgio, F.P., Koszka, K., et al. (2009). A small-molecule inhibitor of tgf-Beta signaling replaces
sox2 in reprogramming by inducing nanog. Cell Stem Cell 5, 491-503.

Ieda, M., Fu, J.D., Delgado-Olguin, P., Vedantham, V., Hayashi, Y., Bruneau, B.G., and
Srivastava, D. (2010). Direct reprogramming of fibroblasts into functional cardiomyocytes by
defined factors. Cell 142, 375-386.

Iwafuchi-Doi, M., and Zaret, K.S. (2014). Pioneer transcription factors in cell reprogramming.
Genes & Dev 28, 2679-2692.

Jaruszewicz, J., Kimmel, M., and Lipniacki, T. (2014). Stability of bacterial toggle switches is
enhanced by cell-cycle lengthening by several orders of magnitude. Physical review E,
Statistical, nonlinear, and soft matter physics 89, 022710.

Jiang, H., Xu, Z., Zhong, P., Ren, Y., Liang, G., Schilling, H.A., Hu, Z., Zhang, Y., Wang, X.,
Chen, S., Yan, Z., and Feng, J. (2015). Cell cycle and p53 gate the direct conversion of human
fibroblasts to dopaminergic neurons. Nat Commun 6, 10100.

Jones, R.G., and Thompson, C.B. (2009). Tumor suppressors and cell metabolism: a recipe for
cancer growth. Genes & Dev 23, 537-548.

Kanherkar, R.R., Bhatia-Dey, N., Makarev, E., and Csoka, A.B. (2014). Cellular Reprogramming
for understanding and treating human disease. Front Cell Dev Biol 2, 67.

Kawamura, T., Suzuki, J., Wang, Y.V., Menendez, S., Morera, L.B., Raya, A., Wahl, G.M., and
Izpisúa Belmonte, J.C. (2009). Linking the p53 tumour suppressor pathway to somatic cell
reprogramming. Nature 460, 1140-1144.

Keszthelyi, A., Minchell, N.E., and Baxter, J. (2016). The Causes and Consequences of
Topological Stress during DNA Replication. Genes 7.

Kim, Y., Zheng, X., Ansari, Z., Bunnell, M.C., Herdy, J.R., Traxler, L., Lee, H., Paquola, A.C.M.,
Blithikioti, C., Ku, M., et al. (2018). Mitochondrial Aging Defects Emerge in Directly
Reprogrammed Human Neurons due to Their Metabolic Profile. Cell Rep 23, 2550-2558.

Kim, D.H., Marinov, G.K., Pepke, S., Singer, Z.S., He, P., Williams, B., Schroth, G.P., Elowitz,
M.B., and Wold, B.J. (2015). Single-cell transcriptome analysis reveals dynamic changes in
lncRNA expression during reprogramming. Cell Stem Cell 16, 88-101.

King, H.W., and Klose, R.J. (2017). The pioneer factor OCT4 requires the chromatin remodeler
BRG1 to support gene regulatory element function in mouse embryonic stem cells. eLife 6,
e22631.

107

Komarov, P.G., Komarova, E.A., Kondratov, R.V., Christov-Tselkov, K., Coon, J.S., Chernov,
M.V., and Gudkov, A.V. (1999). A chemical inhibitor of p53 that protects mice from the side
effects of cancer therapy. Science 285, 1733-1737.

Kotsantis, P., Silva, L.M., Irmscher, S., Jones, R.M., Folkes, L., Gromak, N., and Petermann, E.
(2016). Increased global transcription activity as a mechanism of replication stress in cancer.
Nat Commun 7, 13087.

Kueh, H.Y., Champhekar, A., Nutt, S.L., Elowitz, M.B., and Rothenberg, E.V. (2013). Positive
feedback between PU.1 and the cell cycle controls myeloid differentiation. Science 341, 670-
673.

Kurita, M., Araoka, T., Hishida, T., O'Keefe, D.D., Takahashi, Y., Sakamoto, A., Sakurai, M.,
Suzuki, K., Wu, J., Yamamoto, M., et al. (2018). In vivo reprogramming of wound-resident cells
generates skin epithelial tissue. Nature 561, 243-247.

Kurth, I., Georgescu, R.E., and O’Donnell, M.E. (2013). A solution to release twisted DNA
during chromosome replication by coupled DNA polymerases. Nature 496, 119-122.

Lee, S.W., Oh, Y.M., Lu, Y.L., Kim, W.K., and Yoo, A.S. (2018). MicroRNAs Overcome Cell
Fate Barrier by Reducing EZH2-Controlled REST Stability during Neuronal Conversion of
Human Adult Fibroblasts. Dev Cell 46, 73-84 e77.

Lee, S.W., Oh, Y.M., Lu, Y.L., Kim, W.K., and Yoo, A.S. (2018). MicroRNAs Overcome Cell
Fate Barrier by Reducing EZH2-Controlled REST Stability during Neuronal Conversion of
Human Adult Fibroblasts. Dev Cell 46, 73-84 e77.

Lee, T.I., and Young, R.A. (2013). Transcriptional regulation and its misregulation in disease.
Cell 152, 1237-1251.

Li, H., Collado, M., Villasante, A., Strati, K. Ortega, S., Cañamero, M., Blasco, M.A., and
Serrano, M. (2009). The Ink4/Arf locus is a barrier for iPS reprograming. Nature 460, 1136-
1139.

Li, X., Zuo, X., J, J., Ma, Y., Wang, J., Liu, D., Zhu, J., Du, X., Xiong, L., Du, Y., Xu, J., Xiao, X.,
Wang, J., Chai, Z., Zhao, Y., and Deng, H. (2015). Small-molecule-driven direct reprogramming
of mouse fibroblasts into functional neurons. Cell stem cell 17, 195-203.

Liu, Z., Wang, L., Welch, J.D., Ma, H., Zhou, Y., Vaseghi, H.R., Yu, S., Wall, J.B., Alimohamadi,
S., Zheng, M., Yin, C., Shen, W., Prins, J.F., Liu, J., and Qian, L. (2017). Single-cell
transcriptomics reconstructs fate conversion from fibroblast to cardiomyocyte. Nature 551, 100-
104.

Lyssiotis, C.A., Foreman, R.K., Staerk, J., Garcia, M., Mathur, D., Markoulaki, S., Hanna, J.,
Lairson, L.L., Charette, B.D., and Bouchez, L.C. (2009). Reprogramming of murine fibroblasts to
induced pluripotent stem cells with chemical complementation of Klf4. Proc Natl Acad Sci 106,
8912-8917.

Ma, H., Wert, K.J., Shvartsman, D., Melton, D.A., and Jaenisch, R. (2018). Establishment of
human pluripotent stem cell-derived pancreatic beta-like cells in the mouse pancreas. Proc Natl
Acad Sci U S A 115, 3924-3929.
108

Ma, J., and Wang, M.D. (2016). DNA supercoiling during transcription. Biophys Rev 8, 75-87.

Ma, J., Bai, L., and Wang, M.D. (2013). Transcription under torsion. Science 340, 1580-1583.

Ma, T., Xie, M., Laurent, T., and Ding, S. (2013). Progress in the reprogramming of somatic
cells. Circ Res 112, 562-574.

Maherali, N., Sridharan, R., Xie, W., Utikal, J., Eminli, S., Arnold, K., Stadtfeld, M., Yachechko,
R., Tchieu, J., Jaenisch, R., Plath, K., and Hochedlinger, K. (2007). Directly reprogrammed
fibroblasts show global epigenetic remodeling and widespread tissue contribution. Cell Stem
Cell 1, 55-70.

Malumbres, M., and Barbacid, M. (2003). RAS oncogenes: the first 30 years. Nat Rev Cancer 3,
459-465.

Manzo, S.G., Hartono, S.R., Sanz, L.A., Marinello, J., De Biasi, S., Cossarizza, A., Capranico,
G., and Chedin, F. (2018). DNA topoisomerase I differentially modulated R-loops across the
human genome. Genome Biol 19, 100.
Marchetto, M.C., Winner, B., and Gage, F.H. (2010). Pluripotent stem cells in
neurodegenerative and neurodevelopmental diseases. Hum Mol Genet 19, R71-76.

Marión, R.M., Strati, K., Li, H., Murga, M., Blanco, R., Ortega, S., Fernandez-Capetillo, O.,
Serrano, M., and Blasco, M.A. (2009). A p53-mediated DNA damage response limited
reprogramming to ensure iPS cell genomic integrity. Nature 460, 1149-1153.

Marro, S., Pang, Z.P., Yang, N., Tsai, M.C., Qu, K., Chang, H.Y., Südhof, T.C., and Wernig, M.
(2011). Direct lineage conversion of terminally differentiated hepatocytes to functional neurons.
Cell Stem Cell 9, 374-382.

Mazzoni, E.O., Mahony, S., Closser, M., Morrison, C.A., Nedelec, S., Williams, D.J., An, D.,
Gifford, D.K., and Wichterle, H (2013). Synergistic binding of transcription factors to cell-specific
enhancers programs motor neuron identity. Nat Neurosci 16, 1219-1227.

McNeish, J., Gardner, J.P., Wainger, B.J., Woolf, C.J., and Eggan, K. (2015). From dish to
bedside: lessons learned while translating findings from a stem cell model of disease to a
clinical trial. Cell Stem Cell 17, 8-10.

Merrikh, H., Machon, C., Grainger, W.H., Grossman, A.D., and Soultanas, P. (2011). Co-
directional replication-transcription conflicts lead to replication restart. Nature 470, 554-557.

Mertens, J., Paquola, A.C., Ku, M., Hatch, E., Bohnke, L., Ladjevardi, S., McGrath, S.,
Campbell, B., Lee, H., Herdy, J.R., et al. (2015). Directly Reprogrammed Human Neurons
Retain Aging-Associated Transcriptomic Signatures and Reveal Age-Related
Nucleocytoplasmic Defects. Cell Stem Cell 17, 705-718.

Mor, N., Rais, Y., Sheban, D., Peles, S., Aguilera-Castrejon, A., Zviran, A., Elinger, D., Viukov,
S., Geula, S., Krupalnik, V., et al. (2018). Neutralizing Gatad2a-Chd4-Mbd3/NuRD Complex
Facilitates Deterministic Induction of Naive Pluripotency. Cell Stem Cell 23, 412-425 e410.

109

Morris, S.A., Cahan, P., Li, H., Zhao, A.M., San Roman, A.K., Shivdasani, R.A., Collins, J.J.,
and Daley, G.Q. (2014). Dissecting engineered cell types and enhancing cell fate conversion via
CellNet. Cell 158, 889-902.

Nakagawa, M., Koyanagi, M., Tanabe, K., Takahashi, K., Ichisaka, T., Aoi, T., Okita, K.,
Mochiduki, Y., Takizawa, N., and Yamanaka, S. (2008). Generation of induced pluripotent stem
cells without Myc from mouse and human fibroblasts. Nat Biotechnol 26, 101-106.

Naughton, C., Avlonitis, N., Corless, S., Prendergast, J.G., Mati, I.K., Eijk, P.P., Cockroft, S.L.,
Bradley, M., Ylstra, B., and Gilbert, N. Transcription forms and remodels supercoiling domains
unfolding large-scale chromatin structures (2013). Nat Struct Mol Biol 20, 387-395.

Nieminuszcy, J., Schwab, R.B., Niedzwiedz, W. (2016). The DNA fibre technique – tracking
helicases at work. Methods 108, 92-98.

Okita, K., Nakagawa, M., Hyenjong, H., Ichisaka, T., and Yamanaka, S. (2008). Generation of
mouse induced pluripotent stem cells without viral vectors. Science 322, 949-953.

Papp, B., and Plath, K. (2013). Epigenetics of reprogramming to induced pluripotency. Cell 152,
1324-1343.

Pauklin, S., and Vallier, L. (2013). The cell-cycle state of stem cells determines cell fate
propensity. Cell 155, 135-147.

Percharde, M., Bulut-Karslioglu, A., and Ramalho-Santos, M. (2017a). Hypertranscription in
Development, Stem Cells, and Regeneration. Dev Cell 40, 9-21.

Percharde, M., Wong, P., and Ramalho-Santos, M. (2017b). Global Hypertranscription in the
Mouse Embryonic Germline. Cell Rep 19, 1987-1996.

Pfisterer, U., Kirkeby, A., Torper, O., Wood, J., Nelander, J., Dufour, A., Bjorklund, A., Lindvall,
O., Jakobsson, J., and Parmar, M. (2011). Direct conversion of human fibroblasts to
dopaminergic neurons. Proc Natl Acad Sci U S A 108, 10343-10348.

Phatnanai, H.P. and Greenleaf, A.L. (2006). Phosphorylation and functions of the RNA
polymerase II CTD. Genes and Dev 20, 2922-2936.

Rais, Y., Zviran, A., Geula, S., Gafni, O., Chomsky, E., Viukov, S., Mansour, A.A., Caspi, I.,
Krupalnik, V., Zerbib, M., et al. (2013). Deterministic direct reprogramming of somatic cells to
pluripotency. Nature 502, 65-70.

Ranganathan, S., and Bowser, R. (2010). p53 and cell cycle proteins participate in spinal motor
neuron cell death in ALS. Open Pathol J 4, 11-22.

Sekiya, S., and Suzuki, A. (2011). Direct conversion of mouse fibroblasts to hepatocyte-like
cells by defined factors. Nature 475, 390–393.

Seyfried, T.N., and Huysentruyt, L.C. (2014). On the origin of cancer metastasis. Crit Rev
Oncog 18, 43-73.

110

Shen, C.N., Slack, J.M., and Tosh, D. (2000). Molecular basis of transdifferentiation of pancreas
to liver. Nat. Cell Biol. 2, 879–887.

Shi, Y., Do, J.T., Desponts, C., Hahm, H.S., Schöler, H.R., and Ding, S. (2008) A combined
chemical and genetic approach for the generation of induced pluripotent stem cells. Cell Stem
Cell 2, 525-528.

Shi, Y., Lin, S., Staats, K.A., Li, Y., Chang, W.H., Hung, S.T., Hendricks, E., Linares, G.R.,
Wang, Y., Son, E.Y., et al. (2018). Haploinsufficiency leads to neurodegeneration in C9ORF72
ALS/FTD human induced motor neurons. Nat Med 24, 313-325.

Singh, V.K., Kalsan, M., Kumar, N., Saini, A., and Chandra, R. (2015). Induced pluripotent stem
cells: applications in regenerative medicine, disease modeling, and drug discovery. Front Cell
Dev Biol 3, 1-18.

Skourti-Stathaki, K., and Proudfoot, N.J. (2014). A double-edged sword: R loops as threats to
genome integrity and powerful regulators of gene expression. Genes and Dev 28, 1384-1396.

Slattery, S.D., Newberg, J.Y., Szafran, A.T., Hall, R.M., Brinkley, B.R., and Mancini, M.A.
(2012). A framework for image-based classification of mitotic cells in asynchronous populations.
Assay Drug Dev Technol 10, 161-178.

Smith, D.K., Yang, J., Liu, M., and Zhang, C. (2016). Small molecules modulate chromatin
accessibility to promote NEUROG2-mediated fibroblast-to-neuron reprogramming. Stem Cell
Reports 7, 955-969.

Smith, Z.D., Nachman, I., Regev, A., and Meissner, A. (2010). Dynamic single cell imaging of
direct reprogramming reveals an early specifying event. Nat Biotech 28, 521-526.

Son, E.Y., Ichida, J.K., Wainger, B.J., Toma, J.S., Rafuse, V.F., Woolf, C.J., and Eggan, K.
(2011). Conversion of mouse and human fibroblasts into functional spinal motor neurons. Cell
Stem Cell 9, 205-218.

Soufi, A., Donahue, G., and Zaret, K.S. (2012). Facilitators and impediments of the pluripotency
reprogramming factors' initial engagement with the genome. Cell 151, 994-1004.

Spemann, H., and Mangold, H. (1924). Induction of embryonic primordia by implantation of
organizers from different species. Roux’s Arch Entw Mech 100, 599-638.

Spemann, H. (1938). Embryonic development and induction. Yale Univ

Srivastava, D., and DeWitt, N. (2016). In Vivo Cellular Reprogramming: The Next Generation.
Cell 166, 1386-1396.

Stadfield, M., and Hochedlinger, K. (2010). Induced pluripotency: history, mechanisms, and
applications. Genes Dev 24, 2239-2263.

Stadfield, M., Nagaya, M., Utikal, J., Weir, G., and Hochedlinger, K. (2008). Induced pluripotent
stem cells generated without viral integration. Science 322, 945-949.

111

Sun, C., Zhou, D., Zhang, Z., He, L., Zhang, F., Wang, X., Yuan, J., Chen, Q., Wu, L., and
Yang, Q. (2014). Senescence impairs direct conversion of human somatic cells to neurons. Nat
Commun 5, 4112.

Tachibana, M., Amato, P., Sparman, M., Gutierrez, N.M., Tippner-Hedges, R., Ma, H., Kang, E.,
Fulati, A., Lee, H-S., Sritanaudomchai, H., Masterson, K., Larson, J., Gokhale, S., Stouffer, R.L.,
Wolf, D., and Mitalipov, S. (2013). Human embryonic stem cells derived by somatic cell nuclear
transfer. Cell 153, 1228-1238.

Takahashi, K., and Yamanaka, S. (2006). Induction of pluripotent stem cells from mouse
embryonic and adult fibroblast cultures by defined factors. Cell 126, 663-676.

Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K., and Yamanaka, S.
(2007). Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell
131, 861-872.

Takahashi, S. (2012). Cellular reprogramming – lowering gravity of Waddington’s epigenetic
landscape. J Cell Sci 125, 2553-2560.

Takazawa, T., Croft, G.F., Amoroso, M.W., Studer, L., Wichterle, H., and Macdermott, A.B.
(2012). Maturation of spinal motor neurons derived from human embryonic stem cells. PLoS
One 7, e40154.

Tang, Y., Liu, M.L., Zang, T., and Zhang, C.L. (2017). Direct Reprogramming Rather than iPSC-
Based Reprogramming Maintains Aging Hallmarks in Human Motor Neurons. Frontiers in
molecular neuroscience 10, 359.

Tapia, N., and Scholer, H.R. (2010). p53 connects tumorigenesis and reprogramming to
pluripotency. J Exp Med 207, 2045-2048.

Teves, S., and Henikoff, S (2014). Transcription-generated torsional stress destabilizes
nucleosomes. Nat Struct Mol Biol 21, 88-94.

Thomas, M., White, R.L., and Davis, R.W. (1976). Hybridization of RNA to double-stranded
DNA: formation of R-loops. PNSA 73, 2294-2298.

Thomson, J.A., Itskovitz-Eldor, J., Shapiro, S.S., Waknitz, M.A., Swiergiel, J.J., Marshall, V.S.,
and Jones, J.M. (1998). Embryonic stem cell lines derived from human blastocysts. Science
282, 1145-1147.

Treutlein, B., Lee, Q.Y., Camp, J.G., Mall, M., Koh, W., Shariati, S.A., Sim, S., Neff, N.F.,
Skotheim, J.M., Wernig, M., and Quake, S.R. (2016). Dissecting direct reprogramming from
fibroblast to neuron using single-cell RNA-seq. Nature 534, 391-395.

Tsubouchi, T., Soza-Ried, J., Brown, K., Piccolo, F.M., Cantone, I., Landeira, D., Bagci, H.,
Hochegger, H., Merkenschlager, M., and Fisher, A (2013). DNA synthesis is required for
reprogramming mediated by stem cell fusion. Cell 152, 873-883.

Tuduri, S., Crabbe, L., Conti, C., Tourriere, H., Holtgreve-Grez, H., Jauch, A., Pantesco, V., De
Vos, J., Thomas, A., Theillet, C., et al. (2009). Topoisomerase I suppresses genomic instability
by preventing interference between replication and transcription. Nat Cell Biol 11, 1315-1324.
112

Turnquist, C., Horikawa, I., Foran, E., Major, E.O., Vojtesek, B., Lane, D.P. Lu, X., Harris, B.T.,
and Harris, C.C. (2016). p53 isoforms regulate astrocyte-mediated neuroprotection and
neurodegeneration. Cell Death Differ 23, 1515-1528.

Utikal, J., Polo, J.M., Stadtfeld, M., Maherali, N., Kulalert, W., Walsh, R.M., Khalil, A.,
Rheinwald, J.G., and Hochedlinger, K. (2009). Immortalization eliminates a roadblock during
cellular reprogramming into iPS cells. Nature 460, 1145-1148.

Vierbuchen, T., and Wernig, M. (2012). Molecular roadblocks for cellular reprogramming. Mol
Cell 47, 827-838.

Vierbuchen, T., Ostermeier, A., Pang, Z.P., Kokubu, Y., Sudhof, T.C., and Wernig, M. (2010).
Direct conversion of fibroblasts to functional neurons by defined factors. Nature 463, 1035-1041.

Vogt, M.A., Ehsaei, Z., Knuckles, P., Higginbottom, A., Helmbrecht, M.S., Kunath, T., Eggan, K.,
Williams, L.A., Shaw, P.J., Wurst, W., Floss, T., Huber, A.B., and Taylor, V. (2018). TDP-43
induced p53-mediated cell death of cortical progenitors and immature neurons. Scientific
Reports 8, 8097.

Vos, S.M., Tretter, E.M., Schmidt, B.H., and Berger, J.M. (2011). All tangled up: how cells direct,
manage and exploit topoisomerase function. Nat Rev Mol Cell Biol 12, 827-841.
Vrieseling, E., and Arber, S. (2006). Target-induced transcriptional control of dendritic patterning
and connectivity in motor neurons by the ETS gene Pea3. Cell 127, 1439-1452.

Wainger, B.J., Kiskinis, E., Mellin, C., Wiskow, O., Han, S.S., Sandoe, J., Perez, N.P., Williams,
L.A., Lee, S., Boulting, G., et al. (2014). Intrinsic membrane hyperexcitability of amyotrophic
lateral sclerosis patient-derived motor neurons. Cell Rep 7, 1-11.

Wang, L.H., Mayer, B., Stemmann, O., and Nigg, E.A. (2010). Centromere DNA decatenation
depends on cohesin removal and is required for mammalian cell division. J Cell Sci 123, 806-
813.

Wapinski, O.L.,YiLee, Q., Chen, A.C., Li, R., Corces, M.R., Ang, C.E., Treutlein, B., Xiang, C.,
Baubet, V., Suchy, F.P., Sankar, V., Sim, S, Quake, S.R., Dahmane, N., Wernig, M., and
Chang, H.Y. (2017). Rapid chromatin switch in the direct reprogramming of fibroblasts to
neurons. Cell Reports 20, 3236-3247.

Wapinski, O.L., Vierbuchen, T., Qu, K., Lee, Q.Y., Chanda, S., Fuentes, D.R., Giresi, P.G., Ng,
Y.H., Marro, S., Neff, N.F., et al. (2013). Hierarchical mechanisms for direct reprogramming of
fibroblasts to neurons. Cell 155, 621-635.

Watanabe, A., Yamada, Y., and Yamanaka, S. (2013). Epigenetic regulation in pluripotent stem
cells: a key to breaking the epigenetic barrier. Philos. Trans. R. Soc. Lond. B Biol. Sci. 368,
20120292.

Wen, X., Tan, W., Westergard, T., Krishnamurthy, K., Markandaiah, S.S., Shi, Y., Lin, S.,
Shneider, N.A., Monaghan, J., Pandey, U.B., et al. (2014). Antisense Proline-Arginine RAN
Dipeptides Linked to C9ORF72-ALS/FTD Form Toxic Nuclear Aggregates that Initiate In Vitro
and In Vivo Neuronal Death. Neuron 84, 1213-1225.
113

Wernig, M., Lengner, C.J., Hanna, J., Lodato, M.A., Steine, E., Foreman, R., Staerk, J.,
Markoulaki, S., and Jaenisch, R. (2009). A drug-inducible transgenic system for direct
reprogramming of multiple somatic cell types. Nat Biotechnol 26, 916-924.

Xu, J., Du, Y., and Deng, H. (2015). Direct lineage reprogramming: strategies, mechanisms, and
applications. Cell Stem Cell 16, 119-134.

Yamanaka, S. (2009). Elite and stochastic models for induced pluripotent stem cell generation.
Nature 460, 49-52.

Yan, J., Enge, M., Whitington, T., Dave, K., Liu, K., Sur, I., Schmierer, B., Jolma, A., Kivioja, T.,
and Taipale, J. (2013). Transcription factor binding in human cells occurs in dense clusters
formed around cohesin anchor sites. Cell 154, 801-813.

Yoo, A.S., Sun, A.X., Li, L., Shcheglovitov, A., Portmann, T., Li, Y., Lee-Messer, C., Dolmetsch,
R.E., Tsien, R.W., and Crabtree, G.R. (2011). MicroRNA-mediated conversion of human
fibroblasts to neurons. Nature 476, 228-231.

Yu, H., and Dröge, P. (2014). Replication-induced supercoiling: a neglected DNA transaction
regulator? Trends Biochem Sci 39, 219-220.
Yu, J., Hu, K., Smuga-Otto, K., Tian, S., Stewart, R., Slukvin, I.I., and Thomson, J.A. (2009)
Human induced pluripotent stem cells free of vector and transgene sequences. Science 324,
797-801.

Zegerman, P., and Diffley, J.F.X. (2009). DNA replication as a target of the DNA damage
checkpoint. DNA Repair 8, 1077-1088.

Zhao, T., and Xu, Y. (2010). p53 and stem cells: new developments and new concerns.

Zhao, Y., Yin, X., Qin, H., Zhu, F., Liu, H., Yang, W., Zhang, Q., Xiang, C., Hou, P., Song, Z.,
Liu, Y., Yong, J., Zhang, P., Cai, J., Liu, M., Li, H., Li, Y., Qu, X., Cui, K., Zhang, W., Xiang, T.,
Wu, Y., Zhao, Y., Liu, C., Yu, C., Yuan, K., Lou, J., Ding, M., and Deng, H. (2008). Two
supporting factors greatly improved the efficiency of human iPSC generation. Cell Stem Cell 3,
475-479.

Zhao, Z., Sagare, A.P., Ma, Q., Halliday, M.R., Kong, P., Kisler, K., Winkler, E.A., Ramanathan,
A., Kanekiyo, T., Bu, G., et al. (2015). Central role for PICALM in amyloid-beta blood-brain
barrier transcytosis and clearance. Nat Neurosci 18, 978-987.

Zhou, D., Zhang, Z., He, L., Du, J., Zhang, F., Sun, C., Zhou, Y., Wang, X., Lin, G., Song, K.,
Wu, L., and Yang, Q. (2014). Conversion of fibroblasts to neural cells by p53 depletion. Cell
Reports 9, 1-9.

Zhou, Q., Brown, J., Kanarek, A., Rajagopal, J., and Melton, D.A. (2008). In vivo reprogramming
of adult pancreatic exocrine cells to beta-cells. Nature 455, 627-632.

Zhou, Y., Wang, L., Vaseghi, H.R., Liu, Z., Lu, R., Alimohamadi, S., Yin, C., Fu, J.D., Wang,
G.G., Liu, J., et al. (2016). Bmi1 Is a Key Epigenetic Barrier to Direct Cardiac Reprogramming.
Cell Stem Cell 18, 382-395.

Abstract (if available)

Abstract Although cellular reprogramming enables the generation of new cell types for disease modeling and regenerative therapies, reprogramming remains a rare cellular event. By examining reprogramming of fibroblasts into motor neurons and multiple other somatic lineages, we find that epigenetic barriers to conversion can be overcome by endowing cells with the ability to mitigate an inherent antagonism between transcription and DNA replication. We show that transcription factor overexpression induces unusually high rates of transcription and that sustaining hypertranscription and transgene expression in hyperproliferative cells early in reprogramming is critical for successful lineage conversion. However, hypertranscription impedes DNA replication and cell proliferation, processes that facilitate reprogramming. We identify a chemical and genetic cocktail that dramatically increases the number of cells capable of simultaneous hypertranscription and hyperproliferation by activating topoisomerases. Further, we show that hypertranscribing, hyperproliferating cells reprogram at 100-fold higher, near-deterministic rates. Therefore, relaxing biophysical constraints overcomes molecular barriers to cellular reprogramming.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Generation and characterization of an in vitro model of inner ear sensory hair cells using transcription factor mediated cellular reprogramming

PDF

Dissecting the heterogeneity of mouse hematopoietic stem cells in vivo

PDF

Functional compensation between hematopoietic stem cell clones in vivo

PDF

The role of ERK1/2 in mouse embryonic stem cell fate control

PDF

The organ of Corti sensory progenitor cell competence is controlled by SoxC transcription factors

PDF

Identification of therapeutic targets in human cerebral brain organoid models of neurodegeneration

PDF

Transcriptional and epigenetic mechanisms underlying sensory hair cell differentiation and regeneration

PDF

Developing a robust single cell whole genome bisulfite sequencing protocol to analyse circulating tumor cells

PDF

Exploring stem cell pluripotency through long range chromosome interactions

PDF

Using novel small molecule modulators as a tool to elucidate the role of the Myocyte Enhancer Factor 2 (MEF2) family of transcription factors in leukemia

PDF

Investigating the effects of polycomb repressive complex inhibitors on cancer cell phenotypic plasticity

Asset Metadata

Creator Wunder, Kimberley Nicole (author)

Core Title Investigating molecular roadblocks to enhance direct cellular reprogramming

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Development, Stem Cells and Regenerative Medicine

Publication Date 10/23/2019

Defense Date 06/19/2019

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cellular reprogramming,disease modeling,genomic instability,hypertranscription,neurodegeneration,OAI-PMH Harvest,p53,RepSox,single-cell RNA-seq,topoisomerase,transcription factor,transcription rate

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Segil, Neil (committee chair), Ichida, Justin K. (committee member), Ying, Qilong (committee member)

Creator Email babos@usc.edu,kbabos89@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-227695

Unique identifier UC11675570

Identifier etd-WunderKimb-7878.pdf (filename),usctheses-c89-227695 (legacy record id)

Legacy Identifier etd-WunderKimb-7878.pdf

Dmrecord 227695

Document Type Dissertation

Rights Wunder, Kimberley Nicole

Type texts

Source 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 a...

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