Epigenetic control of neuronal activity dependent gene transcription as a basis for long term memory formation

Copyright
Nicodemus Edrick Oey
2014
iv

Abstract
The brain is a highly plastic structure, capable of changing its responses to the
external stimuli it receives. This evolutionarily conserved concept of
neuroplasticity underlies all forms of memory, from sea slugs, flies, rodents, all
the way to humans. While short-term memories that last for minutes are mediated
by transient changes in neural connections, long-term memories that last for
years require more persistent changes that involve the production of new
proteins. Decades of research have shown that the molecular mechanisms
responsible for the consolidation of long-term memory are unlikely to be mediated
by changes to the static genomic DNA sequence. The last twenty years have
seen the emergence of epigenetics as a highly sophisticated mechanism by
which a neuron can dictate with remarkable specificity which genes should be
expressed at precisely which time in response to activity.

Neuronal activity-dependent gene transcription depends on the action of several
enzymes that respond to activity to specifically regulate the expression of genes
that effectuate downstream functions. The ability of epigenetic regulators to tag
specific locations in the genome for the de-novo transcription of genes has
proven to be essential to learning and memory, as indicated by the disastrous
consequences of their absence in various clinical syndromes of mental
retardation. The present work attempts to study and characterize the events that
mediate long-term memory consolidation from an epigenetics standpoint,
specifically in chromatin modification or “epigenetic tagging” of specific
nucleosomes which seem to be involved in both early and late events of memory
consolidation along the temporal axis of neural activity. Amongst the many
epigenetic regulators important for memory function, the TIP60 protein, in
particular, is of significant interest due to its involvement both in early events of
neuronal activity-dependent gene induction, and also in late events consisting of
epigenetic changes leading to long-lasting memory consolidation.
Using a combination of biochemistry, super-resolution microscopy, chromatin
immunoprecipitation, and mass spectrometry-based techniques, the first part of
this thesis presents findings on the role of the Alzheimer’s Disease-associated
epigenetic enzyme TIP60 and an X-linked Mental Retardation (XLMR)-associated
protein PHF8 in the rapid neuronal activity-dependent transcription of ARC, a
crucial regulator of memory consolidation. The second part of this thesis will
explore the role of TIP60 in mediating the functions of ARC protein itself in the
late epigenetic events that eventually result in memory consolidation. The last
part of the thesis will be devoted to discussing the importance of epigenetic
processes of chromatin modification in general neuronal functions such as
development and survival, as well as specific functions such as memory
consolidation. Finally, looking forward to the future, several of the potentially
endless possibilities in neuroepigenetics such as clinical applications of
epigenetic modifying therapy and ARC-modulating strategies in neuropsychiatric
disorders will be offered.

v

Dedication

As you read the words written on this page, neuronal cells in your brain are firing in
highly patterned electrical activities across their synapses in order to encode the
information you read, some of which you may remember, being eventually stored in
your memory. This remarkable ability of the brain which is composed of over 200
billion neurons and more than 100 trillion synapses is thanks to the amazing
capabilities found in each unique neuron, which is able to change the genes it
expresses at any given time in response to the pattern of activity it receives. Such is
the dynamics of neuronal cells that networks of them are able to underlie our ability
to see, to hear, to smell, to taste, to move, to feel, and most importantly, to think…
I would like to dedicate these four years’ work to my parents, Hoat and Megan Oey,
for bringing me to life, for raising me as a good son and brother, and for allowing me
to travel 15349km to Singapore to start living my dream of being a Clinician-Scientist,
a long, arduous journey which this PhD dissertation is a part of: Mom and Dad, I love
you. I would like to thank my siblings for all their support: my sister Elrika and new
brother-in-law Christopher Prest in Waterloo for allowing me to sing at their wedding,
my younger sister Elvina Oey in New York for the invaluable Occupational Therapist
and Psychologist point-of-view, my younger brother Edbert Oey for muscling me into
the right path no matter how much I strayed, my younger cousin William Tanaka Oey
for being the fun, kind-hearted man that he is and Carissa Oey for taking care of my
parents when I should be the one doing it…. thank you to all my wonderful siblings: I
love you guys. One incredibly special mention goes out to the love of my life,
Christine Chan, PhD candidate, for correcting my drawings of neurons that look like
“Strepsils™”, for those many hours of rehearsals before presentations, and for
helping me clean my incredibly messy room: I love you, Hui Shan.
I would like to thank the VanDongen Laboratory: Shaun Teo and Caroline Wee, for
holding me by the hand when I was first trying to walk (aka run gels), Rajaram

Ezhilarasan for the hard work and dedication, Niamh Higgins, Knvul Sheikh, Annabel
Tan, Gokul Banumurthy our amazing RA’s who work day and night and on the
weekends too, Ju Han the computer genius who taught me Chinese, Mark Dranias
for the discussions on synaptic plasticity, Xiaoyu for investigating long-term memory
in cultured neurons, and especially How Wing Leung, our post-doc, who has taught
me everything I know. Finally, I would like to say to my lab mom and dad, Tony and
Margon VanDongen, “Aren’t you glad I survived?” – thank you Margon for the best
spaghetti I’ve ever tasted, for believing in me when no one else did, and for
defending me when odds were against it, thank you Tony for mentoring me… you
have taught me how to do science, and that is something I will never forget for the
rest of my life. I am forever in your debt. With that, I hope you don’t mind me saying
that I am very much looking forward to the exciting research we have planned
together on the horizon!

vi

Contents
Title Page i
Abstract Signature ii
Copyright iii
Abstract iv
Dedication v
Table of Contents vi
List of Figures x
List of Abbreviations xii

Chapter 1
Review of Literature 1
I. Introduction to Learning and Memory
A. A short history 1

B. Classifications of memory 3
II. Molecular Mechanisms of Long-Term Memory Formation 5
A. Overview: plasticity and activity 5
B. Epigenetics as a mechanism of activity-dependent gene
expression 6
C. ARC: a master regulator of synaptic plasticity 11
D. TIP60: an effector of early and late neuroepigenetic events 15
E. PHF8: a specialized neuronal transcriptional co-activator 16
III. The Timeline of Neuronal Activation 18
Chapter 2
Early Epigenetic Events: the Characterization of a Chromatin-modifying
Complex Composed of PHF8 and TIP60 that Alter H3K9acS10P to Enable
Activity-dependent Transcription of Arc 20
I. Abstract 20
II. Introduction 21
III. Materials and Methods 24
A. Plasmid Construction and Cloning 24
B. Hippocampal and Cortical Neuronal Cell Culture 25
C. Transfections and Neuronal Stimulations 26
D. Conventional Immunofluorescence 26
E. Proximity Ligation In-Situ Assay 27
F. Widefield Microscopy, Calcium imaging, and Data Analysis 28
G. Co-immunoprecipitation and Western Blotting 29
H. Immunoprecipitation followed by Mass Spectrometry 30
vii

I. Chromatin Immunoprecipitation (ChIP) and Triton X-Acetic Acid-
Urea histone gel electrophoresis 30
J. 3-dimensional Structured Illumination Microscopy 32
K. 3-dimensional Stochastic Optical Reconstruction Microscopy 32

IV. Results 33
A. Transcriptional Activators PHF8 and TIP60 Colocalize in the
Interchromatin Space 33
B. The Histone Demethylase PHF8 Physically Associates the
Histone Acetyltransferase TIP60 37
C. PHF8 and TIP60 Form a Dual-Function Complex that Increases
Histone Acetylation on H3K4me3-bearing Chromatin 39
D. PHF8 Removes Transcriptionally Suppressive H3K9me2 and
Associates with Transcriptionally Active H3K9ac 42
E. PHF8 and TIP60 are Activity-dependent and Co-regulate
H3K9acS10P in Response to Neuronal Activity 43
F. The PHF8-TIP60 Complex Modulates activity-induced
H3K9acS10P 46
G. The PHF8-TIP60 Interactome is Rich in Proteins Involved in
Transcription and Includes the Neuronal Splicing Factor PSF 50
H. Super-resolution Microscopy Situates Endogenous PHF8, TIP60,
and PSF Within 30nm of Each Other in the Activated Neuronal
Nucleus 53
V. Discussion 56
Chapter 3
Late Epigenetic Events: the Interaction Between TIP60 and ARC
Functions to Regulate H4K12ac, a Learning-induced Chromatin
Modification Involved in Ageing-associated Memory Impairment 64
I. Abstract 64
II. Introduction 66
III. Materials and Methods 68
A. Constructs and Cloning 68
B. Cell Culture 69
C. Transfections and Stimulations 70
D. Immunofluorescence 70

E. Imaging and Data Analysis 71
F. 3-dimensional Structured Illumination Microscopy 72
G. Photo-activated Localization Microscopy (PALM) and Direct
Stochastic Optical Reconstruction Microscopy (dSTORM) 72
H. Immunoprecipitation and Western Blotting 73
I. Induction of Arc gene expression by stimulation of neural network
activity 74
IV. Results 75
viii

A. ARC Protein Interacts with betaSpIVSigma5, PHF8, PML and
TIP60 and Components of the TIP60 Chromatin Remodeling
Complex 75
B. PML, TIP60, and ARC Form a Tight Complex in the Nucleus of
Activated Neurons 77
C. TIP60 and ARC Overexpression Increases H4K12 Acetylation but
not H3K9, H3K14, H2AK5, or H2BK5 Acetylation 78
D. ARC, PML, and PHF8 Modulate TIP60’s Acetyltransferase Activity
80
E. Endogenous ARC Interacts with TIP60 in a Variety of Dynamic
Nuclear Structures as Seen on Localization Microscopy 81
F. Endogenous ARC is Correlated with High TIP60 Nuclear Levels in
Activated Neurons 84
G. ARC Recruits TIP60 to PML Bodies 85
H. Activity-induced ARC Increases H4K12 Acetylation at a Timepoint
that Correlates with Memory Consolidation in Neurons 86
I. The Enzymatically Inactive Mutant of TIP60 Fails to Induce H4K12
Acetylation in Hippocampal Neurons 88
J. ARC Associates at Single-Molecule Level with the Learning-
Induced Histone Mark H4K12ac 89

V. Discussion 91

Chapter 4
Integrating the Findings: the Elucidation of the Genes and Mechanisms
that lead to Memory Consolidation 98
I. Introduction 98
II. Preliminary Results and Discussion
A. In-vitro Neural Network Activity Leads to Specific Site-Directed
Changes in Chromatin Modification 101
B. In-vivo Novel Environment Enrichment Leads to Specific Patterns
of Chromatin Modification Partly Mediated by PHF8 and TIP60102
C. PHF8 and TIP60 are Activity-Dependent Chromatin-modifying
Enzymes With Different Promoter Occupancy Profiles 103
D. The Transcriptional Activator PHF8 is Found Within Nanometres
of PTB-associated Splicing Factor and Nascent RNA 105
E. The Recruitment of PHF8 to Active Transcriptional Start Sites
Precedes RNA Polymerase II Binding at the Arc and c-Fos
Genomic Loci Following Neuronal Activation 106
F. Specific Regulation of Arc Gene Expression by ERK and p38
MAPK Signaling Pathways 107
G. The Interactome of PHF8, TIP60, and ARC Give Novel Clues to
the Processes that Lead Ultimately to Memory Consolidation 110

ix

Chapter 5
Conclusions and Future Directions: Towards Epigenetically Informed
Translational and Clinical Trials 117
Bibliography 125

Appendix A.
Publications accepted or under review 138


x

List of Figures
Figure 1: PHF8 and TIP60 colocalize and recruit each other in neuronal interchromatin
space. . 35
Figure 2: PHF8 and TIP60 physically associate to form a dual function chromatin-modifying
complex 38
Figure 3: PHF8 removes the repressive histone mark H3K9me2 and associates with the
activating histone mark H3K9ac. 42
Figure 4: Neuronal activity reorganizes PHF8 and TIP60 in the nucleus and effectuate
histone methylation and acetylation changes. 45
Figure 5: PHF8 and TIP60 modulate neuronal activity-induced histone acetylation at
H3K9acS10P and activation of the Arc gene 48
Figure 6: Knockdown of PHF8 impairs activity-dependent induction of H3K9acS10P and Arc
and c-Fos expression 49
Figure 7: PHF8, TIP60, and H3K9acS10P are specifically enriched in the transcriptional start
site of the Arc gene. 50
Figure 8: Common interacting partners between PHF8 and TIP60 function primarily in
transcription and mRNA processing. 52
Figure 9: Endogenous TIP60 is located within 30nm of PHF8 in the activated hippocampal
neuronal nucleus 54
Figure 10: PHF8 and TIP60 form a tripartite complex with the splicing factor PSF/SFPQ 55
Figure 11: Four-color immunofluorescence of a quaternary complex formed between ARC,
PML, bSpectrin, and TIP60 77
Figure 12: ARC protein interacts with two members of the TIP60 chromatin remodeling
complex: the transcriptional coactivator BRG1 and AMIDA 77

Figure 13: Endogenous ARC is able to localize TIP60 to PML bodies 79
Figure 14: ARC+TIP60 overexpression had a mild effect on global H4K12 acetylation 80
Figure 15: ARC has a positive modulatory effect on TIP60-mediated H4K12 acetylation. 81
Figure 16: 3D Stimulated Emission Depletion Microscopy shows association of endogenous
Arc and Tip60. 83
Figure 17: Dual-color super-resolution microcopy of Arc-mEOS2 and endogenous Tip60 in
the activated neuronal nucleus. 84
xi

Figure 18: ARC protein levels correlate with that of TIP60 at the 4-hour mark of sustained
neural activity. 85
Figure 19: ARC recruits Tip60 to PML bodies. 86
Figure 20: ARC expression increases H4K12ac levels. 89
Figure 21: A Tip60 mutant lacking acetyltransferase activity decreases H4K12 acetylation.
91
Figure 22: ARC associates at the single molecule level with H4K12ac, a learning-induced
histone mark 92
Figure 23: No global changes in histone modification seen with 3 hours of network activity.
102
Figure 24: Network activity induces specific chromatin modification changes 102
Figure 25: In-vivo paradigm of Novel Environment Enrichment induces specific increases in
H3K9acS10P in a subset of neurons in the brain 104
Figure 26: NEE induces specific H3K9acS10P within a subset of neurons that have TIP60 and
ARC in the hippocampus. 107
Figure 28: Single-molecule localization microscopy situates PHF8 and its binding partner
PSF in areas of active RNA transcription 109
Figure 29: PHF8 is recruited to TSS of neuronal activity-dependent genes before RNA
Polymerase II 110
Figure 30: Control of specific activity-induced gene expression by modulation of upstream
signaling kinase pathways. 112

Figure 31: The interactome of PHF8, TIP60, and ARC reveals a major uniting mechanism of
mRNA metabolism, transcriptional regulation, and mRNA splicing. 115
Figure 32: A graphical abstract of activity-dependent DNA, histone, RNA, and protein
changes in a single brain cell. 116
Figure 1 – Graphical Summary of Present Dissertation 122

xii

List of Abbreviations
Arc = Activity-Regulated Cytoskeletal-associated gene (NCBI gene ID 11838 in Mus musculus)

CFP = Cyan Fluorescent Protein
DAPI = 4',6-diamidino-2-phenylindole; marker of cellular DNA
Fos = FBJ osteosarcoma oncogene (NCBI gene ID 14281 in Mus musculus)
GFP = Green Fluorescent Protein
H3 = Histone 3
H3K9 = Histone 3 Lysine 9
H3K9ac = Histone 3 Lysine 9 acetylated
H3S10p = Histone 3 Serine 10 phosphorylated
H3K9acS10P = Histone 3 Lysine 9 acetylated Serine 10 phosphorylated
HEK293 = Human Embryonic Kidney cells 293
IEG = Immediate-Early Gene
MYST = MOZ, YBF2, SAS2, TIP60 family of acetyltransferases
nm = nanometer (1×10
−9
m)
NEE = Novel Environmental Enrichment
PHF8 = PHD-Finger protein 8 (NCBI gene ID 320595 in Mus musculus)
Pol II = RNA Polymerase II
PSF = Polypyrimidine Tract Binding protein-associated Splicing Factor, also known as SFPQ

(Splicing Factor Proline/Glutamine rich, NCBI gene ID 71514 in Mus Musculus)
SIM = Structured Illumination Microscopy
STORM = Stochastic Optical Reconstruction Microscopy
TIP60 = Tat-Interacting Protein 60kDa (also known as Kat5, K(lysine) acetyltransferase 5)
YFP = Yellow Fluorescent Protein


1

Chapter 1
Review of Literature

1.I. Introduction to Learning and Memory
1.I.A. A short history
“You have to begin to lose your memory, if only in bits and pieces, to realize that
memory is what makes our lives. Life without memory is no life at all Our memory is
our coherence, our reason, our feeling, even our action. Without it we are nothing.”
― Luis Buñuel
A major goal of neurobiology is to establish how complex behaviors such as
learning and memory are encoded in the cells of the brain itself. In the 28,470
days that constitutes the average human lifespan
1
, a person may come in
contact with over 100,000 other individuals and a multitude of things, places,
tastes, sounds, smells, and other sensations which form the bulk of his
experiences in life, all of which will largely influence the way he thinks and
behaves. We refer to this ability as learning, which is defined as a
measurable, adaptable change in a person’s behavior due to experience.
Memory, on the other hand, is the storage and recall of that experience which
is required for a person to use what was learned. Although there has been

much discord between psychologists using “top-down” approaches and
physiologists who employ “bottom-up” approaches in formulating theories
about learning and memory, studies of direct removal of certain areas of the
brain such as in the case of H.M
2
definitively show that the underlying
mechanisms of memory are found in the nervous system. Experiments
conducted by Karl Lashley in the 1930s were amongst the first to dissect out
the existence of a memory trace in the brain: using surgical lesions to disrupt
2

an already-formed spatial memory in animals, it was concluded that memories
are not localized in any one structure but are rather distributed diffusely in
equipotent areas throughout the cortex
3
.
“This series of experiments has yielded a good bit of information about what and
where the memory trace is not. It has discovered nothing directly of the real nature of
the engram. […] I believe that even the reservation of individual synapses for special
associative reactions is impossible. The alternative is, perhaps, that the dendrites
and cell body may be locally modified in such a manner that the cell responds
differentially, at least in the timing of its firing, according to the pattern of combination
of axon feet through which excitation is received.” [Lashley K. In search of the
engram. SympSocExp Biol. 1950;4:454–82.]

Remarkably even at this early stage of neurobiological research, Lashley had
already held the view that memory formation involved the activity of
thousands to millions of neurons connected in circuits wherein each individual
neuron and its synaptic connections “may be locally modified”. This view was
corroborated by seminal work by Donald Hebb and Jerzy Konorski who

suggested that synaptic connections may be made more effective by neuronal
activity which produce both a short increase in excitability, thought to underlie
short-term memory, and a long-lasting structural change in the cell body and
synapse itself, thought to underlie long-term memory
4
. This theory, which
continues to undergo rigorous validation through studies done at the
anatomical, biochemical, electrophysiological and now molecular level, is
based on the principle that for learning and memory to occur, changes in the
environment must be translated into measurable, adaptable changes in the
physiology of the neuron. The changes exerted by differing patterns of
neuronal activity that result in behavioral modification constitute a mysterious
phenomenon commonly referred to as neuroplasticity.
3

It is on this contextual framework of neuroplasticity that I now address the
main issue of this thesis which is how does the environment affect individual
neuronal behavior to learn and encode meaningful memories? I will begin by
first exploring what types of memory are working at the broad, psychological
level. I will then delve deeper into the proposed mechanisms accounting for
memory at the neural network and cellular level, which are all the result of
many decades of research. Finally, I will focus on the molecular mechanisms
of memory formation, including an overview of the nascent field of
neuroepigenetics on which this thesis is built.

1.I.B. Classifications of Memory
To highlight the gaps in our knowledge about memory formation, it is
necessary to briefly review the currently dominant theory of how memory
works. At the moment of this writing, the generally accepted model, attributed
to Atkinson and Shiffrin in 1968, is that there exist at least three distinct

systems that flow naturally from one to another, all of which function together
to form a lasting memory
5
. Information in the form of stimuli detected by our
senses (vision, hearing, touch) can be either ignored and disappear or
perceived and enter into our sensory register: a “buffer” or temporary storage
which works subconsciously on a time scale of less than one second
6,7
. When
attended to with the proper amount of attention, it is thought that some
information is transferred to short-term memory, which works on a time scale
of less than one minute
8
. Finally, information from short-term memory, through
4

a mechanism called consolidation, can then enter long-term memory, which
may last for days, months, or even in some cases a lifetime
9
.
Regardless of what framework one chooses to study memory, one of the most
crucial questions in neuroscience remains how does information become
represented by the millions of cells in the brain? Between top-down
approaches that look at human behavior, and studies that attempt to look at
this question from the “bottom up” or from the molecules that constitute the
brain, is the “black box” which is the neural network. The brain is made up of
hundreds of thousands of networks each of which is built from thousands of
individual neurons connected together. Due to the elusiveness of the engram,
the idea that memories are stored in a distributed manner in the brain has
become increasingly tangible. In support of this idea, researchers are now

able to simulate biologically plausible networks in silico
10
, which have the
capacity to recognize faces and learn to differentiate different types of music,
thanks to the emergent properties of the network: that is, capabilities that are
not explained by the physiology of each individual neuronal cell. These
computational experiments have since been validated in living neuronal
networks
11
. In this respect, memory has been attributed to the ability of an
interconnected network of neurons to transiently store information through
modification of the strength of the synapses that link them. Although this may
serve as a plausible explanation for working memory that lasts several
seconds at maximum, the fact that long-term memory as we know it may
persist for months and even years point toward the existence of another, more
long-lasting mechanism.

5

1.II. Molecular Mechanisms of Long-Term Memory
Formation

1.II.A. Overview
In contrast to short-term memory which relies on reverberating activity
between neurons and therefore lasts only a few seconds, long-term memory
can last for decades and therefore require more permanent changes at the
structural level which persist in the absence of transient action potentials.
Research into the cellular and molecular mechanisms underpinning complex
cognitive functions began with the groundbreaking discovery that neurons can
adapt in response to the external environment not only during development

but also in adulthood, where neuroplasticity governs diverse phenomena
ranging from vision, somatosensory maps, to learning and memory. Early
studies, performed in many different species, involving intracerebral injections
of mRNA and protein synthesis inhibitors showed that inhibiting protein
synthesis circumjacent to a training period does not affect short-term memory
but abolishes the ability of animals to form long-term memory, indicating that
there is a window of time during which gene transcription and protein
synthesis is crucial for the formation of long-term memory
12-14
. It has therefore
been known for several decades that there is a fundamental mechanistic
difference in the way short- and long-term memory are formed
15
.
Unlike short-term memory, the formation of long-term memory is entirely
dependent on gene transcription and subsequent protein synthesis, which
was first demonstrated by elegant studies in the model organism Aplysia
16,17
.
Since then, extensive work in Drosophila and rodents has validated the notion
6

that activity-dependent neuronal plasticity underlies long-term memory
18-21
.
Central to all of these studies is the basic principle that external stimuli,
through alterations of neurotransmitter release, ion channel opening, Ca
2+

influx, or electrical activity at neuronal synapses, are able to induce changes

in gene expression that occur inside the neuronal nucleus
22-24
. The basic
premise that gene regulation in the nucleus can be directly altered by neural
activity at the synapse lies at the core of the concept of neuronal plasticity that
underlies many complex aspects of cognition and behavior. But how does a
neuron dynamically change its gene expression programs in response to
synaptic activity?

1.II.B. Epigenetics as a mechanism of activity-dependent
gene expression
In the process of differentiation, pluripotent embryonic stem cells divide into
highly specialized cell types such as hepatocytes, melanocytes, or neurons
themselves which, despite having the same genetic material as the original
stem cell, continue to “remember” they are liver, pigment, or brain cells even
as they divide further and replicate their entire genome de novo. Upon
exposure to various antigens, T lymphocyte precursors of the mammalian
immune system commit themselves to an almost limitless array of
differentiated states each of which expresses different genes, forming an
immunological memory in response to transient signals from foreign invaders.
In much the same way, the remarkable ability of neurons to adapt to the
environment can be traced to a fundamental property of all cells, which is that
of cellular memory: this is the concept of epigenetics (epi – from the Greek επί
7

meaning over or outside of), which is the study of all physiological processes
that are not caused by changes in DNA sequence but can nevertheless highly
influence gene expression
25
. Previously thought to be hard-wired only into

developing cells, epigenetic mechanisms are now also thought to function in
the dynamic regulation of gene expression in the post-mitotic neurons of the
adult nervous system
26
. The four cardinal mechanisms of epigenetics include:
1) DNA methylation: the addition of methyl groups to DNA bases, 2) non-
coding RNAs: RNA editing and DNA recoding, 3) nucleosome remodeling: the
higher-order 3-dimensional arrangement of chromatin, and 4) histone code:
post-translational modifications of histone molecules that package DNA
itself
27
.
The present work focuses on the role of chromatin modification as an
important phenomenon in regulating gene expression programs which
mediate activity-dependent plasticity in the nervous system. In neurons,
genetic material does not exist naked in the nucleus but is compacted in a
highly ordered structure consisting of histones and DNA, collectively called
chromatin
28
. Due to the immense levels of compaction, wherein a 3-meter
long stretch of chromosomal DNA is able to fit into a nucleus that is 3-microns
in length, a large majority of the genome is virtually inaccessible to gene
transcriptional machinery. Hence, in a system that requires dynamic gene
expression changes, there is a need for a mechanism to attach molecular
labels to parts of the genome that needs to be activated at the right time.
The basic organization of chromatin is a stretch of 146bp of DNA wrapped
around an octameric core histone complex, which altogether is termed the
nucleosome. Each individual histone, including H2A, H2B, H3, and H4, is an
8

evolutionarily conserved, highly basic protein which has a protruding N-
terminal tail that can be modified post-translationally through processes not
limited to acetylation, phosphorylation, and methylation of specific lysine
residues, all of which can affect gene expression through direct modification
of chromatin structure or by serving as platforms for the binding of
transcription factors. While histone acetylation has been traditionally thought
to be transcriptionally activating, the role of phosphorylation and methylation
can be either activating or repressing, making the dynamic, combinatorial
regulation of these modifications an extremely complicated process.
Neurons in the brain continually change the genes they express in response
to synaptic activity, which is highly specific to the stimuli the organism
receives. The physical mechanism of “tagging” nucleosomes inside the
neuronal nucleus with post-translational histone modifications can be likened
to ascribing dynamic postal codes or addresses which can potentially serve to
dictate exactly “when” and “where” specific genes must be expressed: a
highly desirable ability in neurons given the static nature of genomic DNA.
Importantly, these “tags” or modifications are reversible and continually
changing such that at any given moment in time, each nucleosome may be
bearing different sets of modifications, with varying effects on gene
transcription. The epigenetic modification of histone molecules therefore
represents a rich, highly modifiable way to regulate gene transcription, which
is very attractive as a potential biochemical basis of information storage in the
nervous system.
The workhorses that are responsible for reading, writing, or erasing these
modifications may be collectively termed epigenetic or chromatin regulators.
9

As with the rules of physics, the rules of chromatin regulation follow the
general principle that for every action there is an equal but opposite reaction:
while Histone Acetyltransferases (HATs) tag histones with acetylation marks,

Histone Deacetylases (HDACs) remove acetyl groups. Likewise, Histone
Methyltransferases (HMTs) write methylation marks whereas Histone
Demethylases (HDMs) erase methyl groups from histones. Hundreds of other
proteins are able to “read” these nucleosomal tags and latch on to potentially
very specific areas of the genome, by virtue of PHD (Plant Homeo-Domain),
or bromo-domains which bind to methylated and acetylated histones
respectively.
Perhaps the most striking manifestation of the importance of chromatin
regulation is the clinical observation that defects in the enzymes responsible
for specifying histone modifications lead to specific and disastrous
consequences in cognitive function
29
. Notwithstanding evidence implicating
epigenetic enzymes in numerous neurological and neuropsychiatric
disorders
29,30
, an increasingly large number chromatin-modifying proteins is
now recognized to be crucial in learning and memory
31
. For instance, defects
in the enzyme CREB Binding Protein (CBP), a histone acetyltransferase and
transcriptional co-activator, cause Rubinstein-Taybi syndrome which is
characterized by severe cognitive disability
32
. On the other hand, mutations in
RSK2, a histone kinase that phosphorylates serine-10 of H3, are implicated in
Coffin-Lowry Syndrome, which is also marked by prominent mental
retardation. Finally, inactivating mutations in PHD Finger Protein 8 (PHF8), a
histone demethylase, also cause mental retardation
33

. These seemingly
different mechanisms target different histone marks, yet all of them exert a
10

highly specific neurological phenotype of intellectual disability, prompting the
search for the exact roles of histone acetylation, phosphorylation, and
methylation not only in neuronal development but also in adult neuroplasticity
and memory formation.
While it is important to identify the epigenetic enzymes which contribute to
cognitive function, the next step will be to understand how these enzymatic
machineries are regulated, what chromatin modifications are effectuated, and
ultimately, what gene targets they act upon to produce synaptic, network, and
behavioral changes. The study of the epigenetics of memory is further
complicated by the fact that each individual epigenetic enzyme appears to be
responsible for controlling distinct gene sets with varying temporal
specificities. Given that there exist hundreds of epigenetic regulators affecting
a potentially unlimited number of chromatin and downstream gene targets, the
study of the molecular mechanism of memory formation at least from the
chromatin viewpoint seems like an amazingly daunting task. It can be
conceptualized, however, that for a single neuron, one can systematically
organize the events that eventually lead to memory consolidation in a
temporally defined manner. Much like how the cellular correlate of memory,
Long Term Potentiation, has an early and a late phase, which parallels the
timeline of short-term memory acquisition and long-term memory
consolidation, epigenetic events can also be divided into early events that
occur within minutes of neuronal activation, and late events that take a longer
time to develop. In this model, early epigenetic events may underlie early-
phase LTP and short-term plasticity while late events may be responsible for
late-phase LTP and memory consolidation.
11

1.II.C. Arc – a regulator of synaptic memory
The leading mechanisms to explain memory at the level of neuronal synapses
are Long Term Potentiation (LTP) and Long Term Depression (LTD), which
are defined by the measurable, persistent increase (for LTP) or decrease (for
LTD) in the strength of the post-synaptic neuronal response in response to
varying stimuli
34,35
. These stimuli-dependent changes in synaptic weights are
thought to operate in every single one of the hundred trillion synapses
connecting the neurons in the brain and mediate all forms of activity-
dependent plasticity
36
. While LTP can be thought of as being a positive
regulator of a memory trace by strengthening response to a stimulus, the
seemingly counterintuitive LTD turns out to be crucial for memory as well,
apparently by producing a diminished response to a stimulus
37
. As such, both
early and late LTP and LTD are directly paralleled by many behavioral
correlates of learning and memory. And just like in short vs long-term
memory, only the late phase of LTP and LTD requires novel induction of gene
and protein synthesis
38,39
. Although many proteins have so far been
implicated in synaptic plasticity, one particular gene product, ARC (Activity-
Regulated Cytoskeletal protein), is situated at the nexus of LTP, LTD, and
protein synthesis-dependent memory.
Arc is a single copy gene that is speculated to have appeared late in evolution
as no reasonable homolog exists in invertebrate species though it is highly

conserved in vertebrates, suggesting that this unique gene may have highly
specific functions in higher organisms
40,41
. In excitatory pyramidal neurons of
the cortex and hippocampus where it is enriched, Arc expression has proven
to be one of the most tightly coupled events to neuronal activity and a plethora
12

of behavioral paradigms ranging from simple exposures to sound
42
, emotional
and fear memory
43
, spatial exploration
44
, taste memory
45
, visual stimuli
46
, and
novel environments
47
. Remarkably, total deletion of the Arc gene in the mouse
completely abrogated the ability to form long-term memories without
disrupting short-term memory formation
48
. Subsequently, it was found that Arc
expression was related to virtually all cases of long-term memory formation,
whether it be in birds
49

, rodents
50
, and even humans
51
. Arc may be the only
gene currently known to be induced by neuronal activity and required for all
forms of LTP and LTD: early Arc synthesis is required for early LTP
expression, whereas sustained Arc synthesis is needed for late-phase LTP
maintenance
52,53
; on the other hand, Arc is also required for LTD
54,55
. This
bidirectional involvement suggests that Arc may be a master regulator of
activity-dependent changes in synaptic weights, regardless of the direction of
the change
56
.
In view of its tight coupling with neural activity and its undisputed role in
consolidating long-term memories, much research has focused on
determining the exact molecular functions of Arc and the pathways that
govern Arc expression and regulation. Consistent with its highly crucial role in
memory consolidation, Arc gene expression is tightly controlled at many
levels, including transcription, mRNA degradation, translation, and protein
degradation
57
. Although we now know that diverse mechanisms including
NMDA receptors
58
, the ERK/MAPK pathway

59
, metabotropic Glutamate
receptors
60
, and Voltage-Gated Calcium Channels
61
may all play a role in
regulating Arc expression, the precise signaling cascades that connect
13

synaptic activity with the actual transcription of Arc gene are complex and not
well understood.
Given the complexity in its regulation, there is a high likelihood that the
dynamic, experience-dependent increase in Arc expression may be mediated
by epigenetic mechanisms. Although one study has shown that DNA
methylation may be responsible for shutting down Arc expression at the 24-
hour time point after electroconvulsive seizures
62
, the effects of chromatin
modification on Arc transcription have not yet been elucidated. Interestingly,
recent findings indicate that the rapid induction of Arc transcription that occurs
within minutes of neuronal activity is controlled by RNA Polymerase II which is
“paused” at the promoter region of the Arc gene
63
. In Chapter 2 of this thesis,
I propose a mechanism by which early Arc transcription is regulated by an
epigenetic mechanism involving a dual-function chromatin-modifying complex
and neuronal activity-dependent post-translational histone modification.
While it is important to study how such an important gene is regulated at the
level of transcription, it is the translated ARC protein is ultimately responsible

for its function. Arc translation is also very complex and is known to happen at
multiple sites including dendrites
64
. The fact that emotional memories are
highly Arc-dependent is reflected by the complex regulatory mechanisms
involved in ARC protein translation, being modulated by dopamine and beta-
adrenergic receptor signaling
65
. Indeed, infusions of beta-adrenergic agonists
into the amygdala increases ARC protein levels in the hippocampus and
enhanced the performance of mice trained under an inhibitory avoidance
paradigm, an emotionally relevant memory
66
.