development of a viral toolbox to study the role of m6a rna … · the hippocampus following a...
TRANSCRIPT
Development of a Viral Toolbox to Study the Role of m6A
RNA Methylation in Memory
by
Colleen Gillon
A thesis submitted in conformity with the requirements
for the degree of Master of Science
Department of Physiology
University of Toronto
© Copyright by Colleen Gillon (2016)
ii
Development of a Viral Toolbox to Study the Role of m6A RNA
Methylation in Memory
Colleen Gillon
Master of Science
Department of Physiology
University of Toronto
2016
Abstract
Memory relies critically on local synthesis of proteins at synapses. How cells are able to
precisely and independently coordinate memory-related translation at each of their synapses is
not fully understood. Just as transcription is modulated by epigenetic modifications of DNA
during memory formation, local translation may be regulated by mRNA modifications, like N6
methylation of the adenosine base (m6A). In this project, we investigate the m6A pathway and
find that the demethylase Fto is highly expressed in the mouse brain, and regulated in the CA1 of
the hippocampus following a contextual memory task, along with several other pathway
components and overall levels of m6A in mRNA. To probe the role of m6A in memory
formation, we then develop herpes simplex virus (HSV) vectors to rapidly and selectively
manipulate Fto levels in cells, including the first HSV-mediated CRISPR/Cas9 knockdown
vector, which will increase our ability to flexibly target multiple genes in vivo.
iii
Acknowledgments
This thesis project represents over a year of important growth in my scientific research skills
which would not have been possible without the humour, advice and patience of many people.
I would first like to thank my supervisor, Dr. Sheena Josselyn, for her guidance and support, and
for welcoming me into a wonderful lab full of brilliant researchers, who have generously shared
their knowledge and skills with me over the past two years. Among the members of the Josselyn
and Frankland labs, I would particularly like to thank Dr. Jonathan Epp, Dr. Justin Kenney,
Dr. Valentina Mercaldo, Dr. Asim Rashid and Dr. Gisella Vetere who answered so many of my
questions, laughed at my terrible jokes, taught me numerous techniques, and readily shared their
advice and experience with respect to grad life and research. Above all, however, I am most
grateful to Dr. Brandon Walters, who not only devised this project, but mentored me through
each step, training me in molecular biology, helping me when I made mistakes, and discussing
all aspects of my project with me. With his love of science, patience, and particularly his delicate
sense of humour, he has been an outstanding mentor. I owe him a heavy debt, and only hope that
the Nutella martinis which he will now be receiving from me yearly will be sufficient repayment.
I would also like to thank my committee members, Dr. Zhengping Jia and Prof. John Yeomans
who took the time to guide and advise me through each phase of my Master’s studies.
I've been fortunate to share this experience with my excellent comrades in arms, Lina, Bonny,
Dano, Frances and Patrick. Together, we navigated the murky waters of grad school, celebrating
the highs, and complaining about the lows. I will definitely miss laughing and hunting for free
food with them. Of course, I would not even be here without my family. Over the past few
decades, my Mom and Dad have invested an immense amount of time and energy preparing me
for life, encouraging and nudging me through the tough times, and listening to me drone on
about my research and failed Westerns. I feel very lucky to have them by my side, as well as
Evan and Chantal who this year side-tracked all of our lives by introducing a very cute, chubby-
cheeked distraction. I blame my baby nephew for any delays in finishing this thesis, as he is
quite adorable and cannot yet contradict me. Lastly, I am very thankful to Hubert, who has
encouraged me every step of the way, been subjected to detailed accounts of every successful
iv
and failed experiment, and read every word of this thesis (except the references, I think). He kept
me fed, hydrated and supported through the most difficult stretches of my Master’s, and I
couldn't ask for a better partner.
My Master’s research was generously funded by a grant from the Canadian Institutes of Health
Research, a graduate scholarship from the National Science and Engineering Research Council
of Canada, a SickKids Restracomp award, and a stimulus award from the Department of
Physiology at the University of Toronto.
v
Contributions of Collaborators
Several members of the Josselyn laboratory were involved in producing the work presented in
this thesis. Their contributions are listed here.
Figure 3.1
Brandon Walters: Designed qPCR primers for Alkbh5 and Fto.
Figure 3.2
Brandon Walters/Matthew Yip: Ran experiments for Alkbh5 and Fto.
Brandon Walters: Helped analyse the data.
Figure 3.3
Brandon Walters/Matthew Yip: Ran experiments for m6A in total and mRNA enriched samples.
Figure 3.4
Brandon Walters: Designed the sgRNA targeting Fto in the Cas9 knockdown virus. Prepared and
analysed some of the samples to validate the shRNA and Cas9 knockdown viruses in cell culture.
Rachael Neve lab (MIT): Cloned the Cas9 virus plasmids, and packaged them into virus
particles.
Mika Yamamoto: Packaged the Fto overexpression and shRNA knockdown virus plasmids into
virus particles.
Figure S1
Brandon Walters/Matthew Yip: Ran experiments for Alkbh5 and Fto.
Brandon Walters: Helped analyse the data.
vi
Table of Contents
Acknowledgments.......................................................................................................................... iii
Contributions of Collaborators ........................................................................................................v
List of Tables ...................................................................................................................................x
List of Figures ................................................................................................................................ xi
List of Appendices ........................................................................................................................ xii
List of Abbreviations ................................................................................................................... xiii
Chapter 1 Introduction .....................................................................................................................1
1.1 Summary of the Research Question .....................................................................................1
1.2 The Hippocampus ................................................................................................................2
1.2.1 Memory in the hippocampus ...................................................................................2
1.2.2 Hippocampal anatomy .............................................................................................3
1.2.3 CA1 and spatial memory .........................................................................................4
1.3 Local Translation Regulates Memory Formation ................................................................5
1.3.1 Synaptic plasticity in memory .................................................................................6
1.3.2 Gene transcription in memory .................................................................................7
1.3.3 Gene translation in memory .....................................................................................9
1.3.4 Local gene translation in memory ............................................................................9
1.3.4.1 Regulation of local mRNA transport in memory ....................................12
1.3.4.2 Regulation of local mRNA translation in memory ..................................13
1.4 m6A mRNA Methylation as a Potential Regulator of Local Gene Translation ................15
1.4.1 Common mRNA modifications .............................................................................15
1.4.2 m6A modification of mRNA .................................................................................16
1.4.3 m6A pathway .........................................................................................................18
vii
1.4.3.1 m6A methyltransferases ..........................................................................18
1.4.3.2 m6A demethylases ...................................................................................21
1.4.3.3 m6A binding proteins ..............................................................................23
1.4.4 m6A in the brain ....................................................................................................29
1.5 Study Rationale ..................................................................................................................29
1.6 Hypothesis..........................................................................................................................30
1.7 Review of Chosen Gene Manipulation Methods ...............................................................30
1.7.1 Herpes Simplex Virus (HSV) ................................................................................30
1.7.2 Cas9-based Gene Manipulation .............................................................................31
Chapter 2 Materials and Methods ..................................................................................................35
2.1 Mice ...................................................................................................................................35
2.2 Behavioural Training .........................................................................................................35
2.3 Cell Culture ........................................................................................................................36
2.4 Sample Analysis.................................................................................................................36
2.4.1 RNA isolation and complementary DNA (cDNA) synthesis ................................36
2.4.2 qPCR ......................................................................................................................37
2.4.3 m6A ELISA ...........................................................................................................38
2.5 HSV Viral Vector Design ..................................................................................................38
2.5.1 Design of the FTO overexpression HSV vector (p1005-FTO) ..............................38
2.5.2 Design of Fto sgRNAs ...........................................................................................39
2.5.3 Design of the Fto and control sgRNA Cas9 vectors (Cas9-Fto and Cas9-ctrl) .....39
2.5.4 Design of the Fto and scrambled shRNA vectors (shRNA-Fto and shRNA-
scr)..........................................................................................................................40
2.6 Statistical Analysis .............................................................................................................41
Chapter 3 Results ...........................................................................................................................42
3.1 m6A Pathway Components Show Distinct Tissue Expression Patterns ............................42
viii
3.2 m6A Pathway Component Levels are Regulated During Learning ...................................42
3.3 m6A Levels in mRNA are Regulated During Learning ....................................................45
3.4 In Vitro Validation of HSVs ..............................................................................................45
3.4.1 p1005-FTO overexpression virus ..........................................................................46
3.4.2 Cas9-Fto knockdown virus ....................................................................................46
3.4.3 shRNA-Fto knockdown virus ................................................................................48
Chapter 4 Discussion .....................................................................................................................49
4.1 Fto is Enriched in the Brain ...............................................................................................49
4.2 The m6A Pathway is Extensively Regulated During Memory Formation ........................51
4.2.1 Regulation of m6A by methyltransferase and demethylase levels ........................51
4.2.2 Downstream effects of m6A regulation on transcript processing ..........................53
4.2.3 Biphasic regulation of m6A and other memory-related processes ........................54
4.3 m6A is Specifically Regulated in mRNA During Memory Formation .............................55
4.4 Speculative Model of the Role of m6A in Regulating Local mRNA Translation
During Memory Formation ................................................................................................56
4.5 Overview of the Potential Role of m6A in Memory..........................................................60
4.6 Development of HSV Gene Manipulation Tools to Probe the Role of m6A in
Memory ..............................................................................................................................61
4.6.1 Overexpression and knockdown vectors ...............................................................61
4.6.2 Cas9-based HSV ....................................................................................................63
4.7 Limitations .........................................................................................................................64
4.7.1 Missing pieces in the m6A pathway puzzle ...........................................................64
4.7.2 m6A pathway regulation may not be occurring primarily in principal neurons ....64
4.7.3 m6A regulation in other brain regions during learning .........................................65
4.7.4 Measuring mRNA levels versus protein levels ......................................................65
4.7.5 Limitations related to the use of HSVs ..................................................................66
4.7.5.1 Toxicity of HSVs in cell culture ..............................................................66
ix
4.7.5.2 Limited infection rate using HSVs ..........................................................67
4.7.6 Effects on m6A levels of targeting Fto are not restricted to mRNA .....................67
4.8 Future Aims .......................................................................................................................68
4.8.1 Test whether manipulating FTO and m6A levels impacts memory ......................69
4.8.2 Identify mRNA transcripts showing enriched m6A levels during memory
formation ................................................................................................................70
4.8.3 Establish whether m6A is playing a role in memory at the synapse .....................71
Chapter 5 Conclusion .....................................................................................................................73
References ......................................................................................................................................75
Appendix ........................................................................................................................................92
x
List of Tables
Table 1.1. Role of m6A methyltransferase complex components ................................................ 26
Table 1.2. Role of m6A demethylases .......................................................................................... 27
Table 1.3. Role of m6A binding proteins ..................................................................................... 28
xi
List of Figures
Figure 1.1. The tri-synaptic circuit ................................................................................................. 4
Figure 1.2 Two options for fine-tuned regulation of local gene translation ................................. 11
Figure 1.3. The m6A pathway ...................................................................................................... 17
Figure 1.4. Gene editing with Cas9 and dCas9 ............................................................................. 34
Figure 2.1. Virus designs .............................................................................................................. 40
Figure 3.1. m6A pathway components show distinct tissue expression patterns ......................... 43
Figure 3.2. m6A pathway components are regulated during learning .......................................... 44
Figure 3.3. m6A levels in mRNA are regulated during learning .................................................. 46
Figure 3.4. In vitro validation of HSVs ........................................................................................ 47
Figure 4.1. Speculative model of the role of m6A in regulating local mRNA translation during
memory formation ........................................................................................................................ 59
Figure S1. m6A pathway components levels in context-only or shock-only controls ................. 92
xii
List of Appendices
Figure S1. m6A pathway components levels in context-only or shock-only controls.
xiii
List of Abbreviations
ACTB beta-actin
ALK alpha-ketoglutarate-dependent dioxygenase
ALKBH5 alkB homolog 5
ANOVA analysis of variance
ARC activity-regulated cytoskeleton-associated protein
BDNF brain-derived neurotrophic factor
CA1 cornu Ammonis area 1
CA3 cornu Ammonis area 3
CamKII calcium/calmodulin-dependent protein kinase II
CamKIIA alpha subunit of CamKII
cAMP cyclic adenosine monophosphate
CaN calcineurin
CCAC Canadian Council on Animal Care
cDNA complementary DNA
CFC contextual fear conditioning
CPEB1 cytoplasmic polyadenylation element binding protein 1
CREB cAMP responsive element binding protein
CRISPR clustered regularly interspaced short palindromic repeats
CT threshold cycle
DG dentate gyrus
DNA deoxyribonucleic acid
EC entorhinal cortex
ELAVL1 ELAV-like RNA binding protein 1
ELISA enzyme-linked immunosorbent assay
E-LTP early phase LTP
f6A N6-formyladenosine
FMRP fragile X mental retardation protein
FTO fat mass and obesity-associated protein
GAPDH glyceraldehyde 3-phosphate dehydrogenase
xiv
hm6A N6-hydroxymethyladenosine
HPRT hypoxanthine-guanine phosphoribosyltransferase
IEG immediate early gene
L-LTP late phase LTP
LTD long-term depression
LTP long-term potentiation
m5C 5-methylcytosine
m6A N6-methyladenosine
m7G N7-methylguanosine
MAP2 microtubule-associated protein 2
MEF mouse embryonic fibroblasts
mESC murine embryonic stem cell
METTL14 methyltransferase-like 14
METTL3 methyltransferase-like 3 (or MT-A70)
miRNA micro RNA
MIT Massachusetts Institute for Technology
MRI magnetic resonance imaging
mRNA messenger RNA
N2a neuro2a
NGF nerve growth factor
NIH National Institutes of Health
NMDA N-methyl-D-aspartate
PCR polymerase chain reaction
PpI protein phosphatase I
PRP plasticity-related protein
PSD postsynaptic density
qPCR quantitative real time PCR
RELN reelin
RNA ribonucleic acid
RNP ribonucleoprotein particles
shRNA small hairpin RNA
sgRNA single guide RNA
xv
Sub subiculum
tRNA transfer RNA
UTR untranslated region
WTAP Wilms' tumour 1-associating protein
YTH YT521-B homology
ZBP1 zip-code-binding protein 1
ZIF268 zinc finger protein 225 (or Egr-1)
1
Chapter 1
Introduction
Chapter 1
1.1 Summary of the Research Question
Memory formation requires strengthening and weakening connections between different neurons
in the brain, a process called synaptic plasticity known to depend on both gene transcription and
translation (Bourtchuladze et al., 1998; Castellucci et al., 1970, 1986; Frey et al., 1988, 1996;
Mayford et al., 2012). It was long thought that gene translation occurs mostly in the cell soma,
and that during memory formation, plasticity-related proteins (PRP) are trafficked to their
synaptic targets to enable plasticity (Frey & Morris, 1997; Redondo & Morris, 2011). How
neurons, with their highly complex dendritic arbours, are able to specifically target proteins only
to those synapses undergoing plasticity remains unclear. The discovery that protein synthesis
also occurs in dendrites and spines, and that this local translation is necessary for memory
formation suggested an additional mechanism for regulating local plasticity, namely regulating
local translation (Miller et al., 2002; Ouyang et al., 1999). Indeed, trafficking transcripts to
synapses instead of proteins may be advantageous for neurons, as it allows them to produce
proteins only when and where they are needed in the cell. However, a similar question emerges,
namely how are neurons able to independently and precisely regulate local translation of
messenger ribonucleic acids (mRNA) at each of their synapses? The mechanisms regulating
local translation would have to include a stimulus-induced, dynamic and flexible pathway. In
particular, such a pathway would have to allow for a large number of differentiable signals, to
enable fine-tuned control over local translation at the individual mRNA transcript level. Lastly,
to ensure that each transcript may be processed independently of others, such a signal would
likely have to be present in the actual transcripts themselves. The mechanisms identified to date
account well for the broad regulation of local translation, but do not provide transcript-level
precision. In light of this, the N6-methyl adenosine (m6A) modification that occurs in RNA
emerges as a promising candidate, due to its wide prevalence, flexibility and dynamic nature, for
the fine regulation of local protein synthesis in memory (Fu et al., 2014a).
2
1.2 The Hippocampus
As a first step in investigating the potential role of m6A in memory, I first review the key brain
regions involved in learning and memory. Implicit memory, which includes the abilities to
subconsciously remember procedures, perceptions and motor tasks, is principally supported by
the striatum, neocortex and cerebellum respectively (Squire, 2004). Declarative memory, the
ability to remember facts and events, relies primarily on the hippocampus, and neighbouring
regions, including the amygdala and fornix, as well as the entorhinal (EC), perirhinal and
parahippocampal cortices (Simons & Spiers, 2003). As declarative memories age however, they
become increasingly independent of the hippocampus, depending mainly on neocortical and
prefrontal regions (for review, see Frankland & Bontempi, 2005; Simons & Spiers, 2003).
Among these key memory regions, the hippocampus has been most widely studied, not only
because its dysfunction results in profound memory impairments (Morris et al., 1982;
Rudy et al., 2002; Scoville & Milner, 1957), but also because it constitutes a central hub for
information flow in the brain during the acquisition of declarative memories (Battaglia et al.,
2011; Wheeler et al., 2013). The hippocampus thus emerges as a promising candidate region in
which to begin our investigation of the role of m6A in memory.
1.2.1 Memory in the hippocampus
The first strong clinical evidence for the role of the hippocampus and neighbouring regions in
memory came from patients suffering from epilepsy or schizophrenia who, in severe cases, had
been treated with surgical brain tissue resections of these commonly epileptogenic regions. The
most famous patient to undergo bilateral resection from these regions was H. M. (Milner 1968;
Scoville & Milner 1957). Following surgery, he suffered severe anterograde, and graded
retrograde amnesia, losing the ability to form new memories, or recall memories from several
years preceding his surgery. His case was particularly severe, as the region resected included not
only large portions of the hippocampus, but also the subiculum (Sub), amygdala and EC (Squire,
1992; Squire & Zola-Morgan 1991). By comparing the regions resected in patients suffering
from permanent post-surgical amnesia, Scoville & Milner (1957) identified the hippocampus as
the common factor. Squire et al. (1990) arrived at the same conclusion using magnetic resonance
imaging (MRI) in patients having developed amnesia following a stroke or other brain injury.
3
In light of the early clinical findings pointing strongly to a critical role for the hippocampus in
declarative learning, Hirsh (1974) reviewed hippocampus memory lesion studies in animals and
was the first to propose a specific role for the hippocampus in contextual memory. Since then,
the hippocampus has been shown to be necessary not only for the acquisition of contextual
memory (Fanselow, 2000; Martin & Clark, 2007; Rudy et al., 2002), but also for spatial memory
tasks (Girardeau et al., 2009; Morris et al., 1982) and trace conditioning (McEchron &
Disterhoft, 1999; Misane et al., 2005; Solomon et al., 1986). As these studies show, inhibiting or
disrupting activity in the hippocampus during or shortly after learning significantly impairs
memory recall performance. Together, these findings strongly support the choice of the
hippocampus as a starting point for investigating potential mechanisms underlying memory
formation, like m6A RNA methylation.
1.2.2 Hippocampal anatomy
The ability of the hippocampus to coordinate the complex task of storing and retrieving
memories may in part be due to its highly organized structure which is well conserved from
rodents to humans (Simons & Spiers, 2003). This feature has enabled researchers to distinguish
the different roles of hippocampal subregions in memory. In investigating a potential role for
m6A in memory, it is helpful to focus on specific subregions, as the regulation of m6A may
differ depending on the role a subregion plays in memory, and thus the function for which m6A
may be harnessed (van Gemert et al., 2009).
Located in the medial temporal lobe, the hippocampus is most commonly subdivided into cornu
Ammonis fields (CA1 and CA3) and the dentate gyrus (DG) (Fig. 1.1). The primary connections
between these regions are comprised in a tri-synaptic circuit, first described by the Spanish
neuroscientist Santiago Ramón y Cajal in 1893, and later refined by Swanson et al. (1978). The
principal input of the circuit begins with the neighbouring EC, which receives input from cortical
sensory association areas, and projects onto the DG via the perforant pathway. DG granule cells
then synapse onto CA3 pyramidal cells via the mossy fiber pathway. Lastly, CA1 receives inputs
from CA3 cells via the Schaffer collaterals (Fig. 1.1). CA1, acting as the main output of the
hippocampus, then projects back onto the EC and the Sub. In addition to these commonly
described connections, many additional inputs and outputs exist, contributing to the complexity
of the hippocampus (for review, see Amaral & Witter, 1989).
4
Through their different structures and composition, CA1, CA3 and the DG each contribute
differently to memory formation. Whereas the DG is thought to be particularly important for
ensuring that similar memories are stored distinctly, a process called pattern separation
(Deng et al., 2013; Leutgeb et al., 2007), CA3 is thought to function as an autoassociative
network, binding together different components of a memory. This property of CA3 is thought to
underlie pattern completion, a process that allows an incomplete memory cue to trigger recall of
an entire memory (Lisman, 1999; Kesner, 2007). Last in the circuit, CA1 encodes cognitive
maps of new environments, and plasticity in this region is particularly critical for spatial memory
formation (Moser et al., 2008; O’Keefe & Dostrovsky, 1971; Tsien et al., 1996; Yiu et al.,
2011). For its critical role in spatial memory and position as the main output of this circuit, we
have chosen to focus on CA1 in investigating the role of m6A in memory in the hippocampus.
Figure 1.1. The tri-synaptic circuit. Sagittal section of the rodent hippocampus drawn by Santiago Ramón y Cajal.
Arrows show information flow through the tri-synaptic circuit: EC → DG → CA3 → CA1 → Sub/EC (Adapted
from Ramón y Cajal, 1954).
1.2.3 CA1 and spatial memory
Evidence from both human and rodent studies has shown the CA1 subregion of the
hippocampus, and particularly the dorsal region (Moser et al., 1993), to be particularly important
in memory. Among patients who have suffered hippocampal damage, one patient (R. B.)
5
sustained a bilateral ischemic injury largely restricted to CA1. Even with only CA1 affected,
R. B. displayed a moderately severe anterograde declarative memory impairment, wherein he
retained most memories from before his injury, but showed severe, although not complete,
amnesia for events following his injury (Zola-Morgan et al., 1986). CA1 neurons have since
been shown to be particularly affected in Alzheimer’s disease (van Hoesen et al., 1999),
providing further evidence for a critical role for CA1 in memory in humans.
In rodents, the cognitive maps encoded by place cells in dorsal CA1 are necessary for spatial
memory, and inhibiting their reactivation after learning impairs long-term memory (Carr et al.,
2011; Jadhav et al., 2012; Wilson & McNaughton, 1994). Deng et al. (2013) showed that during
spatial memory recall, many of the cells active during learning are reactivated in dorsal CA1 in
particular, indicating that these neurons are critical components of the memory. Ablation of only
CA1 cells or disruption of their firing is sufficient to impair performance in spatial memory tasks
(Auer et al., 1989; Lenck-Santini et al., 2001; Volpe et al., 1992). Based on these findings, it is
clear that CA1, and dorsal CA1 in particular, plays a key role in spatial memory.
A commonly used spatial task is contextual fear conditioning (CFC) in which rodents learn to
associate an aversive stimulus, typically a foot shock, with a specific context. Similarly to other
spatial tasks, CFC induces synaptic plasticity in CA1 (Hall et al., 2000; Levenson et al., 2002)
and is impaired by hippocampal lesions including CA1 (Chen et al., 1996; Maren et al., 1997).
Since CFC can be learned after a single session, it is particularly conducive to studying
mechanisms underlying memory formation, as these can be temporally restricted to the period
immediately following the single learning session (Sekeres et al., 2012; Wheeler et al., 2013).
For these reasons, we selected CFC as a learning paradigm with which to study the potential role
of m6A in memory formation in dorsal CA1.
1.3 Local Translation Regulates Memory Formation
During formation of memories such as context fear memories, neurons in CA1 and other regions
of the hippocampus undergo synaptic plasticity. Through this process which relies on both gene
transcription and protein translation, synaptic connections between neurons are strengthened or
weakened to encode the new memory (Bourtchuladze et al., 1998; Castellucci et al., 1970, 1986;
6
Frey et al., 1988, 1996; Mayford et al., 2012). Considering that each pyramidal neuron in CA1 is
estimated to have over 30,000 synapses (Megías et al., 2001), the task of ensuring that only those
synapses involved in a new memory are strengthened or weakened is not trivial. The ability of
neurons to do this is partially accounted for by their ability to engage in local protein synthesis,
whereby genes are translated locally in dendrites and spines. However, how cells are able to
traffic individual transcripts to specific dendritic segments or locally label them for translation or
degradation remains an important gap in our understanding of the mechanisms underlying
synaptic plasticity, and memory formation.
1.3.1 Synaptic plasticity in memory
In addition to his early description of the tri-synaptic circuit, Ramón y Cajal (1894) was among
the first to have the insight that the strengthening of connections between neurons might underlie
memory formation in the brain. Almost half a century later, Donald Hebb (1949) formulated his
now famous theory of learning in which he proposed that when a neuron A repeatedly
contributes to the firing of a neuron B, the connection between these neurons is strengthened.
Research in invertebrate systems investigating the relatively simple synaptic connections
underlying reflexes like the gill-withdrawal reflex in Aplysia confirmed these insights, showing
that learning relies on synaptic plasticity, the strengthening and weakening of synaptic
connections between neurons (Castellucci et al., 1970; Kandel & Tauc 1965; Mayford et al.,
2012). Shortly thereafter, the most widely accepted model for activity-dependent strengthening
of synapses, long-term potentiation (LTP), emerged. Indeed, in a critical experiment, Bliss &
Lømo (1973) were able to induce long-lasting potentiation of DG synapses in anaesthetized
rabbits through high-frequency stimulation of the perforant pathway. In other words, after this
potentiation stimulation, inputs from presynaptic neurons excited postsynaptic neurons more
strongly, and this increase was maintained for a prolonged period of time. Almost twenty years
later, Dudek & Bear (1992) showed that the counterpart to LTP, long-term depression (LTD) in
which synaptic connections are weakened, could be produced using low-frequency stimulation.
The discovery of LTP and LTD has provided a useful model for investigating the mechanisms
underlying memory formation at the cellular level. For simplicity, I will focus here on LTP,
which has been most studied for its relationship to memory formation.
7
As pointed out by Lynch (2004), the mechanisms at play in LTP closely parallel those observed
in memory. First, LTP is most readily induced in the hippocampus, a region critically involved in
memory. Second, inhibiting LTP in the hippocampus impairs learning of hippocampal-dependent
tasks. Third, the high-frequency stimulation used to induce LTP resembles the theta rhythms that
occur in the brain naturally during learning. Fourth, similar biochemical changes occur in cells
after LTP as during learning, including gene transcription and translation. Lastly, just as
memories can be stored temporarily or long-term, LTP has an early phase (E-LTP) lasting two to
three hours, and a late phase (L-LTP) lasting hours to weeks (Lynch, 2004). While there are
differences between L-LTP and the mechanisms through which synaptic plasticity occurs
naturally in the brain during learning (for discussion of this topic, see Martin et al., 2000), these
strong parallels make L-LTP an excellent candidate for studying memory at the cellular level.
Typically, as in CA1 where it has been most extensively studied, L-LTP requires activation of N-
methyl-D-aspartate (NMDA) receptors at the postsynaptic terminal of stimulated neurons.
Calcium flowing into the terminal through NMDA channels initiates the signaling cascades
which lead to potentiation or depression of a synapse, through translation and activation of a
large number of PRPs and molecules (Bliss & Collingridge, 1993; for review, see Malenka &
Bear, 2004; for review in CA1, see Luscher & Malenka, 2012). Gene transcription and
translation are key components of these cascades and required for both L-LTP and memory
formation (Bourtchuladze et al., 1998; Castellucci et al., 1986; Frey et al., 1988, 1996).
Although it is clear that these processes are tightly regulated, how neurons are able to coordinate
local translation in particular to ensure that only the appropriate synapses are strengthened or
weakened during memory formation is unknown.
1.3.2 Gene transcription in memory
Similar to gene translation, gene transcription is critical for L-LTP and long-term memory
(lasting at least 24 hours). Inhibiting RNA synthesis, or gene transcription, has been shown to
block L-LTP in Aplysia and rat hippocampal slices (Castellucci et al., 1986; Frey et al., 1996)
and impair memory for spatial and contextual tasks when the hippocampus is targeted (Da Silva
et al., 2008; Igaz et al., 2002; Thut & Lindell, 1974). Two time windows have been identified
during which transcription is required in CA1 for memory formation: the first at the time of
training, and the second three to six hours after training (Igaz et al., 2002). This biphasic pattern
8
is a key feature of memory formation, as it is also seen in immediate early gene (IEG) activity,
gene translation and LTP (Ramírez-Amaya et al., 2005).
Two general mechanisms for regulating gene transcription during memory have been identified.
The first is the translation or activation of IEGs, many of which are transcription factors, which
bind to promoter sequences of PRP genes like brain-derived neurotrophic factor (Bdnf)1 and
nerve growth factor (Ngf) (Barco et al., 2005; Castrén et al., 1993), and stimulate their
transcription. These IEGs notably include cyclic adenosine monophosphate (cAMP) responsive
element binding protein (Creb), zinc finger protein 225 (Zif268, also known as Egr-1) and
activity-regulated cytoskeleton-associated protein (Arc) (Lynch, 2004). Knocking down or
blocking activation of IEGs in the hippocampus impairs both L-LTP and memory formation
(Bourtchuladze et al., 1994; Bozon et al., 2003; Guzowski et al., 2000; Pittenger et al., 2002).
Although efficient, this mechanism for regulating gene transcription during memory lacks
flexibility and gene-specificity as it relies on hard-coded promoter sequences that are common to
a large number of genes.
More recently, a second more flexible and gene-specific mechanism for regulating gene
transcription during memory was uncovered, namely epigenetic modifications. Indeed, following
learning, dynamic changes are seen in the deoxyribonucleic acid (DNA) methylation patterns
around PRP genes. Through these changes in methylation patterns, neurons are able to flexibly
and quickly regulate the expression of individual PRPs, like Bdnf, reelin (Reln), protein
phosphatase I (PpI), Arc and calcineurin (CaN) (Lubin et al., 2008; Miller et al., 2010; Miller &
Sweatt, 2007; Zovkic et al., 2013). Blocking DNA methylation by inhibiting methyltransferase
activity in the hippocampus impairs synaptic plasticity and learning (Levenson et al., 2006;
Miller & Sweatt, 2007), suggesting that epigenetic modifications are critical for regulating gene
transcription in memory. Importantly, the existence of two parallel mechanisms regulating gene
transcription, and in particular the flexibility and reversibility of epigenetic markers, may provide
a hint as to how gene translation is regulated during memory.
1 Following standard mouse nomenclature, in this document, italics (e.g., Bdnf) are used for genes or mRNA
transcripts, and uppercase letters (e.g., BDNF) are used for proteins.
9
1.3.3 Gene translation in memory
Gene translation is critical to memory formation and inhibiting protein synthesis in the
hippocampus impairs L-LTP (Frey et al., 1988; Kelleher et al., 2004; Stanton & Sarvey 1984),
L-LTD (Manahan-Vaughan et al., 2000; Sajikumar & Frey, 2003) and memory. This is
particularly the case during two critical time windows which it shares with IEG activation, gene
transcription and LTP, namely one occurring at the time of training, and the other three to six
hours after (Bourtchuladze et al., 1998; Quevedo et al., 1999).
Protein synthesis was long thought to occur only in the soma of neurons. This raises an important
question for memory formation: with their thousands of synapses, how are neurons able to traffic
newly synthesized PRPs only to the specific synapses to be potentiated or depressed? One
candidate mechanism proposed by Frey & Morris (1997) is synaptic tagging, whereby recently
activated synapses may be able to preferentially sequester PRPs travelling non-specifically in the
neuron (for discussion of this topic, see Redondo & Morris, 2011). More recently, however, the
discovery that several PRPs are synthesized locally in dendrites and synapses suggests that a
more important question may be how neurons coordinate local translation of proteins during
memory. Indeed, trafficking mRNA to synapses instead of proteins provides several advantages.
First, it saves neurons energy by ensuring that proteins are only translated when needed. Second,
it reduces the risk of proteins producing off-target effects during transport to their dendritic
targets. Last, if a localization or translation signal is present in mRNA, it can occur without being
translated, and thus two mRNAs with different signals can produce the same functional protein.
Nonetheless, the same question remains: how are mRNA transcripts targeted to or translated at
only those synapses involved in a memory?
1.3.4 Local gene translation in memory
The first evidence suggesting local gene translation occurs in neurons was the discovery of
translational machinery, particularly polyribosomes, in dendrites and beneath synaptic junctions
in the hippocampus (Steward & Levy, 1982; Steward & Reeves, 1988). Tiedge & Brosius (1996)
later confirmed that dendrites were competent for protein synthesis, as they contain critical
translation elements, like transfer RNAs (tRNAs), initiation and elongation factors, and
cotranslational signal recognition components. More recently, it was shown that following high-
frequency stimulation, there is an influx of polyribosomes from dendritic shafts into spines in the
10
hippocampus, including CA1 (Bourne et al., 2007; Ostroff et al., 2002). Furthermore, spines that
have received an influx of polyribosomes have enlarged postsynaptic densities (PSDs) as
compared to those that have not, indicating synaptic potentiation at these spines (Ostroff et al.,
2002). Moreover, many mRNA transcripts, including several PRPs, have been found in high
levels in dendrites, spines or synaptosomes, including the alpha subunit of calcium/calmodulin-
dependent protein kinase II (CamKIIA) (Mayford et al., 1996; Paradies & Steward, 1997), Arc
(Link et al., 1995), beta-actin (Actb) (Tiruchinapalli et al., 2003), microtubule-associated
protein 2 (Map2) (Garner et al., 1988), and NMDA receptor subunits (Grooms et al., 2006;
Miyashiro et al., 1994).
Although these findings strongly indicate that dendrites and spines are competent for translation,
they do not demonstrate that local gene translation does in fact occur during learning. In 1992,
Torre & Steward were able to show that local protein synthesis does occur in live cultured
dendrites isolated from their cell bodies. Furthermore, in vivo in the hippocampus, synaptic
activity or high-frequency stimulation enriches levels of transcripts such as Arc and CamKIIA in
dendrites, as well as local protein synthesis (Havik et al., 2003; Link et al., 1995). The increase
in CAMKII protein levels that follows high-frequency stimulation depends on local protein
synthesis as it is blocked even when synthesis is only inhibited at the dendrites (Ouyang et al.,
1999). Additionally, selective localization of Arc to activated dendritic segments in the DG,
depends on NMDA receptor activity providing a further link to memory and L-LTP
(Steward et al. 1998; Steward & Worley, 2001). Finally, in investigating the functional role of
local protein synthesis, Bradshaw et al. (2003) specifically inhibited protein synthesis at apical or
basal dendrites, and showed that L-LTP was only impaired at the targeted locations. Together,
these findings strongly suggest that L-LTP requires local gene translation.
Confirming that local protein synthesis is required in vivo for memory formation has proved
harder to tackle technically. Nonetheless, Ainsley et al. (2014) recently showed that CFC leads
to rapid association of dendritic mRNA with local ribosomes. Most notably however,
Miller et al. (2002) disrupted dendritic localization of CAMKIIA during learning by mutating the
localization signal present in its 3' untranslated region (UTR). Although the protein’s function
was unimpaired, mice with this mutation showed an impairment in L-LTP, as well as in spatial,
associative fear conditioning and object recognition memory. These findings further corroborate
the critical role of local gene translation in memory formation.
11
Figure 1.2 Two options for fine-tuned regulation of local gene translation. A. Regulating mRNA transport.
Unique labels (represented here by different colours) may be added to each mRNA transcript, specifying a synaptic
destination. B. Locally regulating mRNA translation. Distinct labels may be added to each mRNA at the spine to
coordinate processing, e.g. translation (green), silencing (yellow) or degradation (red).
Given the evidence that local translation occurs during learning and is required for memory
formation, one can identify two general approaches neurons may use to coordinate protein
synthesis at individual synapses (Fig. 1.2). The first is that neurons regulate local translation by
specifically targeting each mRNA transcript to the correct spine, via unique labels (Fig. 1.2A).
The second is that each mRNA present in a spine is locally and dynamically labelled for
processing, e.g., translation, silencing or degradation, as needed (Fig. 1.2B). It is also possible
that mechanisms of both types are engaged in coordinating local translation together. Importantly
however, among the mechanisms known to regulate both mRNA transport and local translation,
12
none appear to have the flexibility and dynamic nature required to regulate plasticity at the level
of the individual spine and transcript.
1.3.4.1 Regulation of local mRNA transport in memory
Local transport of mRNA occurs primarily via RNA-containing granules like ribonucleoprotein
particles (RNPs) which are transported along microtubules to dendrites (Dynes & Steward, 2007;
Knowles et al., 1996; Vessey et al., 2006). When transported in RNPs, mRNA is typically
maintained in a translationally repressed state to be released following arrival at the destination
(Bramham & Wells, 2007).
How specific mRNAs are labelled for transport is not well understood. One of the best studied
examples is the case of Actb, which possesses a 54-nucleotide element termed "RNA zipcode" in
its 3' UTR (Kislauskis et al., 1994). This zipcode is recognized by zip-code-binding protein 1
(ZBP1), which is thought to coordinate RNP transport of Actb mRNA to dendritic segments.
Indeed, ZBP1 has been shown to colocalize with Actb in dendritic RNPs (Tiruchinapalli et al.,
2003) and silencing ZBP1 prevents granule formation (Farina et al., 2003), impairing Actb
transport to dendrites (Hüttelmaier et al., 2005; Oleynikov & Singer 2003). When bound to Actb,
ZBP1 inhibits translation. However, upon phosphorylation, ZBP1’s binding affinity for RNA is
reduced, relieving translation repression (Hüttelmaier et al., 2005). Thus, ZBP1 appears to
function both as a transport and translation regulator for Actb. Importantly, neuronal
depolarization has been shown to stimulate movement of ZBP1-containing granules to dendrites
in an NMDA receptor mediated way, suggesting that the Actb zipcode may indeed regulate
dendritic localization during LTP and memory formation (Tiruchinapalli et al., 2003).
Other mRNAs have been found to possess similar zipcodes in their 3' UTRs, like CamKIIA
(Miller et al., 2002), which binds cytoplasmic polyadenylation element binding protein 1
(CPEB1) (Huang et al., 2003). Similar to ZBP-1, CPEB1 forms RNPs in cultured hippocampal
neurons (Huang et al., 2003) and following neuronal stimulation, the number of CamKIIA
mRNA-containing granules localized to dendrites increases (Rook et al., 2000). Several other
zipcode-containing mRNAs, like Arc and Bdnf, are also known to be transported to dendrites, but
the binding proteins involved are unknown (for review, see Andreassi & Riccio, 2009). In the
case of Bdnf, transcripts possessing a longer version of the 3' UTR are more likely to localize to
dendrites of hippocampus neurons (An et al., 2008). Furthermore, Flavell et al. (2008) saw some
13
evidence that, following neuronal stimulation, there is a relative increase in shorter versions of
mRNA transcripts with truncated 3' UTRs, however whether this is the case for PRPs in
particular is unknown. If PRPs are among these genes, this could partially account for activity-
related regulation of transcript localization. Thus, altogether, these findings support a role for
3’UTR zipcodes in determining whether certain transcripts are kept in the soma or nucleus, or
transported to dendrites.
However, several lines of evidence show that zipcodes cannot fully account for the complexity
of dendritic localization. First, evidence from Ainsley et al. (2014) suggests that the number of
gene transcripts present in dendrites may be several orders of magnitude greater than previously
thought. Indeed, following CFC, they found over 2,000 mRNAs bound to ribosomes in dendrites
of CA1 pyramidal neurons. With this many dendritically localized transcripts, it is unlikely that
each possesses a different hardcoded signal recognized by a different localization protein.
Nonetheless, no widespread shared dendritic localization signal has been identified to date that
could account for the dendritic localization of so many mRNAs. Second, different mRNAs are
dendritically expressed in different cells, and at different developmental stages. In fact, even
within individual cells, certain mRNAs, like Map2 and the InsP3 receptor are differentially
expressed in proximal and distal dendrites (Steward & Schuman, 2001). Each zipcode thus
would have to exist in multiple versions to allow for this level of precision. Thirdly and most
importantly for memory formation, although zipcodes may signal for dendritic localization with
some precision, they simply cannot generate enough differentiable signals to specify the exact
dendritic segment or synapse to which a transcript should be transported. Overall, it is clear that
mRNA targeting in neurons is too complex to be entirely regulated by hardcoded signals like
zipcodes.
An important parallel can be drawn here with gene transcription where transcription factors
allow broad changes in gene expression during memory, and epigenetic markers enable fine-
tuned regulation at the level of the individual gene. A similar dynamic and flexible system must
be regulating local mRNA transport during LTP and memory formation (Fig. 1.2A).
1.3.4.2 Regulation of local mRNA translation in memory
Although neurons may primarily regulate local gene translation by controlling which transcripts
are targeted to which synapse, it is also possible that each transported transcript bears a signal
14
indicating whether it should be translated, silenced or degraded (Fig. 1.2B). Generally, when
mRNA transcripts are trafficked to spines for local translation, they remain dormant until a
signaling cascade is initiated prompting their translation, as occurs during L-LTP (Bramham &
Wells, 2007). How different genes are translated at different rates and how some are kept silent,
while others are translated is unknown. A mechanism must exist to precisely and dynamically
label different transcripts for translation or degradation during learning.
Promising candidates for transcript-specific regulation of local protein synthesis include CPEB1,
fragile X mental retardation protein (FMRP) and miR-134, a brain-specific micro RNA
(miRNA). All three molecules localize to synapses (Schratt et al., 2006; Wu et al., 1998;
Zalfa et al., 2003) and have been shown to be involved in translation repression (Kim & Richter,
2006; Schratt et al., 2006; Stebbins-Boaz et al., 1999; Zalfa et al., 2003; Zhang et al., 2001) and
memory processes. Specifically, knocking out CPEB1 and FMRP respectively impairs spatial
memory in mice (Berger-Sweeney et al., 2006) and alters synaptic morphology and plasticity of
cultured hippocampal neurons (Braun & Segal, 2000). miR-134 negatively regulates dendritic
spine size (Schratt et al., 2006) and the miRNA signaling pathway is involved in long-term
memory acquisition in Drosophila (Ashraf et al., 2006). Furthermore, CPEB1, FMRP and
miRNAs are regulated by memory-related events like release of BDNF and glutamate receptor
activation (Ashraf et al., 2006; Huang et al., 2002; Shin et al., 2004; Wells et al., 2001;
Weiler et al., 1997). These findings point strongly to a role for these molecules in regulating
local mRNA translation during memory formation.
It is likely that many other translation regulators exist in addition to CPEB1, FMRP and
miRNAs. Several of these may regulate local translation like transcription factors, by binding to
hardcoded signals in mRNA, thus broadly silencing or activating translation at synapses during
memory. Additional mechanisms, like the phosphorylation of certain elongation factors, also
enable cells to regulate overall translation levels (Costa-Mattioli et al., 2009). However, as
mentioned previously, it is also likely that certain translation regulators enable individual
mRNAs to be distinctly regulated, by recognizing more flexible and dynamic signals in mRNA.
This would allow neurons to target certain transcripts at a synapse for translation and others for
silencing or degradation in response to external stimuli. As was suggested for local mRNA
transport, epigenetic modifications of DNA may provide a hint as to the nature of these dynamic
signals.
15
To conclude, memory formation in tasks such as CFC relies on synaptic plasticity in many
regions of the brain, including the dorsal CA1 region of the hippocampus. Long-term synaptic
plasticity in turn depends on both gene transcription and translation, including local translation in
spines and dendrites. In order to specifically strengthen the synaptic connections that underlie a
memory, neurons must be able to carefully regulate local gene translation through dynamic and
transcript-specific regulation of transport and/or translation. How transport and translation are
specifically regulated at the individual transcript- and spine-level remains unknown. We propose
that a flexible and rapidly modifiable regulatory system must exist that is able to distinctly label
transcripts to be targeted to different dendritic segments and/or to be translated at a specific
synapse during memory formation. Just as epigenetic DNA modifications flexibly and quickly
regulate gene transcription during learning, modifications of mRNA may fulfill the same role in
translation regulation in memory.
1.4 m6A mRNA Methylation as a Potential Regulator of Local Gene
Translation
1.4.1 Common mRNA modifications
Similar to genes which are surrounded by promoters and operators that regulate transcription,
mRNAs contain non coding regulatory components in their 3' and 5' UTRs, as well as in their
introns (for review, see Mignone et al., 2002; Wilkie et al., 2003). These hardcoded regions
regulate mRNA splicing, translation, degradation and even broad subcellular localization, as in
the previously described case of CamKIIA. Although these elements can differ between
transcripts, precursor mRNA undergoes only a few specific sequence modifications, namely
capping, polyadenylation and splicing (for review, see Moore & Proudfoot, 2009), and mature
mRNA is subject to few sequence changes prior to degradation (Coller & Parker, 2004; Dunn &
Cowling, 2014). As a result, it is clear that hardcoded signals like zipcodes in mature mRNA
cannot be readily modified during memory formation to finely regulate local translation at the
individual synapse.
However, similarly to DNA whose nucleotides can be epigenetically modified, RNA nucleotides
can undergo chemical modifications. In fact, more than one hundred such modifications exist,
16
most of which involve methylation, namely the substitution of an atom or group by a methyl
group (CH3) on a nucleotide (Czerwoniec et al., 2009). The most well studied modification is the
cap which is critical for the stability of most mRNAs, but also involved in regulating translation
initiation, export, polyadenylation and splicing of mRNA (for review, see Dunn & Cowling,
2014). Although capping is reversible, decapping is generally followed by degradation, as it
destabilizes transcripts (Dunn & Cowling, 2014). Another common modification is 5-
methylcytosine (m5C), which plays a key role in epigenetically regulating gene transcription
during memory, but is much less prevalent and not well studied in mRNA (Squires et al., 2012).
One of the main reasons these modifications have remained poorly understood is the lack of
appropriate tools to detect and locate these sites in transcripts. Recent advances in high
throughput sequencing has allowed in-depth characterization of many of these modifications. It
is through this work that the most prevalent modification of mRNA, m6A, was discovered in the
1970’s (Desrosiers et al., 1974). However, the more recent discovery that m6A is dynamically
regulated is what has brought it to the forefront of research into mRNA modifications
(Dominissini et al., 2012).
1.4.2 m6A modification of mRNA
The m6A modification of mRNA is characterized by the replacement of a hydrogen by a methyl
group on the nitrogen in the N6 position in adenosine (Fig. 1.3) (Kierzek & Kierzek, 2003).
Accounting for around 50% of mRNA modifications (Wei et al., 1975), 0.1-0.4% of RNA
adenosines are estimated to be methylated, with one m6A site occurring per 2,000
ribonucleotides (Fu et al., 2014a; Wei et al., 1975). In human cells, m6A is present in over 7,000
mRNAs, and enriched around stop codons, in 3' UTRs and within long internal exons
(Dominissini et al., 2012; Meyer et al., 2012), and binds to a well-conserved consensus
sequence: RRACH (where R is a purine, A is the methylated adenosine, and H is adenosine,
cytosine or uracil). Although the consensus sequence is strong, most instances of it in mRNA are
not methylated (Dominissini et al., 2012). Taken together, these findings suggest that m6A
patterns in different copies of the same gene could vary enough to signal for a variety of different
fates, as would be needed to finely regulate local transport or translation of mRNA.
In studying the role of m6A in mRNA, it is important to note that m6A is also found in long non
coding RNAs, as well as in ribosomal, small nuclear and transfer RNAs (Fu et al., 2014a). In
17
contrast, in mammalian genomic DNA, where methylation typically occurs on cytosine, m6A is
barely detectable (Fu et al., 2014a; Jia et al., 2011). Together, these instances of m6A account
for less than 6% of m6A sites, demonstrating that m6A is largely restricted to mRNA
(Meyer et al., 2012).
Figure 1.3. The m6A pathway. The m6A methyltransferase complex (including METTL3, METTL14 and WTAP)
catalyzes methylation of the adenosine (A) base. m6A binding proteins like ELAVL1 and YTHDF1, 2 and 3 bind to
or near m6A, influencing mRNA processing (e.g. translation, stability, splicing and transport). m6A demethylases
FTO and ALKBH5 catalyze demethylation of the adenosine base. Whereas demethylation by ALKBH5 is direct,
demethylation by FTO generates two sequential intermediates, hm6A and f6A, each of which can directly hydrolyze
to adenosine (Fu et al., 2014a)
One of the early indicators that m6A does not occur statically in mRNA was the discovery that
although many m6A peaks are well-conserved between humans and mice, certain sites are
methylated only in a subset of transcripts, whereas other transcripts completely lack m6A even at
18
consensus sites (Cory et al., 1976; Dominissini et al., 2012; Horowitz et al., 1984). However, it
was the discovery in 2011 by Jia et al. that m6A can be demethylated, and is thus a reversible
modification that brought the study of m6A to the forefront. Dominissini et al. (2012) then
showed that external stimuli like ultra-violet light, heat shock, cell specific growth factors or
interferon-gamma can induce a dynamic shift in the m6A patterns of mRNA in cells, affecting 5-
30% of m6A peaks. Importantly, this regulation of m6A is accompanied by changes in gene
expression and splicing, suggesting an important functional role for m6A in regulating cellular
responses to external stimuli.
These studies provided the first evidence for m6A as a rapid and reversible regulator of mRNA
processing, and led to a rapid growth of interest in m6A. Additional research has since suggested
that m6A is involved in regulating not only mRNA transport, but also splicing and translation
(Dominissini et al., 2012; Fustin et al., 2013). Specifically, Zhou et al. (2015) showed that
following heat shock, increased methylation in the 5' UTR of certain transcripts enables
accelerated translation. Together, these findings point to a dynamic and flexible role for m6A in
regulating a variety of mRNA processing steps, making it a prime candidate for finely regulating
local protein synthesis in neurons during memory formation.
1.4.3 m6A pathway
In the next few sections, an overview is provided of the research conducted to date looking at the
various components known to be involved in the m6A pathway. These components can be
subdivided into three groups: methyltransferases, demethylases and methylation binding proteins
(Fig. 1.3). It is important to note that research in this field has only just begun to gather
momentum. As a result, it is very likely that new components and functions will be discovered in
the next few years, filling in many of the gaps in our current understanding of the m6A pathway.
Nonetheless, significant progress has been made in sketching a preliminary account of the
functioning of the m6A pathway.
1.4.3.1 m6A methyltransferases
m6A RNA methylation is catalyzed by a methyltransferase complex, comprising three well
conserved components: two methyltransferases (METTL3 or MT-A70, and METTL14) and one
interacting protein (Wilms’ tumour 1-associating protein (WTAP)) (Fu et al., 2014a) (Fig. 1.3).
19
METTL3 and 14 may appear redundant in the complex, as they are highly homologous and each
possesses a binding site for S-adenosylmethionine, a key substrate for catalyzing methylation
(Bokar et al., 1994; Chiang et al., 1996; Ping et al., 2014; Wang, Y. et al., 2014). However, the
fact that METTL14 shows ten times higher methyltransferase activity suggests that METTL3
may also undertake different functions in the complex (Liu et al., 2014). WTAP, for its part, is a
known regulator of mammalian pre-mRNA splicing (Horiuchi et al., 2013; Ping et al., 2014),
and shows no methyltransferase activity (Liu et al., 2014). Instead, it interacts with many
proteins involved in RNA splicing, stability, polyadenylation, and export (Horiuchi et al., 2013),
and thus, may be responsible for recruiting enzymes to the complex to modulate
methyltransferase activity (Fu et al., 2014a).
Evidence for the role of this complex in methylating mRNA comes first from the fact that all
three components of the complex bind mRNA at sequences consistent with the m6A consensus
sequence, mostly in coding sequences or 3' UTRs, where m6A peaks occur in mammalian
mRNA (Dominissini et al., 2012; Liu et al., 2014; Meyer et al., 2012). Furthermore, knocking
down the levels of each methyltransferase component in vitro has repeatedly been shown to
deplete m6A in a variety of cell types, with the biggest effects occurring when WTAP or
METTL14 is knocked down (Bokar et al., 1997; Dominissini et al., 2012; Geula et al., 2015;
Liu et al., 2014; Wang, Y. et al., 2014) (For a detailed list, see Table 1.1). Conversely,
overexpressing METTL3 or METTL14, but not WTAP, also increases m6A levels, with a
synergistic effect occurring when both are overexpressed together (Liu et al., 2014).
Interestingly, one study in porcine adipocytes found no effect of knocking down METTL3 on
m6A levels, but did see an increase in m6A following overexpression (Wang et al., 2015b). This
lends further support to the hypothesis that METTL3 may have functions additional to transcript
methylation. It is clear nonetheless from these studies that the methyltransferase complex
critically regulates m6A levels in mRNA. This regulation is in turn important for cellular
function, and in particular, coordinating several stages of mRNA processing, including
translation, splicing and transport.
The importance of m6A for normal cell function is shown by the effects of dysregulating the
levels of methyltransferase complex components. For instance, knocking down METTL3 in vitro
leads to apoptosis in certain cells (Dominissini et al., 2012), whereas overexpression reduces
adipogenesis in porcine adipocytes (Wang et al., 2015b). Similarly, knocking down METTL3 or
20
METTL14 in murine embryonic stem cells (mESC) leads to loss of self-renewal capability and
aberrant lineage priming resulting in premature cell death (Geula et al., 2015; Wang, Y. et al.,
2014). A comparable effect is seen following knockdown of METTL3 or WTAP in zebrafish
embryos (Ping et al., 2014) (see Table 1.1). Thus, it is clear that interfering with m6A patterns in
mRNA has severe consequences on cellular function. This dysregulation could be due to effects
on gene translation, splicing and/or transport, all of which have been shown to be affected when
m6A methyltransferase complex components are targeted.
When it comes to gene translation and stability, knocking down METTL3 upregulates some
genes while downregulating others, suggesting that m6A can have bidirectional effects on its
targets (Dominissini et al. 2012). Wang, Y. et al., 2014 found that developmental regulator genes
targeted by the methyltransferase complex show an inverse correlation between their m6A
methylation levels and transcript stability, as seen by enriched mRNA levels when m6A is
depleted (Wang, Y. et al., 2014). Expression of developmental regulators in mESCs may
therefore be regulated by m6A to maintain a pluripotent state (Wang, Y. et al., 2014). Recently,
however, Lin et al. (2016) found that a catalytically inactive METTL3 mutant lacking
methyltransferase activity was still able to promote mRNA translation in human cancer cells by
associating directly with ribosomes (see Table 1.1). Thus, although METTL3 is certainly
involved in RNA methylation, it is possible that some of the effects on translation of
manipulating METTL3 are independent of m6A.
In addition to regulating translation, m6A patterns in mRNA have also been linked to splicing.
Indeed, modulating mRNA m6A levels by knocking out METTL3 leads to differential
expression of splice variants (Dominissini et al., 2012). Furthermore, all three components of the
methyltransferase complex are enriched in nuclear speckles (Bokar et al., 1997; Ping et al.,
2014), structures specifically involved in splicing regulation (Lamond et al., 2003). WTAP
depletion reduces this enrichment for all three enzymes, indicating that its presence is required
for the methyltransferase complex to localize to nuclear speckles (Ping et al., 2014) (see
Table 1.1). Together, these findings also support a role for m6A in regulating splicing.
Lastly, in investigating the role of m6A in regulating the cellular circadian clock, Fustin et al.
(2013) showed that overexpressing METTL3 in human cells and mouse embryonic fibroblasts
(MEF) shortens the circadian period, whereas knocking down METTL3 elongates it (see
21
Table 1.1). Further investigation suggested that the period elongation was linked to delayed
export of mature mRNAs involved in regulating the cellular circadian clock. These findings
provide preliminary evidence for a role for m6A in dynamically regulating mRNA transport.
Overall, the m6A methyltransferase complex is clearly involved in regulating m6A levels in
RNA and may play a role in regulating translation and stability, splicing and transport of mRNA
(see Table 1.1). Specifically, it seems that METTL3 and METTL14 directly regulate m6A levels,
whereas WTAP, and perhaps METTL3 as well, through their interactions with other enzymes,
provides an interface for cellular signals involved in mRNA processing to respond to m6A
levels. Although these findings support a potential role for m6A and the methyltransferase
complex in regulating processes like local protein synthesis in memory formation, no research
has been conducted investigating the role of these enzymes in vivo, and particularly in memory.
For this research project, we investigate how the methyltransferase complex is regulated during
memory, with a focus on METTL14, as it is the main active component of the complex, and
METTL3, as it has been shown to regulate translation.
1.4.3.2 m6A demethylases
In 2011, Jia et al. identified the fat mass and obesity-associated protein (FTO) as the first m6A
demethylase. Two years later, Zheng et al. (2013) discovered that AlkB Homolog 5 (ALKBH5)
also demethylates m6A in RNA (Zheng et al., 2013). These discoveries were of great
importance, as they demonstrated that m6A is reversible, and therefore could dynamically
regulate mRNA processing. Since then, several studies have shown that FTO and ALKBH5
expression in vitro inversely correlates with m6A, further demonstrating their roles as m6A
demethylases (Jia et al., 2011; Zheng et al., 2013) (see Table 1.2 for more details).
Previous to this discovery, FTO had been primarily studied for its possible link to obesity
(Church et al., 2010; Dina et al., 2007; Frayling et al., 2007; Fredriksson et al., 2008;
Gerken et al., 2007; McTaggart et al., 2011; Scott et al., 2007; Stratigopoulos et al., 2008;
Tung et al., 2010), but its precise function remained unknown. It had also been linked to
memory, first as it is well expressed in the brain (Church et al., 2010), and second, because
several of the mutation sites in FTO correlate with an increased risk for Alzheimer’s disease,
cognitive decline and reduced brain volume in humans, independently of obesity-related
measures (Bressler et al., 2013; Keller et al., 2011). If m6A is indeed involved in regulating
22
plasticity in the brain, this could account for FTO’s link to cognitive function, and perhaps even
obesity. ALKBH5, for its part, is enriched in testes and has been primarily linked to fertility
(Zheng et al., 2013), suggesting it may not play a prominent role in brain-related functions (see
Table 1.2).
In addition to being expressed in different tissues, FTO and ALKBH5 catalyze oxidative
demethylation of m6A in different ways, as FTO produces oxidation intermediates between m6A
and adenosine, whereas ALKBH5 does not (Chen et al., 2014) (Fig. 1.3). These two
intermediates, N6-hydroxymethyladenosine (hm6A) and N6-formyladenosine (f6A) are stable in
vitro under physiological conditions with half-lives of approximately three hours each and show
reduced binding affinities to m6A binding proteins (Fu et al., 2013). Given that mammalian
mRNAs have median half-lives of around five hours, it is possible that FTO-generated m6A
intermediates are recognized by a different set of binding proteins and play a functional role in
regulating mRNA processing (Fu et al., 2014a) (Fig. 1.3). Together, these findings strongly
suggest that ALKBH5 and FTO, although both regulators of m6A, have different functions and
can thus produce distinct downstream effects on mRNA processing.
As expected based on the findings from manipulating methyltransferase levels in vitro, altering
demethylase levels also impairs cellular function. Indeed, just as knocking down METTL3 and
METTL14 in mESC leads to aberrant premature differentiation, knocking down FTO in mouse
preadipocytes produces the inverse effect, inhibiting differentiation (Zhang et al., 2015). In male
mice, knocking down ALKBH5 impairs fertility due to maturation arrest and apoptosis in
spermatocytes (Zheng et al., 2013) (see Table 1.2). Again, this is likely due to effects of m6A on
mRNA translation and stability, splicing and transport.
First, as seen when methyltransferase components are knocked down, overexpressing FTO in
mice differentially affects different groups of transcripts. This supports a nuanced role for m6A
in regulating mRNA translation, implying that transcripts can be targeted distinctly for
processing based on the needs of a cell (Merkestein et al., 2014). Second, both FTO and
ALKBH5 localize to nuclear speckles, and depleting FTO leads to modifications in splicing
patterns of methylated transcripts (Jia et al., 2011; Zhao et al., 2014; Zheng et al., 2013). Third,
whereas knocking down METTL3 levels delays nuclear export of certain transcripts, knocking
down ALKBH5 accelerates nuclear export, while increasing the rate of RNA synthesis (see
23
Table 1.2). Together these findings demonstrate the flexibility of the m6A pathway and further
corroborate the hypothesis that m6A is used by cells to regulate mRNA processing and cellular
functions dynamically in response to external stimuli. Based on the link between FTO and
memory, one of these functions could be synaptic plasticity.
To date, only one study has investigated the m6A pathway in the brain, focusing on FTO which
is highly expressed in the hypothalamus, cerebellum and hippocampus (Church et al., 2010), and
localizes to neurons (McTaggart et al., 2011). They found that knocking out FTO leads to a
decrease in IEG levels, implying a change in cellular activity (Hess et al., 2013). Importantly,
they also saw a specific increase in methylation of certain PRP transcripts, including glutamate
and dopamine receptor subunits. While no research has been done to further investigate the
contribution of FTO and m6A to synaptic plasticity, it is clear from these studies that both may
play a key role in memory formation. Investigating how FTO and ALKBH5 are regulated
following memory formation will provide insight into this link, and may indeed uncover a
difference between the roles of these two m6A demethylases in memory.
1.4.3.3 m6A binding proteins
The third group in the m6A pathway comprises the proteins that recognize m6A sites in RNA
(Fig. 1.3). Given that the presence of m6A only minimally affects the secondary structure of
mRNA, it is likely that its functions are primarily carried out through recognition by these
binding proteins, and not direct chemical effects on mRNA structure (Kierzek & Kierzek, 2003).
Furthermore, several of these proteins have been linked to mRNA stability, translation, splicing,
and localization, further supporting a role for m6A in mRNA processing.
Among the proteins found to bind m6A sites in mammalian cells, several are members of the
YT521-B homology (YTH) domain family (YTHDF1, 2 and 3) (Dominissini et al., 2012; Wang
et al., 2015a; Wang, X. et al., 2014; Zhang et al., 2010). The most well-studied, YTHDF2, has
been linked to bidirectional changes in target mRNA translation, depending on its binding site.
On the one hand, Wang, X. et al. (2014) showed that YTHDF2 binding in 3’UTRs and coding
sequences of transcripts results in the translocation of mRNA from the translatable pool to decay
sites, like processing bodies. Knocking down YTHDF2 prevents this translocation, leading to an
accumulation of non-translating m6A-containing mRNAs that would typically be degraded
(Wang, X. et al., 2014) (see Table 1.3). On the other hand, Zhou et al. (2015) found that heat
24
shock upregulates YTHDF2 levels in the nucleus, as well as the translation of stress-induced
transcripts via an increase in m6A levels in their 5’UTRs. These increases are abolished when
YTHDF2 is silenced, but potentiated when FTO is knocked down, suggesting that YTHDF2
binding to m6A in stress-related transcripts blocks their demethylation by FTO, and enables their
translation (Zhou et al., 2015) (see Table 1.3). Given that heat shock typically leads to a broad
suppression of translation, m6A thus enables cells under these conditions to specifically
upregulate translation of stress-related proteins only. Together, these findings show that by
binding m6A at different sites, YTHDF2 dynamically regulates mRNA translation.
In addition to YTHDF2, YTHDF1 has also been shown to promote protein synthesis by
translocating its target mRNA to ribosomes and interacting with translation initiation factors (see
Table 1.3). Although both binding proteins share many target transcripts, YTHDF2 binds RNA
at a later stage in processing than YTHDF1, suggesting complementary rather than competitive
roles (Wang et al., 2015a). Further research is needed to elucidate the different roles of
YTHDF1, 2 and 3 in regulating translation and stability, and perhaps also splicing and transport
of mRNA. Nonetheless, the research to date demonstrating the ability of cells to exercise fine-
tuned control over cellular responses to external stimuli by regulating the expression of m6A
pathway components like FTO and YTHDF2 is already very promising, and certainly supports
the potential for this pathway to finely regulate processes like synaptic plasticity in neurons.
In addition to YTHDF1-3, Dominissini et al. (2012) also identified ELAV-like RNA binding
protein 1 (ELAVL1 or HuR) as significantly associated with methylated mRNA. Likely the best
studied proteins in the m6A pathway, ELAVL1 is involved in many aspects of mRNA
processing that have also been linked to m6A, including enhancing stability (for review, see
Brennan et al., 2001), modulating translation levels (Durie et al., 2011), and possibly regulating
splicing (Lebedeva et al., 2011; Mukherjee et al., 2011). Unlike YTHDF1-3 however which bind
the m6A consensus site, ELAVL1 typically binds uracil-rich regions in 3' UTRs, increasing the
stability of transcripts. In fact, although ELAVL1 is associated with methylated mRNA, its
ability to bind mRNA is sensitive to the proximity of m6A, showing different binding affinities
at different intervals from m6A (Wang, Y. et al., 2014). Like YTHDF2, ELAVL1 also binds to
5’UTR regions where it promotes transcript translation (Durie et al., 2010). This appears to be
partially due to its ability to block binding of miRNAs (for review, see Srikantan et al., 2012),
which have been shown to regulate local gene translation during memory formation (Kosik,
25
2006; Schratt et al., 2006), reducing the stability and suppressing the translation of certain
mRNAs (for review, see Fabien et al., 2010), while promoting the translation of others
(Vasudevan et al., 2010) (See Table 1.3). These findings not only suggest a more complex
interaction between ELAVL1 and m6A than with YTHDF1, 2 and 3, but also support the ability
of m6A to flexibly induce a large number of distinct downstream effects on mRNA depending
on its location in a transcript, a property necessary for regulating complex mechanisms like local
gene translation in memory.
More research is needed to identify new m6A binding proteins and elucidate their effects on
mRNA translation, stability, splicing and transport, in particular in vivo. For instance,
heterogeneous nuclear RNPs which form mRNA granules are also associated with methylated
mRNA, but their potential role in regulating transcript localization and transport based on m6A
patterns has yet to be investigated (Fu et al., 2014a). Nonetheless, the large variety of proteins
and molecules linked to m6A demonstrates a great potential for m6A binding proteins to finely
regulate mRNA processing, and confirms the flexible and dynamic nature of this pathway, as
well as its potential as a regulator of local translation during memory formation. In particular,
YTHDF2, which shows a rapid response to external stimuli like heat shock, and ELAVL1, which
is already known to regulate many aspects of mRNA processing also affected by m6A, are
promising candidates for a role in memory formation in the brain. This project will thus focus on
YTHDF2 and ELAVL1 in investigating the effects of memory formation on m6A binding
proteins.
26
Table 1.1. Role of m6A methyltransferase components.
METTL3 METTL14 WTAP
m6A levels
- Reduced after knockdown in HepG2 cells, HeLa cells,
293FT cells, mESCs and zebrafish embryos (Bokar et
al., 1997; Dominissini et al., 2012; Geula et al., 2015;
Liu et al., 2014; Ping et al., 2013; Wang, Y. et al., 2014)
- No effect after knockdown in porcine adipocytes
(Wang et al., 2015b)
- Increased after overexpression in porcine adipocytes
(Wang et al., 2015b)
- Increased after overexpression in HeLa and 293FT cells
(Liu et al., 2014)
- Reduced after knockdown in
HeLa cells, 293FT cells and
mESCs (Liu et al., 2014; Wang, Y.
et al., 2014)
- Increased after overexpression in
HeLa and 293FT cells (Liu et al.,
2014)
- Reduced after knockdown in
HeLa cells, 293FT cells and
zebrafish embryos (Liu et al.,
2014; Ping et al., 2013)
- No effect after overexpression in
HeLa and 293FT cells (Liu et al.,
2014)
Cellular function
- Knockdown impairs self-renewal capability and lineage
priming, leading to cell death in mESCs (Geula et al.,
2015; Wang, Y. et al., 2014)
- Knockdown elongates circadian period in U2OS cells
and MEFs (Fustin et al., 2013)
- Overexpression shortens circadian period in U2OS
cells and MEFs (Fustin et al., 2013)
- Knockdown impairs self-renewal
capability and lineage priming,
leading to cell death in mESCs
(Wang, Y. et al., 2014)
No findings to date
mRNA expression,
stability and translation
- Knockdown has bidirectional effects on transcript
levels in HepG2 cells (Dominissini et al., 2012)
- Knockdown increases developmental transcript levels
in mESCs (Wang, Y. et al., 2014)
- Promotes translation in a methyltransfer-independent
way (Lin et al., 2016)
- Knockdown increases
developmental transcript levels in
mESCs (Wang, Y. et al., 2014)
- Interacts with stabilizing proteins
(Horiuchi et al., 2013)
mRNA splicing
- Enriched in nuclear speckles (Bokar et al., 1997)
- Knockout leads to differential expression of splice
variants in HepG2 cells (Dominissini et al., 2012)
- Enriched in nuclear speckles
(Ping et al., 2014)
- Enriched in nuclear speckles
(Ping et al., 2014)
- Knockdown reduces enrichment
of all methyltranferase components
in nuclear speckles (Ping et al.,
2014)
- Interacts with splicing proteins
(Horiuchi et al., 2013)
mRNA transport - Knockdown delays nuclear export in U2OS cells and
MEFs (Fustin et al., 2013)
No findings to date - Interacts with export proteins
(Horiuchi et al., 2013)
Memory No findings to date No findings to date No findings to date
27
Table 1.2. Role of m6A demethylases.
ALKBH5 FTO
m6A levels
- Increased after knockdown in HeLa
cells (Zheng et al., 2013)
- Reduced after overexpression in
HeLa cells (Zheng et al., 2013)
- Increased after knockdown in HeLa cells (Jia et al., 2011)
- Reduced after overexpression in HeLa cells (Jia et al., 2011)
Cellular function - Knockdown leads to maturation arrest
and apoptosis in spermatocytes, and
impairs fertility (Zheng et al., 2013)
- Knockdown impairs differentiation in mouse 3T3-L1 preadipocytes (Zhang et al.,
2015)
mRNA expression,
stability and translation
- Knockdown increases RNA synthesis
rate in HeLa cells (Zheng et al., 2013)
- Overexpression differentially affects levels of different transcript groups in HepG2
cells (Merkestein et al., 2014)
- Demethylation of transcripts in MEFs during heat shock induces degradation (Zhou et
al., 2015)
mRNA splicing - Enriched in nuclear speckles (Zheng
et al., 2013)
- Enriched in nuclear speckles (Jia et al., 2011)
- Knockdown leads to differential expression of splice variants in mouse 3T3-L1
preadipocytes (Zhao et al., 2014)
mRNA transport - Knockdown accelerates nuclear
export in HeLa cells (Zheng et al.,
2013)
No findings to date
Memory
No findings to date - Point mutations linked to reduced brain volume and increased risk for Alzheimer’s
disease and cognitive decline in humans (Bressler et al., 2013; Keller et al., 2011)
- Enriched in mouse hypothalamus, cerebellum and hippocampus (Church et al., 2010)
- Expressed in neurons in the mouse hypothalamus, cerebellum and hippocampus
(McTaggart et al., 2011)
- Knockout in mice decreases cellular activity in the midbrain (Hess et al., 2013)
- Knockout in mice increases m6A in PRP transcripts (Hess et al., 2013)
- Knockout in mice increases PRP mRNA levels (Hess et al., 2013)
- Knockout in mice decreases PRP levels (Hess et al., 2013)
28
Table 1.3. Role of m6A binding proteins.
ELAVL1 YTHDF1 YTHDF2
mRNA expression,
stability and translation
- Binding promotes translation and
blocks miRNA binding (Durie et
al., 2010)
- Binding enhances RNA stability
(Brennan et al., 2001).
- Binding promotes
translation in HeLa cells
(Wang et al., 2015a)
- Binding induces degradation in HeLa cells (Wang, X., et
al., 2014)
- Binding reduces translation efficiency in HeLa cells
(Wang, X., et al., 2014)
- Binding during heat shock prevents degradation and
promotes translation in MEFs (Zhou et al., 2015)
mRNA splicing - Possible role in regulating splicing
(Lebedeva et al., 2011; Mukherjee
et al., 2011)
No findings to date No findings to date
Memory No findings to date No findings to date No findings to date
29
1.4.4 m6A in the brain
As this overview showed, research to date investigating the m6A pathway has been conducted
almost exclusively in vitro, with only one study looking at the effects of modulating m6A in the
brain on PRP transcripts (Hess et al., 2013). It is known however that m6A is particularly
enriched in the brain, and present in the kidney, liver and lung. Furthermore, the over 4,500
transcripts in the brain that show enrichment in m6A include many PRPs, like Bdnf, and these
transcripts are preferentially targeted by miRNAs expressed at high levels in the brain
(Meyer et al., 2012). Together, this evidence points toward a potential role for the m6A pathway
in providing fine-tuned regulation of mRNA processing in response to external stimuli in the
brain, and possibly in memory formation.
1.5 Study Rationale
In summary, although research on memory has demonstrated that local gene translation is
required in dendrites and spines to enable synaptic plasticity, how neurons are able to
independently regulate translation of different transcripts at different spines when encoding new
memories is not fully understood. The regulatory system that underlies this ability must be
dynamic, allow for a large number of different signals and respond quickly to external stimuli.
The m6A modification of RNA possesses all of these characteristics, and has been shown to play
a key role in dynamically and flexibly regulating gene translation, stability, splicing and
transport. Together, these findings suggest m6A as an excellent candidate for regulating fine-
tuned memory-related processes, like local gene translation.
Given that the m6A pathway has not been studied extensively in the brain or in memory, this
project aims to lay the groundwork for future research into the possible role of m6A in regulating
local protein synthesis during memory formation. This is accomplished first by establishing
whether certain m6A pathway components are enriched in the brain as compared to other organs,
and second by investigating how these expression levels are affected in the dorsal CA1 of the
hippocampus following a memory task. Lastly, in order to enable future investigation of the role
of m6A in memory, a viral toolbox is designed and validated to probe the effects of
dysregulating the m6A pathway during memory formation.
30
1.6 Hypothesis
The overarching hypothesis for this project is that the m6A pathway is involved in memory
formation. Specifically, we hypothesize that among the m6A pathway components investigated,
specifically Mettl3, Mettl14, Alkbh5, Fto, Elavl1 and Ythdf2, some show enrichment in the brain
as compared to other organs. We further hypothesize that the levels of these components, and of
m6A are regulated during memory formation. We then design viral tools to manipulate the levels
of Fto and validate their effects in vitro on Fto mRNA and m6A levels.
1.7 Review of Chosen Gene Manipulation Methods
1.7.1 Herpes Simplex Virus (HSV)
For the tool development component of this project, the herpes simplex virus (HSV) was
selected to deliver transgenes to neurons and behaviourally investigate their effects on memory
formation. This choice is based on the many advantages HSV offers as compared to other
commonly used transduction tools, including adeno-associated virus (AAV). The first
consideration is cell-type specificity, as we are primarily interested in generating tools to study
the role of m6A in neuronal plasticity. Although AAVs can be targeted specifically to neurons
through the use of different serotypes or promoters (Shetsova et al., 2005; van den Pol et al.,
2009), HSVs are naturally neurotropic and consistently infect only non-dividing mature neurons,
making them a superior choice for neuron-specific work (Fink et al., 1996; Neve et al., 2005).
The second consideration is time course of transgene expression. As the m6A pathway is clearly
involved in a broad range of cellular mechanisms, it is preferable to employ a system that allows
rapid expression, reducing the risk that neurons will engage compensatory mechanisms to
counteract the effects of the transgene, thus confounding any results. HSVs are thus ideal, as
maximal transgene expression is reached within 24 to 72 hours, with expression dissipating
completely by seven days post-infection (Carlezon et al., 1998). This transient expression also
allows behaviour to be specifically correlated to the level of transgene expression, which is
particularly useful in establishing whether the effects of the transient manipulation are permanent
or temporary. In contrast, AAVs are better suited to long-term studies, as maximal transgene
31
expression is typically reached several weeks after infection, and persists for many months, if not
permanently (Lo et al., 2004; van den Pol et al., 2009).
Lastly, the consideration of packaging size is of particular importance for this project. Indeed, as
will be discussed in the next section, one of the viral tools we have developed uses the RNA-
guided endonuclease Cas9 to knockout FTO. Given that Cas9 mRNA spans 4.1 kilobases (kb)
(Senís et al., 2014), and the maximum packaging capacity of most viruses is 4.0-8.0 kb, with
AAVs at the low end (Neve et al., 2005), viral delivery of Cas9 can be particularly difficult, as
additional components like a reporter gene, promoter regions and virus packaging elements must
also be included in viral vectors. Certain groups have attempted to circumvent this problem using
AAVs by co-transducing cells with two viruses (Swiech et al., 2015) or using smaller
orthologues of Cas9 (Ran et al., 2015). HSV avoids this problem entirely, as its packaging
capacity can reach up to 40 kb.
It is important to note that due to the use of helper virus in packaging HSV vectors, HSVs have a
higher toxicity level than AAVs. However, improvements in packaging methods have enabled
minimal uses of helper virus, largely reducing cytotoxic effects (Neve et al., 2005). HSVs have
indeed been successfully used in multiple studies to investigate the role of different genes in
memory in different brain regions, including the hippocampus (Cole et al., 2012; Josselyn et al.,
2001; Han et al., 2009; Sekeres et al., 2010, 2012).
Thus, for its ability to specifically target neurons, as well as its rapid and transient expression,
and significantly larger packaging size, we have chosen to design HSV vectors to deliver
transgenes targeting the m6A pathway to neurons in the hippocampus.
1.7.2 Cas9-based Gene Manipulation
Understanding the role of different genes in regulating brain function is a key goal of
neuroscience research, and typically requires the use of gene editing tools. A common approach
to manipulating gene expression in vivo is viral-mediated delivery of genetic material to
overexpress or knockdown a gene of interest. Traditionally, overexpression has been achieved by
delivering copies of the gene’s coding sequence expressed under a constitutive promoter,
whereas knockdown is typically mediated by RNA interference molecules, like small hairpin
RNAs (shRNAs) that target a gene’s transcripts for degradation (Wilson & Doudna, 2013).
32
Although effective, these tools have certain important limitations. First, since a vector
overexpressing a gene must contain its entire coding sequence, which typically spans several
kbs, targeting several genes at once rapidly increases viral vector size and is thus limited by
maximum virus packaging size, even in HSV (Neve et al., 2005). Second, a gene expressed from
a viral vector does not undergo the same mRNA processing as its endogenous counterpart, due in
part to the absence of the intronic and intergenic regulatory regions that typically surround a
gene. This limits the insight overexpression can provide into the endogenous function of a gene.
In the case of shRNA knockdown, designing a shRNA that effectively targets the transcript of
interest without inducing off-target effects is complex, as the efficiency and specificity of a
shRNA depend on its secondary structure, as well as the structure of the target mRNA region
(Moore et al., 2010). Thus, although virally-mediated overexpression and knockdown are useful
for probing gene function, it will be helpful to develop new methods that circumvent these
disadvantages. For these reasons, although the toolbox developed for targeting the m6A pathway
in this project includes traditional overexpression and shRNA viruses, we have also developed a
Cas9 knockdown virus in order to open the door to developing new, more effective and flexible
gene editing tools.
Identified in the mid-2000s as a natural adaptive immune system in bacteria, the clustered RNA
guided regularly interspaced short palindromic repeats (CRISPR) Cas9 system has quickly
become one of the most popular tools for gene editing (Bolotin et al., 2005; Mojica et al., 2005;
Pourcel et al., 2005). This is largely due to the elegance of this system where an endonuclease,
Cas9, is guided by a single guide RNA (sgRNA) spanning approximately 40 base pairs to cleave
a complementary region of DNA (Jinek et al., 2012). Given that cellular repair of double-
stranded breaks in DNA is prone to error, Cas9 cleavage is able to rapidly and efficiently
introduce frameshift mutations leading to loss of function in target genes (Heidenreich et al.,
2003) (Fig. 1.4A).
Compared to shRNA, sgRNA design is very simple and off-target effects can more easily be
avoided by ensuring that the designed sgRNA shows minimal homology with non-target DNA
sites (Fu et al., 2014b). However, since Cas9 acts by mutating DNA sequences, its effects on a
cell are permanent, even when delivered by a virus that allows transient gene expression like
HSV. To increase the flexibility of the Cas9 system, Gilbert et al. (2014) designed a mutant Cas9
lacking nuclease activity (dCas9). Instead of cutting the target DNA region, dCas9 can act as a
33
scaffold, targeting repressors or promoters to a gene, and temporarily modulating its endogenous
expression. As described by Walters et al. (2015), this can be achieved by including in the
sgRNA a stem loop sequence recognized by certain RNA-binding proteins which in turn recruit
effectors to the dCas9 complex (Fig. 1.4B). Like Cas9, dCas9 can easily be targeted to several
genes by including several different sgRNAs in the viral vector, at relatively little cost in terms
of vector size. If different stem loops are included in different sgRNAs, the target genes can even
be modulated in different directions within the same cell, which may be particularly helpful in
investigating complex pathways like the m6A pathway.
Thus, the advantages associated with using Cas9 and the rapid expansion of its use in gene
editing make it an excellent choice for developing new viral tools to investigate the role of the
m6A pathway memory. To date, however, Cas9 has only been used once in the brain of adult
mice and it was delivered using two AAVs, targeting three genes (Swiech et al., 2015).
Knockdown rates ranged from 17-60%, and were sufficient to produce a behavioural effect.
Given the advantages of HSV over AAV for our purposes detailed in the previous section, we
have elected to express Cas9 and its associated sgRNA using an HSV vector. As a result, this
project constitutes one of the first uses of HSV to deliver Cas9 to cells. If successful, it will pave
the way for the rapid development of a variety of HSVs using dCas9 to target different m6A
pathway components in different directions, allowing for extensive interrogation of the role of
this pathway in memory.
34
Figure 1.4. Gene editing with Cas9 and dCas9. A. Cas9 binds to the target region in gene A, complementary to the targeting portion of the sgRNA, and causes a
double stranded break, which is likely to lead to the introduction of a loss-of-function mutation during repair. B. dCas9 binds to the target region in gene B,
complementary to the targeting portion of the sgRNA. Gene activators or repressors bound to stem loops in the sgRNA are thus recruited to gene B leading
respectively to an increase or decrease in gene expression.
35
Chapter 2
Materials and Methods
Chapter 2
2.1 Mice
Adult (10-14 weeks old) male F1 hybrid (C57 BL/6NTac × 129S6/SvEvTac) mice were used for
all experiments. Mice were group housed (2-5 mice per cage) on a 12h light/dark cycle and
provided with food and water ad libitum. All experimental procedures were conducted in
accordance with the guidelines of the Canadian Council on Animal Care (CCAC), National
Institutes of Health (NIH), and approved by the Animal Care Committee at the Hospital for Sick
Children.
2.2 Behavioural Training
Mice were handled daily for three days prior to CFC. During CFC, mice were placed in a unique
fear conditioning chamber for 5 min. The first 2 min of habituation were followed by three
0.5 mA foot shocks delivered 1 min apart, and an additional minute after the final shock. Mice
were then returned to their home cages in a holding chamber for 30 min, 1h, 2h, or 4h, after
which they were anaesthetized with isofluorane and their brains were flash frozen in isopentane
and stored at -80°C. Hippocampi were later dissected from the whole brain on ice, and dorsal
CA1 was isolated and stored at -80°C (Fig. 3.2, S1).
For gene enrichment studies, three control groups were included to establish whether any effects
found were due to contextual memory formation specifically, or simply exposure to a novel
environment or stress alone: (1) Naïve mice, (2) Shock-only mice, (3) Context-only mice. Naïve
mice were kept in their home cages in the animal colony. Both the shock-only and context-only
control mice underwent similar procedures to the CFC mice. However, instead of CFC, the
shock-only mice received an immediate 0.5 mA foot shock upon entering the fear conditioning
chamber, and were immediately returned to their home cages. Under these conditions, mice are
not able to form a contextual memory, due to a lack of exposure to the context in which the
36
shock is received (Wiltgen et al., 2006). The context-only control mice were exposed to the fear
conditioning chamber for 5 min without foot shocks. All three control groups were age-matched
to the CFC group.
2.3 Cell Culture
Neuro2a (N2a) mouse neuroblastoma cells (ATCC, CCL-131) were grown in high glucose
Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 2 mM
glutamine, non-essential amino acids and penicillin/streptomycin, and maintained in an
atmosphere of 5% CO2 and 95% air at 37°C. N2a cells were seeded at different concentrations in
6-well plates, such that wells would reach approximately 70% confluency when harvested. One
day after seeding, cells were infected with 1-2 μL of virus (p1005-FTO, p1005(+), see details
below) or 2-4 μL (Cas9-Fto, Cas9-ctrl, shRNA-Fto or shRNA-scr, see details below) in 6-well
and 12-well plates respectively. Cells were harvested in the appropriate lysis buffer for
quantitative real time polymerase chain reaction (qPCR) and m6A enzyme-linked
immunosorbent assay (ELISA) (Fig. 3.4) 12h or 2-4 days after transduction.
2.4 Sample Analysis
2.4.1 RNA isolation and complementary DNA (cDNA) synthesis
RNA from whole organs (Fig. 3.1), dorsal CA1 (Fig. 3.2, 3.3, and S1) or N2a cells (Fig. 3.4) was
isolated using the EZ-10 RNA isolation kit (Bio-Basic, BS82322). For mRNA enrichment
(Fig. 3.3A), 26 of the 30 µl of isolated RNA were further purified using the MagJet mRNA
enrichment kit (Thermo Scientific, K2811) and concentrated to 14 µl using the RNEasy
MinElute cleanup kit (Qiagen, 74204). Total RNA was quantified using a Nanodrop 1000
(Thermo Scientific). To ensure that the Fto mRNA sequence present in the p1005-FTO viral
plasmid would not affect gene enrichment results, RNA from cells or tissue infected with p1005-
FTO or p1005(+) (Fig. 3.4A-C) underwent DNase treatment using the DNA-free DNA removal
kit (Ambion, AM1906). For gene enrichment studies or mRNA normalization, RNA was
37
converted into cDNA (400 ng for Fig. 3.1, 7, 9, S1; 2 µL mRNA for Fig. 3.3B) using the High
Capacity cDNA synthesis kit (Thermo Scientific, 4368814).
2.4.2 qPCR
qPCR was performed on 6 ng of cDNA, using the EvaGreen Mastermix (Diamed, ABM
Mastermix-S) in a 10 μL reaction and run on a CFX96 real-time qPCR detection system
(BioRad) using 500 nM primer dilutions (see below for sequences). To normalize m6A-
containing mRNA, 2 μL of cDNA were loaded in a standard 10 μL qPCR reaction (Fig. 3.3B).
Relative gene enrichment was calculated from threshold cycles (CT) using the 2-∆∆CT method (see
below for equation) (Schmittgen & Livak, 2008), and two internal controls for normalization:
glyceraldehyde 3-phosphate dehydrogenase (Gapdh) and hypoxanthine-guanine
phosphoribosyltransferase (Hprt). Control samples used for normalization were brain for organ
analysis (Fig. 3.1), naïve mice for CFC analysis (Fig. 3.2, 3.3, and S1) and cells treated with
control viruses for virus validation (Fig. 3.4).
2-∆∆CT measure of relative enrichment, where ∆∆CT =
[(CT gene of interest – geometric mean of CT internal controls) for sample A
- (CT gene of interest – geometric mean of CT internal controls) for sample B],
with sample A being the treated sample, and sample B being the control sample.
Primers (5’-3’, forward-reverse):
Alkbh5: (GGCGTTCCTTAATGTCCTGA, AGTTCCAGTTCAAGCCCATC)
Elavl1: (GACACCAFAAATCCCACTCAT, GCCAATCCCAACCAGAACA)
Fto: (CTCAGCCACTCAAACTCCAC, TCTTAGAACGCTGTCAGTTGG)
Gapdh: (GTGGAGTCATACTGGAACATGTAG, AATGGTGAAGGTCGGTGTG)
Hprt: (GGAGTCCTGTTGATGTTGCCAGTA, GGGACGCAGCAACTGACATTTCTA
Mettl3: (GTTCCTTGGCTGTTGTGGTAT, CTGCCTCCGATGTTGATCTG)
Mettl14: (ATTCTCCTGGAGCCCCCTCT, ACCCCACTTTCGCAAGCATACT)
Ythdf2: (CAGTCTATGCAGAACCGCTT, ACACTATGAGAAACGCCAAGAG)
38
2.4.3 m6A ELISA
m6A-containing RNA was quantified using the EpiQuik RNA methylation quantification kit
(Epigentek, P-9005). Either 200 ng of total RNA (Fig. 3.3A, Fig. 3.4C, F, I) or 8 µl of
concentrated mRNA (Fig. 3.3B) were run and fit to a standard curve. Total RNA samples were
then normalized against the total amount (ng) of RNA (Fig. 3.3A, Fig. 3.4C, F, I), whereas
mRNA samples were normalized to the geometric mean of two internal controls (Hprt, Gapdh)
determined using qPCR (Fig. 3.3B), as the concentration of mRNA was below the detection
threshold of the Nanodrop 1000.
2.5 HSV Viral Vector Design
In order to establish in future experiments whether regulation of the m6A pathway is critical to
memory formation, viruses were designed to manipulate the levels of a central pathway
component in neurons: the demethylase FTO. An FTO overexpression virus and two Fto
knockdown viruses (shRNA- and Cas9-based) were designed to enable us to investigate the
effects of counteracting or enhancing the natural regulation of Fto following CFC on memory.
2.5.1 Design of the FTO overexpression HSV vector (p1005-FTO)
Primers were designed to amplify the mouse Fto mRNA coding sequence (NCBI Reference
Sequence: NM_011936.2) with KpnI (5’) and XhoI (3’) restriction sites (5’-3’, forward-reverse:
ATAGGTACCCTTTAGTAGCAGCATGAAGCGC,
CCTCTCGAGTGCTTCCTCTAGGATCTTGCTT). Fto was amplified from whole brain cDNA
using the Phusion flash high-fidelity PCR master mix (Thermo Scientific, F548S), and cloned
into the pENTR1A gateway entry vector (Invitrogen, A10462). Positive clones, as shown by
digestion and sequencing analyses, were then cloned into the HSV p1005(+) vector (Clark et al.,
2002, kindly provided by Dr. Rachael Neve, Massachusetts Institute of Technology (MIT), MA)
using the Gateway LR Clonase II enzyme mix (Invitrogen, 11791100), placing Fto under an
IE4/5 promoter. Notably, p1005(+) co-expresses green fluorescent protein (GFP) under a CMV
promoter. The resulting plasmid (Fig. 2.1A) was sequenced and midi-prepped from a positive
clone and packaged into virus particles using a helper virus. The virus was then purified on a
sucrose gradient, pelleted and resuspended in 10% sucrose. The average titer obtained for the
39
recombinant virus stocks was typically 1.8 x 109 infectious units/ml. p1005(+) was used as a
control for all experiments using p1005-FTO (Fig. 3.4A-C).
2.5.2 Design of Fto sgRNAs
Three sgRNAs targeting the different exons of Fto were designed using Desktop Genetics Ltd. to
ensure maximal on-target and minimal off-target activity. These guides were synthesized
(Integrated DNA Technologies) with flanking sites corresponding to BbsI restriction sites (see
sequences below), phosphorylated (NEB, M0201), annealed and cloned into the pSpCas9(BB)-
2A-GFP (PX458) validation plasmid from Dr. Feng Zhang, MIT, MA (Addgene, 48138)
(Ran et al., 2013) using the BbsI restriction enzyme. Plasmids were then transfected into
NIH 3T3 embryonic murine fibroblasts (ATCC, CRL-1658) using Lipofectamine 3000
(Invitrogen, L3000001). Cells were harvested and processed for qPCR as described above. The
sgRNA that generated the highest level of Fto knockdown, namely the one targeting exon 3, was
selected for the viral plasmid design (data not shown).
sgRNA (5’-3’, 3’-5’):
targeting exon 1: CACCG TTCCCGCTCTCGTTCCTCCG,
AAAC CGGAGGAACGAGAGCGGGAA C
targeting exon 2: CACCG CCAGAAACTGAGGCTCCTTG,
AAAC CAAGGAGCCTCAGTTTCTGG C
targeting exon 3: CACCG GCAGTGTGAGAAAGGCCTC,
AAAC GAGGCCTTTCTCACACTGC C
2.5.3 Design of the Fto and control sgRNA Cas9 vectors (Cas9-Fto and Cas9-ctrl)
The sgRNA targeting exon 3 of Fto was cloned by collaborators (Dr. Rachael Neve, MIT, MA)
into an HSV vector expressing Cas9-2A-GFP under a Synapsin promoter and the sgRNA under a
U6 promoter (Fig. 2.1B). A control vector containing no sgRNA was also designed. The average
titer obtained for these recombinant virus stocks was 6.0 x 108 infectious units/ml.
40
2.5.4 Design of the Fto and scrambled shRNA vectors (shRNA-Fto and shRNA-
scr)
A shRNA targeting Fto (Sigma, SHCLNG-NM_011936) and a scrambled shRNA (Addgene,
1864) were obtained. Primers were designed to amplify the U6 promoter and shRNA regions
with BamHI (5’) and NotI (3’) restriction sites (5’-3’, forward-reverse:
GTCCCGGATCCTCACCGAGGGCCTAT, TATGCGGCCGCTGATTCGGTCAACGAGG).
Regions were amplified from the pLKO.1 vector using the Phusion flash high-fidelity PCR
master mix (Thermo Scientific, F548S) and 0.25 M Betaine solution (Sigma, B0300) to reduce
hairpin formation. The sequences were then cloned into p1005(+) and the resulting vector
(Fig. 2.1C) was packaged as described for p1005-FTO. The average titer obtained for these
recombinant virus stocks was 2.2 x 109 infectious units/ml.
shRNA sequences (5’-3’):
Fto shRNA: CCGGGTCTCGTTGAAATCCTTTGATCTCGAGATCAAAGGATTTCA
ACGAGACTTTTTG
Scrambled shRNA: CCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTA
ACCTTAGG
Figure 2.1. Virus designs. Orange wedges show where sequences were cloned into the p1005(+) vector for
p1005-FTO and shRNA-Fto. AmpR = ampicillin resistance gene; oriS = HSV origin of replication. A. p1005-FTO
overexpression vector (9.5 kb). Fto coding sequence (CDS) inserted under an IE4/5 promoter. eGFP expressed
under a CMV promoter. B. Cas9-Fto knockdown vector (9kb). Fto sgRNA expressed under a U6 promoter. Cas9-
2A-GFP expressed under a Synapsin promoter. C. shRNA-Fto knockdown vector (9 kb). Fto shRNA inserted under
U6 promoter after the IE4/5 promoter. eGFP expressed under a CMV promoter.
41
2.6 Statistical Analysis
For organ and CFC enrichment experiments, one-way analyses of variance (ANOVA) were used
to test for population differences (Fig. S1), followed either by Tukey honest significant
differences (Fig. 3.1) or Dunnett (Fig. 3.2 and 3.3) post-hoc analyses. Specifically, the Dunnett
post-hoc test was used for tests where only comparisons between the four or six different
conditions and the naïve control group were considered (Dunnett, 1955). No significant
differences were observed in shock-only or context-only groups between different time points for
any of the genes analyzed (Fig. S1), and thus these groups were collapsed into a single context-
only or a single shock-only group (Fig. 3.2). For virus validation experiments, unpaired t-tests
were conducted to compare the cells infected with viruses manipulating Fto levels to control
cells, for each time point (Fig. 3.4), with Welch’s corrections when required for unequal
variances (Fig. 3.4A-B). Adjusted p-values below 0.05 are considered significant. Statistical
analyses were performed using GraphPad Prism 7.00.
42
Chapter 3
Results
Chapter 3
3.1 m6A Pathway Components Show Distinct Tissue Expression
Patterns
In order to establish whether any m6A pathway components were enriched in the brain compared
to other regions, we measured the mRNA levels of the different components in the brain, liver,
kidney and lung. Levels of the m6A methyltransferase Mettl3 were similar across all different
tissues (Fig. 3.1A, F3,10 = 1.12, p > 0.05), whereas Mettl14 (Fig. 3.1B, F3,10 = 7.82, p < 0.05) and
both m6A binding proteins, Elavl1 (Fig. 3.1E, F3,10 = 3.97, p < 0.05) and Ythdf2 (Fig. 3.1F,
F3,10 = 3.94, p < 0.05), showed differences as revealed by one-way ANOVAs. Tukey post-hoc
analyses revealed that Mettl14 expression in the brain was higher than in the lung, whereas
Elavl1 and Ythdf2 expression was higher in the liver than in the lung. The same analyses
revealed that both m6A demethylases displayed opposite patterns of expression from one
another. Whereas Alkbh5 showed higher expression in the liver and kidney than in the brain and
lung (Fig. 3.1C, F3,10 = 25.24, p < 0.05), Fto expression in the brain was enriched in the brain as
compared to all three other organs, but also higher in the liver than in the lung (Fig. 3.1D,
F3,10 = 25.11, p < 0.05).
3.2 m6A Pathway Component Levels are Regulated During Learning
To determine whether m6A pathway components are regulated during learning, mice underwent
CFC and the mRNA levels of the different components in dorsal CA1 were measured at different
time points following CFC. For the control groups that underwent context-only or shock-only
treatments, none of the m6A pathway components investigated showed differences in expression
levels at the different time points as revealed by one-way ANOVA (Fig. S1). Thus, all time
points for each treatment were merged into one group for comparison to the CFC time points.
43
Figure 3.1. m6A pathway components show distinct tissue expression patterns. A. No tissue enrichment found
for Mettl3 (F3,10 = 1.12, p > 0.05). B. Mettl14 is enriched in the brain compared to the lung (F3,10 = 7.82, p < 0.05).
C. Alkbh5 is enriched in the liver and kidney compared to the brain and lung (F3,10 = 25.24, p < 0.05). D. Fto is
enriched in the brain compared to the liver, kidney and lung, and enriched in the liver compared to the lung
(F3,10 = 25.11, p < 0.05). E. Elavl1 is enriched in the liver compared to the lung (F3,10 = 3.97, p < 0.05). F. Ythdf2 is
enriched in the liver compared to the lung (F3,10 = 3.94, p < 0.05). n = 4 (brain, kidney), 3 (liver, lung). *p < 0.05.
44
Figure 3.2. m6A pathway components are regulated during learning. A. Mettl3 is upregulated 30 min after CFC
compared to naïve controls (F6,114 = 7.33, p < 0.05). B. Mettl14 is downregulated 30 min after CFC compared to
naïve controls (F6,102 = 7.51, p < 0.05). C. No changes found for Alkbh5 (F6,82 = 1.94, p > 0.05). D. Fto is
downregulated 30 min and 2h after CFC compared to naïve controls (F6,117 = 6.61, p < 0.05). E. Elavl1 is
downregulated 30 min and 2h after CFC compared to naïve controls (F6,114 = 3.41, p < 0.05). F. Ythdf2 shows
changes in regulation, but pairwise comparisons did not reveal differences (F3,10 = 2.75, p < 0.05). Dashed black line
represents naïve control levels. n = 19-23 (naïve), 18-37 (context-only), 18-33 (shock-only), 8-10 (30 min after
CFC), 7-8 (1h after CFC), 8-11 (2h after CFC), 7-8 (4h after CFC). *p < 0.05.
45
Following CFC, both methyltransferases showed changes in expression levels as revealed by
one-way ANOVA (Mettl3: Fig. 3.2A, F6,114 = 7.33, p < 0.05; Mettl14: Fig. 3.2B, F6,102 = 7.51,
p < 0.05). Dunnett post-hoc analyses revealed that both Mettl3 and Mettl14 levels were different
30 min after CFC as compared to naïve mice, with Mettl3 levels increasing by nearly 70%, and
Mettl14 levels decreasing by just over 30%. Fto levels also showed changes in expression
(Fig. 3.2D, F6,117 = 6.61, p < 0.05) with a decrease at both 30 min and 2h after CFC of just over
35% and 50% respectively as compared to naïve mice. In contrast, the other demethylase,
Alkbh5, did not show changes in expression following CFC (Fig. 3.2C, F6,82 = 1.94, p > 0.05).
Lastly, both m6A binding proteins showed changes in expression (Elavl1: Fig. 3.2E,
F6,114 = 3.41, p < 0.05; Ythdf2: Fig. 3.2F, F3,10 = 2.75, p < 0.05). Lastly, Elavl1 showed a 40%
decrease at 30 min following CFC. However, pairwise comparisons of Ythdf2 levels in the
different conditions with naïve mice did not reveal any differences.
3.3 m6A Levels in mRNA are Regulated During Learning
As the levels of the m6A pathway components were regulated following CFC, we then
investigated whether m6A levels in total RNA and mRNA were regulated at different time points
following CFC. Whereas m6A levels in total RNA did not show changes following CFC as
revealed by one-way ANOVA (Fig. 3.3A, F4,22 = 0.02, p > 0.05), m6A levels in mRNA enriched
samples did (Fig. 3.3B, F4,22 = 5.48, p < 0.05). Specifically, Dunnett post-hoc analyses revealed a
change in m6A levels in mRNA both 30 min and 4h following CFC, corresponding to an
approximately 140% increase compared to naïve mice.
3.4 In Vitro Validation of HSVs
Given that Fto showed enrichment in the brain, and was regulated during learning, we selected
the demethylase FTO as the target for the HSVs we designed to manipulate the m6A pathway.
These viruses were then validated in vitro in N2a cells, a line of immortalized neural precursor
cells, to verify their effects on Fto and m6A.
46
Figure 3.3. m6A levels in mRNA are regulated during learning. A. No changes were found for m6A levels in
total RNA (F4,22 = 0.02, p > 0.05). B. m6A levels in mRNA enriched samples are upregulated 30 min and 4h after
CFC (F4,22 = 5.48, p < 0.05). Dashed black line represents naïve control levels. n = 19-23 (naïve), 18-37 (context-
only), 18-33 (shock-only), 8-10 (30 min after CFC), 7-8 (1h after CFC), 8-11 (2h after CFC), 7-8 (4h after CFC).
n = 10 (naïve), 10 (30 min after CFC), 4 (1h after CFC), 5 (2h after CFC), 4 (4h after CFC). *p < 0.05.
3.4.1 p1005-FTO overexpression virus
As compared to cells infected with the control virus, p1005(+), N2a cells infected with the
p1005-FTO overexpression virus had higher levels of Fto both 12h (Fig. 3.4A, t5 = 3.67,
p < 0.05) and 2 days (Fig. 3.4B, t7 = 6.05, p < 0.05) after infection. Specifically, Fto levels in
p1005-FTO infected cells were approximately 1200% and 600% higher at 12h and 2 days
respectively, as compared to control cells. m6A levels in total RNA were 40% lower after 2 days
in p1005-FTO cells as compared to control cells (Fig. 3.4C, t4 = 16.0, p < 0.05).
3.4.2 Cas9-Fto knockdown virus
As compared to cells infected with the control virus, Cas9-ctrl, N2a cells infected with the Cas9-
Fto knockdown virus had lower levels of Fto both 2 (Fig. 3.4D, t10 = 4.0, p < 0.05) and 4 days
(Fig. 3.4E, t10 = 3.31, p < 0.05) after infection. Specifically, Fto levels in Cas9-Fto infected cells
were approximately 25% and 55% lower at 2 and 4 days respectively, as compared to control
cells. m6A levels in total RNA were 50% higher after 4 days in Cas9-Fto cells as compared to
control cells (Fig. 3.4F, t4 = 2.84 , p < 0.05).
47
Figure 3.4. In vitro validation of HSVs. A. Fto levels were upregulated compared to controls 12h after infection
with the p1005-FTO overexpression virus (t5 = 3.67, p < 0.05). B. Fto levels were upregulated compared to controls
2 days after infection with the p1005-FTO overexpression virus (t7 = 6.05, p < 0.05). C. m6A levels were
downregulated compared to controls 2 days after infection with the p1005-FTO overexpression virus (t4 = 16.0,
p < 0.05). D. Fto levels were downregulated compared to controls 2 days after infection with the Cas9-Fto
knockdown virus (t10 = 4.0, p < 0.05). E. Fto levels were downregulated compared to controls 4 days after infection
with the Cas9-Fto knockdown virus (t10 = 3.31, p < 0.05). F. m6A levels were upregulated compared to controls
4 days after infection with the Cas9-Fto virus (t4 = 2.84, p < 0.05). G. Fto levels were downregulated compared to
controls 2 days after infection with the shRNA-Fto knockdown virus (t8 = 3.59, p < 0.05). H. Fto levels were
downregulated compared to controls 3 days after infection with the shRNA-Fto knockdown virus (t10 = 2.99,
p < 0.05). I. m6A levels were upregulated compared to controls 3 days after infection with the shRNA-Fto virus
(t4 = 5.97, p < 0.05). n = 7-9 (p1005-FTO, mRNA), 4-6 (mRNA Cas9- and shRNA-Fto), 3 (m6A). *p < 0.05.
48
3.4.3 shRNA-Fto knockdown virus
As compared to cells infected with the control virus, shRNA-scr, N2a cells infected with the
shRNA-Fto knockdown virus had lower levels of Fto both 2 (Fig. 3.4G, t8 = 3.59, p < 0.05) and
3 days (Fig. 3.4H, t10 = 2.99, p < 0.05) after infection. Specifically, Fto levels in Cas9-Fto
infected cells were approximately 70% and 30% lower at 2 and 3 days respectively, as compared
to control cells. m6A levels in total RNA were 35% higher after 3 days in Cas9-Fto cells as
compared to control cells (Fig. 3.4I, t4 = 5.97 , p < 0.05).
49
Chapter 4
Discussion
Chapter 4
Memory formation critically relies on local mRNA translation (Miller et al., 2002; Ouyang et al.,
1999). However, the mechanism regulating the trafficking and translation of individual
transcripts to and at specific spines in neurons during memory formation is poorly understood.
Such a mechanism must be stimulus-induced, dynamic and flexible, like a modifiable signal
present in mRNA transcripts that enables differential processing of individual transcripts. The
m6A modification of RNA emerges as a promising candidate as it appears to possess all these
characteristics (Fu et al., 2014a), but has not yet been studied in memory. This study aimed to
begin this investigation by establishing whether certain components of the m6A pathway are
enriched in the brain and regulated during memory formation, and developing tools for future
probing of the role of m6A in memory. Our results show that the m6A demethylase Fto is
particularly enriched in the brain, and that several m6A pathway components are regulated
following CFC. Overexpression and knockdown viruses targeting Fto were designed and
validated in vitro for future probing of the role of m6A in memory.
4.1 Fto is Enriched in the Brain
We hypothesized that some of the components of the m6A pathway would be enriched in the
brain, as compared to other organs, and this is indeed the case for the demethylase Fto which
shows over 50% higher expression in the brain than in the lung, kidney, or lung (Fig. 3.1D).
Although the other components are not enriched in the brain, they do show different relative
gene expression patterns from one another across the different organs. Specifically, the other
demethylase, Alkbh5, displays an almost opposite pattern to Fto, as it is expressed 200-300%
more in the liver and kidney respectively than in the brain (Fig. 3.1C). In addition, many
components show much lower relative enrichment in the lung (Fig. 3.1). As m6A has not been
studied extensively in the lung, it is unclear whether this finding reflects a less dynamic
engagement of the m6A pathway in this organ or whether further investigation of the pathway
50
will reveal additional components enriched in the lung. Altogether, therefore, these findings
suggest that the m6A pathway is engaged differently by different organs, possibly enabling them
to harness m6A for distinct purposes. In addition, the enrichment patterns of Fto suggest it may
be particularly important for brain-related m6A functions.
In supporting distinct roles for FTO and ALKBH5 in the brain, these results complement
previous findings that showed that FTO may play a role in brain-related functions
(Bressler et al., 2013; Church et al., 2010; Keller et al., 2011), whereas Alkbh5 is particularly
expressed in testes and involved in spermatogenesis (Zheng et al., 2013). Indeed, as with Alkbh5,
expression patterns across organs and cells often hint at the function of specific genes
(Cahoy et al., 2008; Zetterström et al., 1996; Zheng et al. 2013). Given that m6A is involved in
many distinct cellular functions, like differentiation (Geula et al., 2015), adipocytosis
(Zhang et al., 2015) and circadian rhythm regulation (Fustin et al., 2013), it is possible that the
expression levels of its pathway components are critical to allowing it to have different functions
in different tissues. This shifting of the balance between the demethylases toward Fto, for
instance, might enable the m6A pathway in the brain to exploit its particularities, like the
particular transcripts it targets or the intermediates, hm6A and f6A, that are produced during
demethylation by FTO, but not ALKBH5 (Chen et al., 2014; Fu et al., 2013). For instance, if, as
suggested by the evidence from Hess et al. (2013), FTO does specifically target PRP transcripts
in the brain, its enriched levels in the brain might be key to allowing the m6A pathway to
regulate memory-related processes like synaptic plasticity. Thus, altogether, our findings further
corroborate a potential role for FTO in the brain and in memory.
Although the methylation and binding components of the m6A pathway that were investigated
did not show any specificity in the brain, they are well expressed in this organ and may
nonetheless be involved in shaping brain-related m6A function, not through their basal
expression, but through their dynamic regulation during memory formation. Thus, in addition to
supporting a particular role for FTO in regulating m6A-related functions in the brain, our results
also support a possible role for the other pathway components, as these are also well expressed in
the brain. If these components are indeed involved in memory, it is likely that, similar to many
PRPs and IEGs (Barco et al., 2005; Miller & Sweatt, 2007; Ramírez-Amaya et al., 2005), their
levels are regulated in memory-related brain regions like dorsal CA1 during memory formation.
51
4.2 The m6A Pathway is Extensively Regulated During Memory
Formation
Our second hypothesis was that m6A pathway components are regulated during memory
formation. In agreement with the hypothesis, several components show changes in gene
expression in the dorsal CA1 region of the hippocampus at different time points following CFC.
This is not the case for shock-only and context-only controls (Fig. S1), suggesting that the
regulation of the m6A pathway seen after CFC is not simply due to stress or exposure to a novel
environment, but rather to the formation of a contextual memory.
As Fig. 3.2 shows, the m6A pathway is most widely regulated 30 min after CFC. Indeed, the
methyltransferase Mettl14, the demethylase Fto and the binding protein Elavl1, are all
downregulated, whereas Mettl3 is upregulated. Furthermore, as seen in Fig. 3.3, these changes
are accompanied by an increase in overall m6A levels in mRNA, but not total RNA. Then, 2h
after CFC, both Fto and Elavl1 are downregulated again, whereas m6A levels remain at baseline
levels until they rise again 4h after CFC. By relating these changes to what is known from the
literature about the different components, one can start to piece together the dynamics of the
m6A pathway.
4.2.1 Regulation of m6A by methyltransferase and demethylase levels
The overall increase in mRNA m6A at 30 min (Fig. 3.3B) may be accounted for by both the
upregulation of Mettl3 and the downregulation of Fto. Indeed, although METTL3 and METTL14
are thought to methylate m6A as part of a complex (Fu et al., 2014a), overexpressing either
component alone leads to an increase in m6A levels (Liu et al., 2014; Wang et al., 2015b), and
knocking down either reduces m6A levels (Bokar et al., 1997; Dominissini et al., 2012; Geula et
al., 2015; Liu et al., 2014; Ping et al., 2013; Wang, Y. et al., 2014). Thus, both
methyltransferases appear to be able to function independently from one another. This would
allow them to have distinct functions in memory, as suggested by the fact that they are regulated
in opposite directions 30 min after CFC (Fig. 3.2A-B).
For instance, if METTL3 and METTL14 target different groups of transcripts in the brain, such
as PRPs and proteins involved in baseline cell metabolism respectively, their differential
regulation at 30 minutes could enable a specific increase in synaptic targeting and translation of
52
PRPs during memory formation. Indeed, although one study found that METTL3 is able to act
independently of its ability to methylate adenosine by directly promoting mRNA translation
(Lin et al., 2016), it has also clearly been implicated in methylating RNA (Bokar et al., 1997;
Dominissini et al., 2012; Geula et al., 2015; Liu et al., 2014; Ping et al., 2013; Wang et al.,
2015b; Wang, Y. et al., 2014). Thus METTL3 could be primarily driving the enrichment in
mRNA m6A levels seen at 30 minutes, targeting transcripts involved in memory formation
specifically, and compensating for the decrease in METTL14.
The other factor that could explain the upregulation of m6A at 30 min is the downregulation of
the demethylase Fto 30 min after CFC (Fig. 3.2D). Interestingly, Alkbh5 does not show any
regulation (Fig. 3.2C), fitting with the idea that its role in the m6A pathway is less prominent in
the brain than in other tissues. It is also possible that ALKBH5 is critical to maintaining baseline
transcript processing in the cell, whereas FTO is involved in memory-related processing.
Importantly however, Fto is downregulated again at 2h (Fig. 3.2D), when no change is seen in
m6A levels, and returns to baseline at 4h when m6A is enriched again (Fig. 3.3B). This could be
accounted for if major changes in m6A occurring in the nucleus drown out local changes in m6A
at the synapses involved in a memory. For instance, if the methyltransferase complex functions
primarily in the nucleus, methylating transcripts for transport to the synapses, and Fto acts
primarily at synapses, demethylating arriving transcripts, an increase in mRNA m6A levels at the
synapse at 2h caused by Fto downregulation could be undetected if m6A levels are held constant
in the nucleus. Then at 4h, the observed increase in m6A levels could be driven by an increase in
methylation in the nucleus. Since neither Mettl3 nor Mettl14 is upregulated at 4h, it is likely that
an additional methyltransferase component is upregulated or activated at this time, and possibly
enabling specific increases in m6A in PRP transcripts required for memory formation. Such
components could include WTAP, although it does not catalyze methylation and its
overexpression has not been shown to increase m6A levels in vitro (Liu et al. 2014), or
additional as yet unidentified methyltransferase components.
Overall, these findings show that at least three, and possibly more regulators of m6A levels, each
of which may target different sets of transcripts, are all regulated during memory formation. This
demonstrates the remarkable flexibility of the pathway, further supporting a potential role in
regulating complex cellular functions. How these changes in m6A levels in different transcripts
53
affect memory formation depends, however, on how these changes are recognized by m6A
binding proteins and enable changes in transcript processing.
4.2.2 Downstream effects of m6A regulation on transcript processing
In changing the levels of m6A in transcripts during memory formation, cells are likely enabling
changes in transcript processing by m6A binding proteins like ELAVL1 and YTHDF2. The
regulation of these proteins following CFC provides indications of how this may be occurring.
During the first wave of m6A enrichment, at 30 min, neither of the binding proteins is
upregulated. Elavl1, which has independently been linked to promoting mRNA translation and
stability (Brennan et al., 2001; Durie et al., 2010), and regulating splicing (Lebedeva et al., 2011;
Mukherjee et al., 2011), is in fact downregulated at this time point (Fig. 3.2E). This indicates that
ELAVL1’s functions in the m6A pathway are being suppressed at this stage in memory
formation, in favour perhaps of YTHDF2, or other binding proteins like YTHDF1 or 3. Indeed,
YTHDF2 has been shown to compete with FTO and promote translation during heat shock
(Zhou et al., 2015). Thus, it is possible that the decrease in Fto releases the brake on YTHDF2,
allowing it to bind m6A-containing transcripts and increase their translation. However,
depending on where m6A occurs in transcripts, YTHDF2 binding may also promote degradation
(Wang, X. et al., 2014). Thus, the location of m6A in transcripts could enable translation of
PRPs at 30 min and degradation of transcripts suppressing synaptic plasticity via YTHDF2.
Interestingly, when m6A levels rise again at 4h, neither Elavl1 nor Ythdf2 is upregulated
(Fig. 3.2E-F). This may be due to the functions of both components being equally engaged at this
stage in memory formation. An alternative explanation, however, is that just as changes in m6A
may occur primarily in the nucleus, ELAVL1 and YTHDF2 may operate primarily at synapses
during memory formation. Depending on the length of the dendritic arbour, it can take RNPs
containing mRNA an hour or two to travel to spines (Dynes & Steward, 2007; Knowles et al.,
1996; Steward et al., 1998), and thus Fto and Elavl1 may be modulated at 2h in response to the
arrival of mRNAs affected by the increase in m6A levels in the nucleus at 30 min. It would then
follow that the increase in m6A levels at 4h would be linked to changes in Fto and Elavl1 levels
at 5 or 6h. This possibility is further discussed in the model presented in Fig. 4.1.
Lastly, it is possible that it is the location of new m6A sites in mRNA, and not the levels of
different binding proteins, that allows some proteins to be engaged rather than others. Both
54
YTHDF2 and ELAVL1 are sensitive to the location of m6A in a transcript (Wang, X. et al.,
2014; Wang, Y. et al., 2014; Zhou et al., 2015), and thus may not need to be regulated
themselves in order to drive an increase in translation of PRPs and decrease in translation of
plasticity-suppressing proteins. This feature of m6A adds an important layer of complexity to the
m6A pathway, and further supports its flexibility and ability to finely regulate complex cellular
functions in memory.
4.2.3 Biphasic regulation of m6A and other memory-related processes
Evidently the m6A pathway is highly flexible, and dynamically regulated during memory
formation. Although causal evidence is needed to confirm the role of m6A in memory, its
biphasic regulation during memory formation provides a strong parallel with many key memory-
related processes. Indeed, IEG expression, gene transcription, protein synthesis and LTP all show
biphasic regulation during memory formation, with a first wave typically occurring in the first
three hours following a memory task, and the second wave, three to six hours after (Lynch, 2004;
Igaz et al., 2002; Quevedo et al., 1999; Ramírez-Amaya et al., 2005). The biphasic pattern
observed for m6A fits well within these windows, occurring at 30 min and 4h. Although, the
waves of Fto and Elavl1 downregulation waves are closer together, occurring at 30 min and 2h,
their lack of synchrony with m6A could be accounted for if, as proposed, they are regulated at
spines, and m6A is upregulated in the nucleus. Thus, neurons could first upregulate m6A at the
nucleus at 30 min to promote transport of PRP transcripts involved in the first wave of memory-
related translation, while enabling local translation of transcripts already present in spines by
decreasing Fto and Elavl1 levels. The downregulation of Fto and Elavl1 at 2h might then enable
translation of newly arrived PRP transcripts (Fig. 4.1). This process might then be repeated at 4h
for PRPs involved in the second wave of memory-related translation (Lynch, 2004). Thus, in
demonstrating the temporal similarities between the regulation of the m6A pathway, and that of
transcription and translation during memory formation, our results further corroborate the
potential of m6A as a regulator of fine-tuned local mRNA translation in memory. Importantly,
this hypothesis is also supported by the fact that the changes in m6A levels following CFC occur
specifically in mRNA.
55
4.3 m6A is Specifically Regulated in mRNA During Memory
Formation
As shown in Fig. 3.3, both increases in m6A levels appear to be specific to mRNA. This finding
points strongly to a role for m6A in regulating transcript processing during memory formation,
and possibly local translation. Importantly, the fact that these increases are not reflected in total
RNA is not surprising given that 94% of m6A sites occur in mRNA (Meyer et al., 2012) and, by
weight, mRNA constitutes only about 4% of the total RNA isolated from a mouse brain
(Alberts et al., 2015). Thus, any changes in m6A levels specific to mRNA would be drowned out
in the total RNA sample.
To date, none of the methyltransferases or demethylases have been shown to specifically target
mRNA over other RNAs such as tRNA and rRNA. These findings demonstrate however that
such targeting is indeed occurring. The ability of FTO to target non coding RNA in addition to
mRNA, at least in vitro, is clearly demonstrated by our findings in Fig. 3.4 where manipulating
Fto levels affected overall m6A levels in total RNA. Whether this is also the case in the brain, or
whether it is other components, like METTL3, that specifically enable an increase in mRNA
m6A levels remains to be established. Nonetheless, it is clear that m6A pathway components are
able to target mRNA specifically, and it is also possible that they are able to target specific
transcripts, like PRPs, as well.
All in all, the finding that both increases in m6A in CA1 following CFC are specific to mRNA
certainly implicates m6A in regulating memory-related mRNA processing, and possibly in
coordinating critical aspects of local mRNA translation, like transport and local translation of
PRPs. This is further supported by Hess et al. (2013)’s finding that FTO may primarily target
PRP transcripts in the brain, and several studies that have linked m6A to transcript transport and
translation in vitro (Dominissini et al., 2012; Durie et al., 2010; Fustin et al., 2013;
Merkestein et al., 2014; Wang, X. et al., 2014; Wang, Y. et al., 2014; Wang et al., 2015b;
Zheng et al., 2013; Zhou et al., 2015). Thus, the enrichment of Fto in the brain, and dynamic
regulation of m6A pathway components, and of mRNA m6A levels in dorsal CA1 during
memory formation all corroborate a potential role for m6A in regulating plasticity, and possibly
the local translation of PRPs. To bring together our findings and the ideas proposed here into a
56
cohesive picture, I propose a speculative model of how m6A may be regulating local mRNA
translation during memory formation.
4.4 Speculative Model of the Role of m6A in Regulating Local mRNA
Translation During Memory Formation
As simplifications of and speculations about reality, most models are false, and some more false
than others. Nonetheless, it can be useful to build a speculative model when observing novel
phenomena. Indeed, models provide a framework in which disparate pieces of a puzzle are
brought together to produce specific hypotheses. These hypotheses can then be tested to establish
more realistic models (Wimsatt, 2007). Thus, in order to explore the possible role the m6A
pathway could play in regulating local mRNA translation, I have devised a speculative model of
their interactions in Fig. 4.1. Specifically, this model shows how the changes found in Fig. 3.2
and 3.3 in m6A pathway component levels during memory formation could contribute to
regulating mRNA transport to spines and local translation of mRNA at synapses during memory
formation or L-LTP. The example given is of a pyramidal neuron and one of its dendritic spines
located in the CA1 region of a mouse learning a memory task. As previously suggested, this
model proposes that in regulating memory, the methyltransferase complex may function
primarily in the nucleus, adding m6A signals to individual transcripts as they are produced,
whereas the demethylase FTO and binding proteins may operate primarily at the synapse in
spines, responding to the m6A signals when transcripts arrive.
In this model, upon transcription, each mRNA is labelled with two m6A signals by the
methyltransferase complex: one synapse specific transport signal, and one translation-related
signal. The colours used to distinguish the signals from one another in Fig. 4.1 correspond to
different patterns of m6A sites in transcripts. In the naïve state (Fig. 4.1A), when all pathway
components are at basal states, transcripts receive an m6A transport signal specifying their
synaptic destination and an m6A translation suppressing signal. They are then packaged into
RNPs (Bramham & Wells, 2007), likely through recognition by m6A binding proteins (not
shown in Fig. 4.1), and shipped to their synaptic destination. At the spine, FTO suppresses
translation of transcripts by removing m6A translation promoting signals from transcripts, and
preventing YTHDF2 or other m6A binding proteins from recruiting the transcripts to the
57
translation machinery (Merkestein et al., 2014; Zhao et al., 2014; Zhou et al., 2015). ELAVL1
binds to the m6A translation suppressing signal on certain transcripts, stabilizing and silencing
them (Brennan et al., 2001).
At the 30 min time point (Fig. 4.1B), after the memory task, the upregulation of METTL3 and
downregulation of METTL14 change the composition of the methyltransferase complex in the
nucleus, leading to an increase in the number of transcripts bearing an m6A signal and being sent
to synapses. Furthermore, these transcripts, involved in the first wave of plasticity-related
translation, now bear an m6A translation promoting signal, such that they may be translated
immediately upon arrival. At the synapse, the downregulation of FTO and ELAVL1 releases
translation suppression, allowing PRP transcripts that are already present to be recruited to the
translation machinery, through binding to YTHDF2 or independently (Zhou et al., 2015).
At the 1h time point (Fig. 4.1C), m6A pathway components return to baseline levels, and the rate
of transcript methylation in the nucleus slows down to a normal rate in the presence of basal
METTL3 and METTL14 levels. Newly generated transcripts receive an m6A translation
suppressing signal, as in the naïve state. At the spine, translation is silenced by the return of FTO
and ELAVL1. Recently translated PRPs strengthen the synapse (Bliss & Collingridge, 1993;
Grooms et al., 2006; Lynch, 2004; Mayford et al., 1996; Miyashiro et al., 1994; Paradies &
Steward, 1997), shown in Fig. 4.1 as an increase in receptor numbers at the synapse. Meanwhile,
RNPs travel through the dendritic arbour toward their synaptic destination (Dynes & Steward,
2007; Knowles et al., 1996; Steward et al., 1998).
At the 2h time point (Fig. 4.1D), the state in the nucleus remains the same. At the spine, FTO and
ELAVL1 are downregulated again, as RNPs reach their destination and deliver a new batch of
mRNA. Transcripts are recruited to the translation machinery, through binding to YTHDF2 or
independently, and PRPs are locally synthesized (Ramírez-Amaya et al., 2005).
58
59
Figure 4.1. (previous page) Speculative model of the role of m6A in regulating local mRNA translation during
memory formation. A pyramidal neuron and one of its dendritic spines (the target of orange m6A synapse specific
targeting signals) located in the dorsal CA1 region of the hippocampus of a mouse learning a memory task are
shown. In this model, methyltransferase complexes operate primarily in the nucleus, adding m6A signals to
individual transcripts as they are produced, whereas the demethylase FTO and binding proteins operate primarily at
spines, responding to the m6A signals when transcripts arrive. Upon transcription, each mRNA is labelled with two
m6A signals by the methyltransferase complex: one synapse specific transport signal, and one translation-related
signal. A. Naïve state: pathway components are at basal states, transcripts with m6A translation suppressing signals
are packaged and shipped in RNPs to synaptic destinations. At the spine, translation is suppressed due to transcript
demethylation by FTO, preventing YTHDF2 from inducing translation, and ELAVL1 binding to m6A translation
suppression signals. B. 30 min after memory task: changes in METTL3 and METTL14 levels change the
composition of the methyltransferase complex, leading to an increase in transcript methylation and trafficking, and
adding of m6A translation promoting signals to transcripts. At the spine, FTO and ELAVL1 are downregulated,
releasing translation suppression. PRP transcripts are then recruited to the translation machinery, through binding to
YTHDF2 or independently. C. 1h after memory task: m6A pathway components return to baseline levels, transcript
methylation in the nucleus slows down, and switches back to adding m6A translation suppressing signals to
transcripts. At the spine, FTO and ELAVL1 inhibit translation. Recently translated PRPs strengthen the synapse.
This is shown as an increase in receptors inserted at the synapse. Meanwhile, RNPs travel through the dendritic
arbour toward their synaptic destination. D. 2h after memory task: At the spine, FTO and ELAVL1 are
downregulated again. RNPs reach their destination and deliver a new batch of mRNAs, which are recruited to the
translation machinery, through binding to YTHDF2 or independently. PRPs are locally synthesized. E. 4h after
memory task: FTO and ELAVL1 return to baseline in the spine, inhibiting protein translation. The second batch of
PRPs contribute to strengthening the synapse. This is again shown as an increase in receptors inserted at the
synapse. In the nucleus, the second phase of transcription is initiated and additional PRP transcripts are methylated
and shipped to their synaptic destination. F. Longer-lasting synaptic potentiation critical to memory formation is
solidified through arrival and translation of the second batch of transcripts. Through m6A pathway regulation, the
spine targeted by orange m6A synapse specific targeting signals has thus specifically been potentiated to help
encode the memory task.
At the 4h time point (Fig. 4.1E), FTO and ELAVL1 return to baseline at the spine, halting
protein translation, and the second batch of PRPs contributes to strengthening the synapse. In the
nucleus, the second phase of transcription is initiated and additional transcripts involved in the
second wave of plasticity-related translation are methylated and shipped to their synaptic
destinations (Igaz et al., 2002). As time points beyond 4h were not explored in this study, the
specifics of the model end here. However, it is assumed that upon arrival at the spine, these
transcripts are likely translated and contribute to the longer-lasting potentiation observed in
synapses critical to a new memory (Bliss & Collingridge, 1993; Bradshaw et al., 2003; Lynch,
60
2004). Thus, by regulating m6A component levels in PRP transcripts, the neuron was able to
specifically target this synapse to be potentiated and encode the new memory (Fig. 4.1F).
4.5 Overview of the Potential Role of m6A in Memory
As mentioned previously, it is likely that key components of the m6A pathway are missing from
the model, and that the details proposed in Fig. 4.1 are different from the actual mechanisms at
play in real neurons. Nonetheless, the model provides a general idea of how m6A might fine-
tune regulation of local mRNA translation, and more importantly demonstrates how powerful
and flexible the dynamics of such a richly regulated pathway are.
If m6A pathway components are indeed involved in regulating local mRNA translation, it is
likely through interactions with known broad regulators of dendritic mRNA transport and local
protein synthesis. By recognizing m6A patterns in mRNA, m6A binding proteins, for example,
could be directing these regulators to specific transcripts. Specifically, if ELAVL1 and YTHDF2
are indeed involved in regulating translation in spines, it is likely that they interact with
previously identified local protein synthesis regulators like CPEB, FMRP and miR-134 (Kim &
Richter, 2006; Schratt et al., 2006; Stebbins-Boaz et al., 1999; Zalfa et al., 2003; Zhang et al.,
2001). If, these or other m6A binding proteins are acting in the nucleus, they may also direct
dendritic transport proteins, like RNPs (Dynes & Steward, 2007; Knowles et al., 1996;
Vessey et al., 2006), to transcripts specifically labelled by m6A. Thus, m6A may provide the
mechanism by which broad mRNA processing mechanisms can be targeted specifically to those
transcripts required for memory formation.
Furthermore, although the speculative model presented in Fig. 4.1 suggests that some m6A
pathway components are active and regulated in spines, it is possible that the pathway is entirely
regulated in the cell body. Indeed, since the findings presented in Fig. 3.2 and 3.3 reflect overall
changes in m6A pathway component levels in dorsal CA1, and not changes within different
neuronal compartments, one cannot conclude that the m6A pathway is acting in spines. In fact,
expression patterns of m6A pathway components have not yet been well studied within neurons,
and thus it is not yet known which components are expressed in dendrites. If it is in fact the case
that m6A components are indeed primarily expressed and regulated in the cell body, this would
61
be compatible with the role proposed in Fig. 4.1 for m6A of regulating mRNA transport from the
nucleus, but not the proposed role of locally regulating translation in spines. However, as protein
synthesis is also required in the soma during memory formation (Huber et al., 2000; Lasek et al.,
1970), m6A pathway components could still be involved in fine-tuned regulation of somatic
translation. A similar model to that presented in Fig. 4.1 could therefore be considered in which
FTO, ELAVL1 and YTHDF2 promote degradation or translation of transcripts in the soma. In
this model, m6A would enable fine-tuned regulation of the levels of specific PRPs before they
are trafficked to their cellular destination, for example through synaptic tagging, and enable
neuronal plasticity and memory formation (Frey & Morris, 1997; Redondo & Morris, 2011).
Further research looking at where m6A pathway component levels are regulated, and what stage
of mRNA processing they affect in memory will help establish which of these two models best
describes the role of m6A.
However, before these models and their proposals can be tested, a causal link must first be
established showing that m6A regulation is indeed critical for memory formation. Second, it will
be important to identify which transcripts show changes in m6A levels during memory formation
and how these changes affect their processing. Third, establishing where in the cell different
m6A pathway components are regulated during memory formation will provide important
insight into whether m6A is indeed involved in local regulation of mRNA processing in memory,
or whether it is acting primarily in the cell body.
To begin tackling the first question, it is critical to develop tools to target the pathway directly
and observe the effects on memory. Given its enriched levels in the brain, links to cognitive
function in humans (Bressler et al., 2013; Keller et al., 2011), and regulation during memory
formation, Fto is an excellent target for which to begin developing these tools.
4.6 Development of HSV Gene Manipulation Tools to Probe the Role
of m6A in Memory
4.6.1 Overexpression and knockdown vectors
Our third aim was to develop HSVs to manipulate Fto, and m6A levels in cells. Three viruses
were developed, namely one overexpression virus and two knockdown viruses, and their effects
62
on Fto levels were measured, as well as their functional effects on manipulating m6A levels. All
viruses express well in N2a cells, as seen by the proportion of cells expressing GFP which was
approximately 80-90% for cells infected with p1005(+), p1005-FTO, shRNA-scr and shRNA-
Fto, and 60-70% for cells infected with Cas9-Fto and Cas9-ctrl, likely due to the two-fold lower
titer of these viruses. The p1005-FTO overexpression virus robustly increases Fto levels, and
decreases overall m6A levels, demonstrating its functional effect (Fig. 3.4A-C). The Cas9-Fto
and shRNA-Fto viruses show the opposite effects, both leading to a decrease in Fto levels, and a
corresponding increase in m6A levels in total RNA (Fig. 3.4D-I). Thus, all viruses worked as
expected in vitro.
It is important to note first that, as the mechanisms of action of the different viruses are different,
distinct time points were selected for each virus. The shortest time points (12h and 2 days) were
used for p1005-FTO which increases FTO directly by increasing Fto levels. Longer time points
(2 and 3 days, or 2 and 4 days respectively) were used for the shRNA-Fto and Cas9-Fto viruses
to allow basal FTO levels to decrease naturally, and reveal the effects of knocking down Fto
mRNA on m6A. Lastly, since Cas9 must first produce a mutation in the Fto gene for a
knockdown to occur, an extra day was added to the second time point for the Cas9-Fto virus.
At 2 days, the knockdown of 25% generated by Cas9-Fto seems lower than the 75% knockdown
produced by shRNA-Fto (Fig. 3.4D, G). This may be due to the lower titer of the Cas9-Fto virus,
as well as the additional step in its knockdown mechanism. As expected, at 4 days the Cas9-Fto
knockdown increases, reaching approximately 50% (Fig. 3.4E). Contrary to expectations,
however, there is a substantial drop of approximately 50% in the level of Fto knockdown from 2
to 3 days for shRNA-Fto (Fig. 3.4G-H). A similar drop is seen in the efficiency of the Fto
overexpression from 12 hours to 2 days for p1005-FTO (Fig. 3.4A-B). This drop in efficiency
may be due to the cytotoxicity caused by the HSV helper virus which is involved in packaging
the viral vectors and is present, albeit at low concentrations, in the viral stocks (Fink et al., 1996;
Neve et al., 2005). Indeed, an increase in cell death was seen in transduced cells, expressing
GFP, particularly for p1005(+), p1005-FTO, shRNA-scr and shRNA-Fto. Cells infected with
Cas9-Fto and Cas9-ctrl may have been spared due to the lower titers of these viruses.
Furthermore, as cytotoxicity affects normal metabolic processes in cells like gene expression, it
is likely that viral gene expression is also affected (Lim, 2013). If this is the case, cells infected
with Cas9-Fto should be less sensitive, as persistent viral gene expression is not required for the
63
knockdown to persist. In fact, once Cas9 has effectively generated a loss of function mutation in
the Fto gene, the Fto knockdown is permanent in that cell and its daughter cells.
Nonetheless, given that the viruses were still effective in manipulating Fto levels at the later time
points, these time points were selected for the m6A analyses in order to allow sufficient time for
the changes in Fto mRNA levels to induce changes in FTO protein levels, and consequently in
m6A levels. In all three cases, the expected changes in m6A were observed, with a possibly
stronger effect seen for Cas9-Fto compared to shRNA-Fto (Fig. 3.4F, I). This may be due to the
difference in days from infection, and the better persistence of the Cas9 knockdown.
Overall, all viruses functioned generally as expected, and their rapid and robust effects support
the future use of these viruses in vivo to target the m6A pathway during learning and investigate
the effects on memory. Given that HSV infects neurons differently from cells in vitro
(Fink et al., 1996; Lim, 2013), it is likely that the overexpression and knockdown levels
observed in vivo will be different from those observed here. Thus, further validation of these
viruses in cultured neurons and in vivo will be critical to verifying and measuring their impact on
Fto and m6A levels in the brain.
4.6.2 Cas9-based HSV
The Cas9-based HSV in particular not only constitutes the first use of Cas9 in HSV, but will also
enable one of the first uses of Cas9 in the brain (Swiech et al., 2015). One of the principal
advantages of using Cas9 to investigate the m6A pathway, is that once the effects of knocking
down Fto on memory have been investigated, it will be relatively simple to expand our
investigation of the pathway by designing new sgRNAs targeting different m6A pathway
components, and adding them to the viral vector in order to knockdown several components at
once (Walters et al., 2015). Eventually, the vector may be redesigned to express dCas9, which
has no endonuclease activity, but can be used to recruit activators and repressors to different
genes (Gilbert et al., 2014). This will enable an even greater flexibility in targeting and
modulating gene expression levels. Overall, the successful validation of this first Cas9-based
HSV is very promising and will likely enable faster progress not only in understanding the role
of the m6A pathway in the brain and in memory, but also in addressing other gene-based
research questions.
64
4.7 Limitations
4.7.1 Missing pieces in the m6A pathway puzzle
As mentioned previously, it is likely that key components of the m6A pathway in the brain are
missing. First, as research into m6A has only recently begun to gather momentum, many
components of the m6A pathway have probably not yet been identified. Second, the dynamics
observed in the levels of methyltransferases and demethylases do not fully account for the
increases in m6A seen at 30 min and 4h (Fig. 3.3B), or the lack of change at 2h when the
demethylase Fto is downregulated (Fig. 3.2D). Third, none of the components identified in the
m6A pathway to date have been shown to specifically target mRNA. Thus, although some, like
METTL3, may indeed show specificity in the brain, it is also possible that additional components
confer this specificity during memory formation. Fourth, although only two m6A binding
proteins were investigated, namely ELAVL1 and YTHDF2, if m6A is indeed involved in
regulating complex mechanisms like local mRNA translation, it is very likely that many
additional binding proteins are involved in the pathway, and dynamically regulated during
memory formation. Thus, although this study provides a first overview of how the m6A pathway
is regulated during memory formation, many pieces of the puzzle may still be missing, some of
which will likely provide key insight into the role of the m6A pathway in memory. Further
research identifying new m6A pathway components and investigating how they are regulated
during memory is required to help fill in the pieces of the puzzle.
4.7.2 m6A pathway regulation may not be occurring primarily in principal
neurons
Due to our dissection technique, in which the entire dorsal CA1 is dissected and processed, the
pool of RNA in the samples used for Fig. 3.2, 3.3 and S1 comes not only from principal neurons,
but also from glial cells and interneurons. As a result, the changes observed in Fig. 3.2 and 3.3
may not reflect regulation of the m6A pathway in neurons, as is assumed in the proposed model
linking m6A to local mRNA translation in neurons (Fig. 4.1). To specifically verify in which
cells the m6A pathway is being regulated, future experiments could be run on RNA isolated from
manually sorted fluorescently labeled pyramidal or glial cells, and could compare regulation of
the m6A pathway in the different cell types (Hempel et al., 2007).
65
Regardless, the HSVs designed in this project to target Fto expression levels will likely shed
some light on this question. Indeed, as HSVs are neurotropic, infecting only non-dividing,
mature neurons (Fink et al., 1996; Neve et al., 2005), any effects seen on memory following
manipulation of Fto using these viruses will reflect changes in FTO and m6A occurring in
neurons, and thus reflect the role of the m6A pathway in neurons in regulating memory.
4.7.3 m6A regulation in other brain regions during learning
The dorsal CA1 region of the hippocampus was selected as a starting point for investigating the
possible role of m6A in memory in the brain, due to its critical role in spatial memory and
position as the main output of the hippocampus (Amaral & Witter, 1989; Auer et al., 1989;
Lenck-Santini et al., 2001; Moser et al., 1993; Volpe et al., 1992). If m6A is indeed involved in
memory, it is likely that the m6A pathway is regulated in different memory tasks, as well as in
several other regions involved in memory, like the DG and prefrontal cortex, both of which play
critical roles in memory and undergo synaptic plasticity during learning (Laroche et al., 2000;
O’Malley et al., 2000; Saxe et al., 2006; Steward & Worley, 2001). In order to develop a
stronger understanding of the role of m6A in memory and establish whether it is involved in
basic memory-related functions, like regulating local mRNA translation, which are critical to all
cells undergoing plasticity, it will be helpful to explore its role in different tasks and regions
involved in learning and memory.
4.7.4 Measuring mRNA levels versus protein levels
In order to evaluate the variations in the levels of m6A pathway components, we chose to
measure mRNA levels instead of protein levels. Indeed, compared to other techniques like
Western blotting, qPCR enables higher throughput, and more precise and sensitive
measurements of enrichment levels (Aebersold et al., 2013; Svec et al., 2015). However, given
that the m6A pathway components investigated in this study are of interest here for their
functions in protein form, it is important to ensure that the changes in mRNA levels observed are
in fact reflected at the protein level. To do this, one could use Western blot analyses or ribosome
profiling, and compare the findings to the qPCR results (Cho et al., 2015). However, a study
looking at over 1000 genes by Cho et al. (2015) showed that during memory formation, changes
in protein levels in the brain are primarily driven by increases in protein synthesis in the first 5 to
10 minutes, then by changes in gene transcription from 30 min to 4h. In other words, after 30
66
min, changes in protein levels are well reflected by changes in mRNA transcript levels. As all of
our time points occurred within this second window, the use of qPCR to measure gene levels was
deemed appropriate to measure changes in m6A pathway components following CFC, and likely
accurately reflects protein levels. Future studies could nonetheless confirm these findings by
measuring protein levels as well.
As for the virus validation component of this project, in order to ensure that the changes induced
in Fto transcript levels were having a functional effect in the cells, we measured the effects on
m6A levels. As m6A levels should only be affected if the FTO protein is being properly
overexpressed or knocked down, and all three viruses showed the expected changes in m6A
levels (Jia et al., 2011), we concluded that the viruses were manipulating FTO levels as intended.
4.7.5 Limitations related to the use of HSVs
4.7.5.1 Toxicity of HSVs in cell culture
As mentioned above, there was a substantial drop in the level of Fto overexpression from 12h to
2 days for p1005-FTO (Fig. 3.4A-B), as well as in the efficiency of the Fto knockdown from 2 to
3 days for shRNA-Fto (Fig. 3.4G-H). Given that increased levels of cell death were also seen at
these time points, for both control and Fto manipulating viruses, it is likely that this is due to the
cytotoxicity of the HSV helper virus used. Helper virus is used to package transgene vectors into
viral particles, as it contains the necessary HSV genes, some of which induce cell lysis. It is then
diluted to a minimal level, but not completely eliminated from the final viral stock, and as such
can be toxic and affect the health and metabolism of cells in vitro (Fink et al., 1996; Lim, 2013;
Neve et al., 2005). Importantly, when infecting non-dividing cells such as neurons, many HSV-
1s, including the type used in this project, enter a latent state in which lytic genes are not
expressed (Fink et al., 1996; Lim, 2013). As a result, cell death is much less of a concern in vivo,
and viruses of this type are widely and successfully used for transgene expression in a variety of
brain regions, including the hippocampus (Cole et al., 2012; Han et al., 2009; Josselyn et al.,
2001; Sekeres et al., 2010, 2012). Future validation of the viruses in vivo will thus be important
to confirm the effects seen on Fto and m6A in vitro, establish the size of these effects in the
brain, and verify that the increased cell death and drop in efficiency are indeed due to helper
virus, as neither effect should be seen in neurons.
67
4.7.5.2 Limited infection rate using HSVs
HSVs offer several advantages for our investigation of the role of the m6A pathway in memory
over other viruses used for gene delivery, like AAVs. As mentioned above, they are neurotropic,
enable rapid and short-term expression of transgenes (Carlezon et al., 1998), and allow
sufficiently large packaging sizes for large proteins like Cas9 (Neve et al., 2005; Senís et al.,
2014). However, although this varies according to the region targeted and vector used, HSVs
typically infect only 6-20% of cells in a brain region (Sekeres et al., 2012; Yiu et al., 2014).
Depending on the memory-related mechanism targeted, manipulating gene expression in such a
small proportion of cells may be sufficient to produce an effect on memory (Han et al., 2009;
Hsiang et al., 2014). Indeed, if manipulating Fto levels in approximately 20% of CA1 principal
neurons impacts memory formation, this will strongly support a critical role for FTO and m6A in
memory. However, if no effect is seen, it may be necessary to verify that this is due to FTO and
not the size of the infection by confirming the results using viruses with higher infection rates,
such as AAVs (Swiech et al., 2015) which can be engineered to specifically target neurons
through different promoters or serotypes (Shetsova et al., 2005; van den Pol et al., 2009).
4.7.6 Effects on m6A levels of targeting Fto are not restricted to mRNA
Lastly, as mentioned previously, none of the demethylases or components of the
methyltransferase complex have been shown to specifically target m6A in mRNA. Furthermore,
as Fig. 3.4 shows, overexpressing or knocking down Fto in vitro leads to observable changes in
m6A levels in total RNA, confirming that FTO also demethylates non coding RNA (Fu et al.,
2014a). Thus, although these HSVs can be used to establish a causal link between FTO and
memory, the findings will not provide an answer as to whether regulation of m6A in mRNA
specifically is critical for learning and memory, as suggested in the proposed model linking m6A
to local mRNA translation (Fig. 4.1).
In order to address this question more specifically, and establish whether m6A is indeed involved
in regulating mRNA processing in memory, and possibly local translation, it will be necessary to
identify which components of the m6A pathway enable specific targeting of mRNA. Indeed, it is
clear from Fig. 3.4 that it is possible to specifically increase m6A levels in mRNA. Once the
components involved in regulating this effect during memory formation are identified, new
HSVs can be designed to target them with relatively little difficulty, by adding or replacing
68
sgRNAs in the Cas9-Fto vector (Walters et al., 2015). These viruses can then be used to
manipulate m6A levels specifically in mRNA during memory formation, and begin to answer
questions specifically about the potential role of m6A in regulating mRNA processing in
memory.
4.8 Future Aims
Overall, the results presented here show that the m6A pathway is extensively regulated during
memory formation. Furthermore, they show that both the m6A demethylase Fto and m6A levels
in mRNA display a biphasic regulation similar to many other critical memory-related processes
like IEG activation, gene transcription, protein synthesis and LTP. Our results also indicate that
mRNA is particularly targeted by the m6A pathway during memory formation, supporting a role
for m6A in regulating components of memory-related mRNA processing, like transport and
translation. Lastly, they demonstrate the flexibility of the m6A pathway, supporting its ability to
regulate complex processes like local mRNA translation. Altogether, these findings strongly
support a role for m6A in memory, and further suggest a possible role in regulating local mRNA
translation. Nonetheless, as this study is the first to explore the m6A pathway in the brain during
memory, more experiments must be conducted before the link between m6A and memory can be
established.
The first of these proposed experiments uses the HSVs designed in this project to test whether
manipulating Fto levels in CA1 during learning affects memory. If this is the case, it would
support a critical role for m6A in memory formation. The second experiment aims to further
explore the result showing an increase in m6A in mRNA during memory formation, in order to
identify the mRNA transcripts in which m6A levels are increased. If these transcripts are
primarily PRPs, it would strongly support a role for m6A in regulating memory-related mRNA
processing. The last experiment aims to establish whether the m6A pathway is in fact engaged at
the synapse during memory formation. If this is the case, it would corroborate the potential role
of m6A in regulating local mRNA translation during memory formation. Thus, each experiment
will help fill in a piece of the puzzle linking m6A to memory.
69
4.8.1 Test whether manipulating FTO and m6A levels impacts memory
One of the principal aims of this project was to develop HSVs to target the m6A pathway during
learning. Using these tools, we will be able to establish whether, as we hypothesize, m6A
regulation is critical in memory formation. Indeed, although our results show that the m6A
pathway is regulated during learning, it is possible that these changes are not critical to memory,
or reflect basic cellular mechanisms independent of memory.
The first step in testing our hypothesis will be, as mentioned previously, to validate the HSVs
designed in this project and establish the size of their effects in vivo. To do this, we will infuse
them into dorsal CA1, collect the tissue three to five days later (Sekeres et al., 2010) and
measure Fto and m6A levels in vivo. Results should be qualitatively similar to those obtained in
culture (Fig. 3.4). Once the HSVs are fully validated, the next step will be to verify how
manipulating FTO levels affects memory formation. Since our results showed that Fto levels are
downregulated following CFC (Fig. 3.2), and m6A levels are upregulated (Fig. 3.3), we
hypothesize that (1) knocking down Fto levels will enhance the naturally occurring regulation,
producing a stronger memory, whereas (2) overexpressing Fto will act counter to natural
regulation of m6A, and thus impair memory.
To test these hypotheses, we will first perform stereotaxic infusions of the HSVs into area CA1
of the hippocampus of adult mice (Sekeres et al., 2010). Following recovery, mice infused with a
knockdown virus (1) will be trained with a weak CFC protocol, receiving for example one
0.5 mA shock over the course of a 5-min exposure to a new context. Mice will then be returned
to the context 24h later and their levels of freezing will be recorded as measures of the strength
of their memory (Chen et al., 1996; Maren et al., 1997). Based on our hypothesis, mice infused
with knockdown viruses should freeze more than mice infused with the control virus, showing an
enhancement of a weak memory. Mice infused with the overexpression virus (2) will be trained
in a strong CFC protocol, receiving for example three 0.5 mA shocks over the course of a 5-min
exposure to a new context. Based on our hypothesis, upon testing 24h later, mice infused with
the overexpression virus should freeze less than mice infused with the control virus, showing an
impairment of a strong memory.
If the results of these experiments do not show an effect of manipulating Fto on memory, it will
suggest that although m6A is regulated during memory formation, it is not critical. If however
70
the results of these experiments are as hypothesized, they would provide the first causal evidence
demonstrating a role for m6A regulation in memory. Given the flexibility of the Cas9 system, the
viruses targeting Fto could quickly be redesigned to target other pathway components and start
to piece together their distinct roles in regulating m6A in memory.
4.8.2 Identify mRNA transcripts showing enriched m6A levels during memory
formation
As shown in Fig. 3.3, there is an upregulation of m6A in mRNA specifically following CFC.
During heat shock, both Dominissini et al. (2012) and Zhou et al. (2015) observed an increase in
m6A sites and translation of transcripts encoding stress-related proteins. Similarly, we
hypothesize that if the upregulation of m6A is indeed related to memory, it will primarily affect
memory-related transcripts.
To test this hypothesis, we will collect RNA from similar cohorts to those in Fig. 3.2, namely
naïve mice, and mice sacrificed 30 min to 4h after CFC. We will then run a methylated RNA
immunoprecipitation assay on the samples, allowing only transcripts containing m6A to be
retained. These samples will then be sequenced to identify which transcripts show an increase in
m6A 30 min and 4h following CFC (Dominissini et al., 2012). Assuming that the transcripts
showing higher m6A levels are also preferentially translated, we would expect to see many
locally translated PRP transcripts on the list, like CamKIIA, (Mayford et al., 1996; Paradies &
Steward, 1997), Arc (Link et al., 1995), Actb (Tiruchinapalli et al., 2003), Map2 (Garner et al.,
1988), and NMDA receptor subunits (Grooms et al., 2006; Miyashiro et al., 1994).
However, the m6A system is very flexible, which is why we proposed that it could be involved
in the fine-tuned processing of mRNA. As a result, the location of m6A in a transcript affects
which binding proteins recognize it, and thus whether it is translated or degraded, for example,
(Durie et al., 2010; Wang, X. et al., 2014; Wang, Y. et al., 2014; Zhou et al., 2015). Thus, it is
likely that not all transcripts showing increased m6A levels will show an increase in translation.
Ribosome profiling could thus be used to identify those transcripts recruited to the translation
machinery following methylation (Cho et al., 2015) and establish whether they are primarily
PRPs.
71
Altogether, the results of this experiment will provide key insight as to what transcripts are being
targeted by the m6A pathway during memory formation for processing. If the transcripts are not
PRPs, it will suggest that m6A may be regulating more basic cellular functions than memory.
However, if they are primarily PRPs, it would strengthen the hypothesis that m6A regulates
mRNA processing of PRP transcripts during memory formation.
4.8.3 Establish whether m6A is playing a role in memory at the synapse
Local mRNA translation is known to be critical to memory formation (Miller et al., 2002;
Ouyang et al., 1999), and yet the underlying mechanisms governing it remain unknown. The
flexibility and dynamic nature of the m6A pathway makes it an excellent candidate for regulating
complex processes like mRNA translation, as proposed in Fig. 4.1. It is for this reason that its
potential role in memory is of particular interest, and thus this final experiment proposes to
establish whether m6A is indeed playing a role in memory at the synapse.
In order to do this, we will look at where m6A pathway components are expressed and regulated
in neurons. If the pathway is important for local mRNA translation, some of its components must
be active at the synapse where local translation occurs. In our model (Fig. 4.1), we propose that
m6A binding proteins, like Elavl1, and the demethylase Fto are active and regulated at the
synapse during memory formation. In order to verify whether this is the case, we will first use
antibodies against ELAVL1, FTO and m6A to establish the subcellular distribution of these
pathway components in hippocampal cultured neurons.
If there is basal expression of these components in dendrites and spines, we will collect the
hippocampi of mice having undergone CFC protocol as in Fig. 3.2. Samples will be separated
into nuclear, cytoplasmic and synaptosomal fractions (Boyl et al., 2007), and processed for a
Western blot analysis of protein abundance. Using appropriate internal controls for each fraction,
we will first compare the level of components, like ELAVL1, FTO and m6A, in the nucleus,
cytoplasm and synaptosomes of naïve mice, then investigate whether any changes in distribution
occur 30 min after CFC. If ELAVL1, FTO or m6A is regulated at the synapse specifically during
memory formation, there should be a relative change in the abundance of the components in the
synaptosomal fraction compared to the cytoplasmic or nuclear fraction 30 min after CFC
compared to naïve controls.
72
If none of the components of the m6A pathway are regulated at the synapse, it will suggest that
m6A is primarily involved in regulating processes like mRNA transport and possibly somatic
translation during memory formation (Dominissini et al., 2012; Fustin et al., 2013; Zhao et al.,
2014). If however some of the components are regulated locally, this would strongly support a
role for m6A in regulating local mRNA translation.
73
Chapter 5
Conclusion
Memory relies critically on the synthesis of plasticity-related proteins, not only in the soma, but
also locally at synapses. How cells are able to precisely and independently coordinate memory-
related translation at each of their synapses remains unknown. Recently, m6A, the most abundant
modification of RNA, was shown to be dynamically regulated and involved in translational
control. Similar to epigenetic modifications of DNA which enable precise control over
transcription in the nucleus during memory formation, mRNA modifications may provide fine-
tuned control over translation throughout the neuron, including locally at individual synapses.
In this project we investigated the role of the m6A RNA methylation pathway in memory.
Firstly, we found that components of the m6A pathway are enriched in different tissues, with the
demethylase Fto being highly expressed in the mouse brain. We then showed that during
formation of a contextual memory, levels of several m6A pathway components, including Fto,
are transiently modulated in the dorsal CA1 of the hippocampus. Correspondingly, there is a
biphasic increase in overall levels of m6A in mRNA. To probe the role of m6A in memory
formation, we then developed and validated HSV vectors to rapidly and selectively overexpress
or knockdown FTO in cells. Among these is the first HSV-mediated CRISPR/Cas9 knockout.
Harnessing this tool will allow increased flexibility and rapid integration of new gene targets to
more thoroughly investigate the role of the m6A pathway in memory.
As this project is one of the first to look at the possible role of m6A in memory, the results
presented here remain exploratory, but promising. Indeed, we show not only that the m6A
pathway is remarkably flexible, but also that it is dynamically regulated during memory
formation, in particular in mRNA. Together, these findings provide the first compelling evidence
showing that m6A may be critically involved in regulating highly complex processes like local
mRNA translation in memory. FTO has already been linked to dementia and cognitive
impairments. Gaining deeper insight into the role of m6A in memory may not only improve our
74
ability to treat diseases involving memory dysfunction, but also strengthen our grasp of the
intricate biological processes that shape who we are and what we do.
75
References
Aebersold, R., Burlingame, A.L. & Bradshaw, R.A. Western blots versus selected reaction
monitoring assays: time to turn the tables? Mol Cell Proteomics 12, 2381-2 (2013).
Ainsley, J.A., Drane, L., Jacobs, J., Kittelberger, K.A. & Reijmers, L.G. Functionally diverse
dendritic mRNAs rapidly associate with ribosomes following a novel experience. Nat
Commun 5, 4510 (2014).
Alberts, B., Johnson, A., Lewis, J., Morgan, D., Raff, M., Roberts, K. & Walter, P. Molecular
Biology of the Cell (Garland Science, New York, 2015).
Amaral, D.G. & Witter, M.P. The three-dimensional organization of the hippocampal
formation: a review of anatomical data. Neuroscience 31, 571-91 (1989).
An, J.J., Gharami, K., Liao, G.Y., Woo, N.H., Lau, A.G., Vanevski, F., ... Xu, B. Distinct
role of long 3' UTR BDNF mRNA in spine morphology and synaptic plasticity in
hippocampal neurons. Cell 134, 175-87 (2008).
Andreassi, C. & Riccio, A. To localize or not to localize: mRNA fate is in 3′ UTR ends.
Trends Cell Biol 19, 465-74 (2009).
Ashraf, S.I., McLoon, A.L., Sclarsic, S.M. & Kunes, S. Synaptic protein synthesis associated
with memory is regulated by the RISC pathway in Drosophila. Cell 124, 191-205 (2006).
Auer, R.N., Jensen, M.L. & Whishaw, I.Q. Neurobehavioral deficit due to ischemic brain
damage limited to half of the CA1 sector of the hippocampus. J Neurosci 9, 1641-1647
(1989).
Barco, A., Patterson, S.L., Alarcon, J.M., Gromova, P., Mata-Roig, M., Morozov, A. &
Kandel, E.R. Gene expression profiling of facilitated L-LTP in VP16-CREB mice reveals
that BDNF is critical for the maintenance of LTP and its synaptic capture. Neuron 48, 123-37
(2005).
Battaglia, F.P., Benchenane, K., Sirota, A., Pennartz, C.M. & Wiener, S.I. The hippocampus:
hub of brain network communication for memory. Trends Cogn Sci 15, 310-8 (2011).
Ben Achour, S. & Pascual, O. Glia: the many ways to modulate synaptic plasticity.
Neurochem Int 57, 440-5 (2010).
Berger-Sweeney, J., Zearfoss, N.R. & Richter, J.D. Reduced extinction of hippocampal-
dependent memories in CPEB knockout mice. Learn Mem 13, 4-7 (2006).
Bliss, T.V. & Collingridge, G.L. A synaptic model of memory: long-term potentiation in the
hippocampus. Nature 361, 31-9 (1993).
76
Bliss, T.V. & Lømo, T. Long-lasting potentiation of synaptic transmission in the dentate area
of the anaesthetized rabbit following stimulation of the perforant path. J Physiol 232, 331-56
(1973).
Bokar, J.A., Rath-Shambaugh, M.E., Ludwiczak, R., Narayan, P. & Rottman, F.
Characterization and partial purification of mRNA N6-adenosine methyltransferase from
HeLa cell nuclei. Internal mRNA methylation requires a multisubunit complex. J Biol Chem
269, 17697-704 (1994).
Bokar, J.A., Shambaugh, M.E., Polayes, D., Matera, A.G. & Rottman, F.M. Purification and
cDNA cloning of the AdoMet-binding subunit of the human mRNA (N6-adenosine)-
methyltransferase. RNA 3, 1233-47 (1997).
Bolotin, A., Quinquis, B., Sorokin, A. & Ehrlich, S.D. Clustered regularly interspaced short
palindrome repeats (CRISPRs) have spacers of extrachromosomal origin. Microbiology 151,
2551-61 (2005).
Bourne, J.N., Sorra, K.E., Hurlburt, J. & Harris, K.M. Polyribosomes are increased in spines
of CA1 dendrites 2 h after the induction of LTP in mature rat hippocampal slices.
Hippocampus 17, 1-4 (2007).
Bourtchuladze, R., Abel, T., Berman, N., Gordon, R., Lapidus, K. & Kandel, E.R. Different
training procedures recruit either one or two critical periods for contextual memory
consolidation, each of which requires protein synthesis and PKA. Learn Mem 5, 365-74
(1998).
Bourtchuladze, R., Frenguelli, B., Blendy, J., Cioffi, D., Schutz, G. & Silva, A.J. Deficient
long-term memory in mice with a targeted mutation of the cAMP-responsive element-
binding protein. Cell 79, 59-68 (1994).
Boyl, P.P., Di Nardo, A., Mulle, C., Sassoè‐Pognetto, M., Panzanelli, P., Mele, A., ... Vara,
H. Profilin2 contributes to synaptic vesicle exocytosis, neuronal excitability, and novelty‐seeking behavior. EMBO J 26, 2991-3002 (2007).
Bozon, B., Kelly, A., Josselyn, S.A., Silva, A.J., Davis, S. & Laroche, S. MAPK, CREB and
zif268 are all required for the consolidation of recognition memory. Philos Trans R Soc Lond
B Biol Sci 358, 805-14 (2003).
Bradshaw, K.D., Emptage, N.J. & Bliss, T.V.P. A role for local protein synthesis is required
for hippocampal late LTP. Eur J Neurosci 11, 3150-2 (2003).
Bramham, C.R. & Wells, D.G. Dendritic mRNA: transport, translation and function. Nat Rev
Neurosci 8, 776-89 (2007).
Braun, K. & Segal, M. FMRP involvement in formation of synapses among cultured
hippocampal neurons. Cerebral Cortex 10, 1045-52 (2000).
Brennan, C.M. & Steitz, J.A. HuR and mRNA stability. Cell Mol Life Sci 58, 266-77 (2001).
77
Bressler, J., Fornage, M., Demerath, E.W., Knopman, D.S., Monda, K.L., North, K.E., ...
Boerwinkle, E. Fat mass and obesity gene and cognitive decline: the Atherosclerosis Risk in
Communities Study. Neurology 80, 92-9 (2013).
Brown, V., Jin, P., Ceman, S., Darnell, J.C., O'Donnell, W.T., Tenenbaum, S.A., ... Warren,
S.T. Microarray identification of FMRP-associated brain mRNAs and altered mRNA
translational profiles in fragile X syndrome. Cell 107, 477-87 (2001).
Cahoy, J.D., Emery, B., Kaushal, A., Foo, L.C., Zamanian, J.L., Christopherson, K.S., ...
Barres, B.A. A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new
resource for understanding brain development and function. J Neurosci 28, 264-78 (2008).
Carlezon, W.A., Jr., Thome, J., Olson, V.G., Lane-Ladd, S.B., Brodkin, E.S., Hiroi, N., ...
Nestler, E.J. Regulation of cocaine reward by CREB. Science 282, 2272-5 (1998).
Carr, M.F., Jadhav, S.P. & Frank, L.M. Hippocampal replay in the awake state: a potential
substrate for memory consolidation and retrieval. Nat Neurosci 14, 147-53 (2011).
Castellucci, V.F., Frost, W.N., Goelet, P.H., Montarolo, P.G., Schacher, S., Morgan, J.A., ...
Kandel, E.R. Cell and molecular analysis of long-term sensitization in Aplysia. J Physiol
(Paris) 81, 349-57 (1986).
Castellucci, V.F., Pinsker, H., Kupfermann, I. & Kandel, E.R. Neuronal mechanisms of
habituation and dishabituation of the gill-withdrawal reflex in Aplysia. Science 167, 1745-8
(1970).
Castrén, E., Pitkanen, M., Sirvio, J., Parsadanian, A., Lindholm, D., Thoenen, H. &
Riekkinen, P.J. The induction of LTP increases BDNF and NGF mRNA but decreases NT-3
mRNA in the dentate gyrus. Neuroreport 4, 895-8 (1993).
Chen, C., Kim, J.J., Thompson, R.F. & Tonegawa, S. Hippocampal lesions impair contextual
fear conditioning in two strains of mice. Behav Neurosci 110, 1177-80 (1996).
Chen, W., Zhang, L., Zheng, G., Fu, Y., Ji, Q., Liu, F., ... He, C. Crystal structure of the RNA
demethylase ALKBH5 from zebrafish. FEBS Lett 588, 892-8 (2014).
Chiang, P.K., Gordon, R.K., Tal, J., Zeng, G.C., Doctor, B.P., Pardhasaradhi, K. & McCann,
P.P. S-Adenosylmethionine and methylation. FASEB J 10, 471-80 (1996).
Cho, J., Yu, N.K., Choi, J.H., Sim, S.E., Kang, S.J., Kwak, C., ... Kaang, B.K. Multiple
repressive mechanisms in the hippocampus during memory formation. Science 350, 82-7
(2015).
Church, C., Moir, L., McMurray, F., Girard, C., Banks, G.T., Teboul, L., ... Cox, R.D.
Overexpression of Fto leads to increased food intake and results in obesity. Nat Genet 42,
1086-92 (2010).
Clark, M.S., Sexton, T.J., McClain, M., Root, D., Kohen, R. & Neumaier, J.F.
Overexpression of 5-HT1B receptor in dorsal raphe nucleus using Herpes Simplex Virus
gene transfer increases anxiety behavior after inescapable stress. J Neurosci 22, 4550-62
(2002).
78
Cole, C.J., Mercaldo, V., Restivo, L., Yiu, A.P., Sekeres, M.J., Han, J.H., ... Josselyn, S.A.
MEF2 negatively regulates learning-induced structural plasticity and memory formation. Nat
Neurosci 15, 1255-64 (2012).
Coller, J. & Parker, R. Eukaryotic mRNA decapping. Annu Rev Biochem 73, 861-90 (2004).
Cory, S., Genin, C. & Adams, J.M. Modified nucleosides and 5'-end groups in purified
mouse immunoglobulin light chain mRNA and rabbit globin mRNA detected by borohydride
labelling. Biochim Biophys Acta 454, 248-62 (1976).
Czerwoniec, A., Dunin-Horkawicz, S., Purta, E., Kaminska, K.H., Kasprzak, J.M., Bujnicki,
J.M., ... Rother, K. MODOMICS: a database of RNA modification pathways. 2008 update.
Nucleic Acids Res 37, D118-21 (2009).
Da Silva, W.C., Bonini, J.S., Bevilaqua, L.R., Medina, J.H., Izquierdo, I. & Cammarota, M.
Inhibition of mRNA synthesis in the hippocampus impairs consolidation and reconsolidation
of spatial memory. Hippocampus 18, 29-39 (2008).
Deng, W., Mayford, M. & Gage, F.H. Selection of distinct populations of dentate granule
cells in response to inputs as a mechanism for pattern separation in mice. Elife 2, e00312
(2013).
Desrosiers, R., Friderici, K. & Rottman, F. Identification of methylated nucleosides in
messenger RNA from Novikoff hepatoma cells. Proc Natl Acad Sci U S A 71, 3971-5 (1974).
Dina, C., Meyre, D., Gallina, S., Durand, E., Korner, A., Jacobson, P., ... Froguel, P.
Variation in FTO contributes to childhood obesity and severe adult obesity. Nat Genet 39,
724-6 (2007).
Dominissini, D., Moshitch-Moshkovitz, S., Schwartz, S., Salmon-Divon, M., Ungar, L.,
Osenberg, S., ... Rechavi, G. Topology of the human and mouse m6A RNA methylomes
revealed by m6A-seq. Nature 485, 201-6 (2012).
Dudai, Y. The restless engram: consolidations never end. Annu Rev Neurosci 35, 227-47
(2012).
Dudek, S.M. & Bear, M.F. Homosynaptic long-term depression in area CA1 of hippocampus
and effects of N-methyl-D-aspartate receptor blockade. Proc Natl Acad Sci U S A 89, 4363-7
(1992).
Dunn, S. & Cowling, V.H. Myc and mRNA capping. Biochim Biophys Acta 1849, 501-5
(2015).
Dunnett, C.W. A multiple comparison procedure for comparing several treatments with a
control. J Amer Statist Assoc 50, 1096-121 (1955).
Durie, D., Lewis, S.M., Liwak, U., Kisilewicz, M., Gorospe, M. & Holcik, M. RNA-binding
protein HuR mediates cytoprotection through stimulation of XIAP translation. Oncogene 30,
1460-9 (2011).
Dynes, J.L. & Steward, O. Dynamics of bidirectional transport of Arc mRNA in neuronal
dendrites. J Comp Neurol 500, 433-47 (2007).
79
Fabian, M.R., Sonenberg, N. & Filipowicz, W. Regulation of mRNA translation and stability
by microRNAs. Annu Rev Biochem 79, 351-79 (2010).
Fanselow, M.S. Contextual fear, gestalt memories, and the hippocampus. Behav Brain Res
110, 73-81 (2000).
Farina, K.L., Huttelmaier, S., Musunuru, K., Darnell, R. & Singer, R.H. Two ZBP1 KH
domains facilitate beta-actin mRNA localization, granule formation, and cytoskeletal
attachment. J Cell Biol 160, 77-87 (2003).
Fink, D.J., DeLuca, N.A., Goins, W.F. & Glorioso, J.C. Gene transfer to neurons using
herpes simplex virus-based vectors. Annu Rev Neurosci 19, 265-87 (1996).
Flavell, S.W., Kim, T.K., Gray, J.M., Harmin, D.A., Hemberg, M., Hong, E.J., ... Greenberg,
M.E. Genome-wide analysis of MEF2 transcriptional program reveals synaptic target genes
and neuronal activity-dependent polyadenylation site selection. Neuron 60, 1022-38 (2008).
Frankland, P.W. & Bontempi, B. The organization of recent and remote memories. Nat Rev
Neurosci 6, 119-30 (2005).
Frayling, T.M., Timpson, N.J., Weedon, M.N., Zeggini, E., Freathy, R.M., Lindgren, C.M.,
... Rayner, N.W. A common variant in the FTO gene is associated with body mass index and
predisposes to childhood and adult obesity. Science 316, 889-94 (2007).
Fredriksson, R., Hagglund, M., Olszewski, P.K., Stephansson, O., Jacobsson, J.A.,
Olszewska, A.M., ... Schioth, H.B. The obesity gene, FTO, is of ancient origin, up-regulated
during food deprivation and expressed in neurons of feeding-related nuclei of the brain.
Endocrinology 149, 2062-71 (2008).
Frey, U., Frey, S., Schollmeier, F. & Krug, M. Influence of actinomycin D, a RNA synthesis
inhibitor, on long-term potentiation in rat hippocampal neurons in vivo and in vitro. J Physiol
490 ( Pt 3), 703-11 (1996).
Frey, U., Krug, M., Reymann, K.G. & Matthies, H. Anisomycin, an inhibitor of protein
synthesis, blocks late phases of LTP phenomena in the hippocampal CA1 region in vitro.
Brain Res 452, 57-65 (1988).
Frey, U. & Morris, R.G. Synaptic tagging and long-term potentiation. Nature 385, 533-6
(1997).
Fu, Y., Dominissini, D., Rechavi, G. & He, C. Gene expression regulation mediated through
reversible m6A RNA methylation. Nat Rev Genet 15, 293-306 (2014).
Fu, Y., Jia, G., Pang, X., Wang, R.N., Wang, X., Li, C.J., ... He, C. FTO-mediated formation
of N6-hydroxymethyladenosine and N6-formyladenosine in mammalian RNA. Nat Commun
4, 1798 (2013).
Fu, Y., Sander, J.D., Reyon, D., Cascio, V.M. & Joung, J.K. Improving CRISPR-Cas
nuclease specificity using truncated guide RNAs. Nat Biotechnol 32, 279-84 (2014).
80
Fustin, J.M., Doi, M., Yamaguchi, Y., Hida, H., Nishimura, S., Yoshida, M., ... Okamura, H.
RNA-methylation-dependent RNA processing controls the speed of the circadian clock. Cell
155, 793-806 (2013).
Garner, C.C., Tucker, R.P. & Matus, A. Selective localization of messenger RNA for
cytoskeletal protein MAP2 in dendrites. Nature 336, 674-7 (1988).
Gerken, T., Girard, C.A., Tung, Y.C., Webby, C.J., Saudek, V., Hewitson, K.S., ... Schofield,
C.J. The obesity-associated FTO gene encodes a 2-oxoglutarate-dependent nucleic acid
demethylase. Science 318, 1469-72 (2007).
Geula, S., Moshitch-Moshkovitz, S., Dominissini, D., Mansour, A.A., Kol, N., Salmon-
Divon, M., ... Hanna, J.H. Stem cells. m6A mRNA methylation facilitates resolution of naive
pluripotency toward differentiation. Science 347, 1002-6 (2015).
Gilbert, L.A., Horlbeck, M.A., Adamson, B., Villalta, J.E., Chen, Y., Whitehead, E.H., ...
Weissman, J.S. Genome-Scale CRISPR-Mediated Control of Gene Repression and
Activation. Cell 159, 647-61 (2014).
Girardeau, G., Benchenane, K., Wiener, S.I., Buzsaki, G. & Zugaro, M.B. Selective
suppression of hippocampal ripples impairs spatial memory. Nat Neurosci 12, 1222-3 (2009).
Grooms, S.Y., Noh, K.M., Regis, R., Bassell, G.J., Bryan, M.K., Carroll, R.C. & Zukin, R.S.
Activity bidirectionally regulates AMPA receptor mRNA abundance in dendrites of
hippocampal neurons. J Neurosci 26, 8339-51 (2006).
Guzowski, J.F., Lyford, G.L., Stevenson, G.D., Houston, F.P., McGaugh, J.L., Worley, P.F.
& Barnes, C.A. Inhibition of activity-dependent arc protein expression in the rat
hippocampus impairs the maintenance of long-term potentiation and the consolidation of
long-term memory. J Neurosci 20, 3993-4001 (2000).
Hall, J., Thomas, K.L. & Everitt, B.J. Rapid and selective induction of BDNF expression in
the hippocampus during contextual learning. Nat Neurosci 3, 533-5 (2000).
Han, J.H., Kushner, S.A., Yiu, A.P., Hsiang, H.L., Buch, T., Waisman, A., ... Josselyn, S.A.
Selective erasure of a fear memory. Science 323, 1492-6 (2009).
Håvik, B., Røkke, H., Bårdsen, K., Davanger, S. & Bramham, C.R. Bursts of high-frequency
stimulation trigger rapid delivery of pre-existing alpha-CaMKII mRNA to synapses: a
mechanism in dendritic protein synthesis during long-term potentiation in adult awake rats.
Eur J Neurosci 17, 2679-89 (2003).
Hebb, D.O. The organization of behavior; a neuropsychological theory (Wiley, New York,
1949).
Heidenreich, E., Novotny, R., Kneidinger, B., Holzmann, V. & Wintersberger, U. Non‐homologous end joining as an important mutagenic process in cell cycle‐arrested cells.
EMBO J 22, 2274-83 (2003).
Hempel, C.M., Sugino, K. & Nelson, S.B. A manual method for the purification of
fluorescently labeled neurons from the mammalian brain. Nat Protoc 2, 2924-9 (2007).
81
Hess, M.E., Hess, S., Meyer, K.D., Verhagen, L.A., Koch, L., Brönneke, H.S., ... Bruning,
J.C. The fat mass and obesity associated gene (Fto) regulates activity of the dopaminergic
midbrain circuitry. Nat Neurosci 16, 1042-8 (2013).
Hirsh, R. The hippocampus and contextual retrieval of information from memory: a theory.
Behav Biol 12, 421-44 (1974).
Horiuchi, K., Kawamura, T., Iwanari, H., Ohashi, R., Naito, M., Kodama, T. & Hamakubo,
T. Identification of Wilms' tumor 1-associating protein complex and its role in alternative
splicing and the cell cycle. J Biol Chem 288, 33292-302 (2013).
Horowitz, S., Horowitz, A., Nilsen, T.W., Munns, T.W. & Rottman, F.M. Mapping of N6-
methyladenosine residues in bovine prolactin mRNA. Proc Natl Acad Sci U S A 81, 5667-71
(1984).
Hsiang, H.-L.L., Epp, J.R., van den Oever, M.C., Yan, C., Rashid, A.J., Insel, N., ...
Frankland, P.W. Manipulating a “cocaine engram” in mice. J Neurosci 34, 14115-27 (2014).
Huang, Y.S., Carson, J.H., Barbarese, E. & Richter, J.D. Facilitation of dendritic mRNA
transport by CPEB. Genes Dev 17, 638-53 (2003).
Huang, Y.S., Jung, M.Y., Sarkissian, M. & Richter, J.D. N‐methyl‐d‐aspartate receptor
signaling results in Aurora kinase‐catalyzed CPEB phosphorylation and αCaMKII mRNA
polyadenylation at synapses. EMBO J 21, 2139-48 (2002).
Huber, K.M., Kayser, M.S. & Bear, M.F. Role for rapid dendritic protein synthesis in
hippocampal mGluR-dependent long-term depression. Science 288, 1254-7 (2000).
Hüttelmaier, S., Zenklusen, D., Lederer, M., Dictenberg, J., Lorenz, M., Meng, X., ... Singer,
R.H. Spatial regulation of beta-actin translation by Src-dependent phosphorylation of ZBP1.
Nature 438, 512-5 (2005).
Igaz, L.M., Vianna, M.R., Medina, J.H. & Izquierdo, I. Two time periods of hippocampal
mRNA synthesis are required for memory consolidation of fear-motivated learning. J
Neurosci 22, 6781-9 (2002).
Jadhav, S.P., Kemere, C., German, P.W. & Frank, L.M. Awake hippocampal sharp-wave
ripples support spatial memory. Science 336, 1454-8 (2012).
Jia, G., Fu, Y., Zhao, X., Dai, Q., Zheng, G., Yang, Y., ... Yang, Y.G. N6-methyladenosine in
nuclear RNA is a major substrate of the obesity-associated FTO. Nat Chem Biol 7, 885-7
(2011).
Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J.A. & Charpentier, E. A
programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science
337, 816-21 (2012).
Josselyn, S.A., Shi, C., Carlezon, W.A., Neve, R.L., Nestler, E.J. & Davis, M. Long-term
memory is facilitated by cAMP response element-binding protein overexpression in the
amygdala. J Neurosci 21, 2404-12 (2001).
82
Kandel, E.R. & Tauc, L. Heterosynaptic facilitation in neurones of the abdominal ganglion of
Aplysia depilans. J Physiol 181, 1-27 (1965).
Kelleher, R.J., Govindarajan, A., Jung, H.-Y., Kang, H. & Tonegawa, S. Translational
Control by MAPK Signaling in Long-Term Synaptic Plasticity and Memory. Cell 116, 467-
479 (2004).
Keller, L., Xu, W., Wang, H.X., Winblad, B., Fratiglioni, L. & Graff, C. The obesity related
gene, FTO, interacts with APOE, and is associated with Alzheimer's disease risk: a
prospective cohort study. J Alzheimers Dis 23, 461-9 (2011).
Kesner, R.P. Behavioral functions of the CA3 subregion of the hippocampus. Learn Mem 14,
771-81 (2007).
Kierzek, E. & Kierzek, R. The thermodynamic stability of RNA duplexes and hairpins
containing N6-alkyladenosines and 2-methylthio-N6-alkyladenosines. Nucleic Acids Res 31,
4472-80 (2003).
Kim, J.H. & Richter, J.D. Opposing polymerase-deadenylase activities regulate cytoplasmic
polyadenylation. Mol Cell 24, 173-83 (2006).
Kislauskis, E.H., Zhu, X. & Singer, R.H. Sequences responsible for intracellular localization
of beta-actin messenger RNA also affect cell phenotype. J Cell Biol 127, 441-51 (1994).
Knowles, R.B., Sabry, J.H., Martone, M.E., Deerinck, T.J., Ellisman, M.H., Bassell, G.J. &
Kosik, K.S. Translocation of RNA granules in living neurons. J Neurosci 16, 7812-20
(1996).
Kosik, K.S. The neuronal microRNA system. Nat Rev Neurosci 7, 911-20 (2006).
Lamond, A.I. & Spector, D.L. Nuclear speckles: a model for nuclear organelles. Nat Rev Mol
Cell Biol 4, 605-12 (2003).
Laroche, S., Davis, S. & Jay, T.M. Plasticity at hippocampal to prefrontal cortex synapses:
dual roles in working memory and consolidation. Hippocampus 10, 438-46 (2000).
Lasek, R.J. Protein transport in neurons. Int Rev Neurobiol 13, 289-324 (1970).
Lebedeva, S., Jens, M., Theil, K., Schwanhausser, B., Selbach, M., Landthaler, M. &
Rajewsky, N. Transcriptome-wide analysis of regulatory interactions of the RNA-binding
protein HuR. Mol Cell 43, 340-52 (2011).
Lenck-Santini, P.P., Save, E. & Poucet, B. Evidence for a relationship between place-cell
spatial firing and spatial memory performance. Hippocampus 11, 377-90 (2001).
Leutgeb, J.K., Leutgeb, S., Moser, M.B. & Moser, E.I. Pattern separation in the dentate gyrus
and CA3 of the hippocampus. Science 315, 961-6 (2007).
Levenson, J.M., Roth, T.L., Lubin, F.D., Miller, C.A., Huang, I.-C., Desai, P., ... Sweatt, J.D.
Evidence that DNA (cytosine-5) methyltransferase regulates synaptic plasticity in the
hippocampus. J Biol Chem 281, 15763-73 (2006).
83
Levenson, J.M., Weeber, E., Selcher, J.C., Kategaya, L.S., Sweatt, J.D. & Eskin, A. Long-
term potentiation and contextual fear conditioning increase neuronal glutamate uptake. Nat
Neurosci 5, 155-61 (2002).
Lim, F. HSV-1 as a model for emerging gene delivery vehicles. ISRN Virol 2013 (2013).
Lin, S., Choe, J., Du, P., Triboulet, R. & Gregory, R.I. The m6A methyltransferase METTL3
promotes translation in human cancer cells. Mol Cell.
Link, W., Konietzko, U., Kauselmann, G., Krug, M., Schwanke, B., Frey, U. & Kuhl, D.
Somatodendritic expression of an immediate early gene is regulated by synaptic activity.
Proc Natl Acad Sci U S A 92, 5734-8 (1995).
Lisman, J.E. Relating hippocampal circuitry to function: recall of memory sequences by
reciprocal dentate-CA3 interactions. Neuron 22, 233-42 (1999).
Liu, J., Yue, Y., Han, D., Wang, X., Fu, Y., Zhang, L., ... He, C. A METTL3-METTL14
complex mediates mammalian nuclear RNA N6-adenosine methylation. Nat Chem Biol 10,
93-5 (2014).
Lo, W.D., Qu, G., Sferra, T.J., Clark, R., Chen, R. & Johnson, P.R. Adeno-associated virus-
mediated gene transfer to the brain: duration and modulation of expression. Human gene
therapy 10, 201-213 (1999).
Lubin, F.D., Roth, T.L. & Sweatt, J.D. Epigenetic regulation of BDNF gene transcription in
the consolidation of fear memory. J Neurosci 28, 10576-86 (2008).
Luscher, C. & Malenka, R.C. NMDA receptor-dependent long-term potentiation and long-
term depression (LTP/LTD). Cold Spring Harb Perspect Biol 4, a005710 (2012).
Lynch, M.A. Long-term potentiation and memory. Physiol Rev 84, 87-136 (2004).
Malenka, R.C. & Bear, M.F. LTP and LTD: an embarrassment of riches. Neuron 44, 5-21
(2004).
Manahan-Vaughan, D., Kulla, A. & Frey, J.U. Requirement of translation but not
transcription for the maintenance of long-term depression in the CA1 region of freely moving
rats. J Neurosci 20, 8572-6 (2000).
Maren, S., Aharonov, G. & Fanselow, M.S. Neurotoxic lesions of the dorsal hippocampus
and Pavlovian fear conditioning in rats. Behav Brain Res 88, 261-74 (1997).
Martin, S.J. & Clark, R.E. The rodent hippocampus and spatial memory: from synapses to
systems. Cell Mol Life Sci 64, 401-31 (2007).
Martin, S.J., Grimwood, P.D. & Morris, R.G. Synaptic plasticity and memory: an evaluation
of the hypothesis. Annu Rev Neurosci 23, 649-711 (2000).
Mayford, M., Baranes, D., Podsypanina, K. & Kandel, E.R. The 3′-untranslated region of
CaMKIIα is a cis-acting signal for the localization and translation of mRNA in dendrites.
Proc Natl Acad Sci U S A 93, 13250-5 (1996).
84
Mayford, M., Siegelbaum, S.A. & Kandel, E.R. Synapses and memory storage. Cold Spring
Harb Perspect Biol 4, a005751 (2012).
McEchron, M.D. & Disterhoft, J.F. Hippocampal encoding of non-spatial trace conditioning.
Hippocampus 9, 385-96 (1999).
McEvoy, M., Cao, G., Llopis, P.M., Kundel, M., Jones, K., Hofler, C., ... Wells, D.G.
Cytoplasmic polyadenylation element binding protein 1-mediated mRNA translation in
Purkinje neurons is required for cerebellar long-term depression and motor coordination. J
Neurosci 27, 6400-11 (2007).
McTaggart, J.S., Lee, S., Iberl, M., Church, C., Cox, R.D. & Ashcroft, F.M. FTO is
expressed in neurones throughout the brain and its expression is unaltered by fasting. PLoS
One 6, e27968 (2011).
Megías, M., Emri, Z., Freund, T.F. & Gulyas, A.I. Total number and distribution of
inhibitory and excitatory synapses on hippocampal CA1 pyramidal cells. Neuroscience 102,
527-40 (2001).
Merkestein, M., McTaggart, J.S., Lee, S., Kramer, H.B., McMurray, F., Lafond, M., ...
Ashcroft, F.M. Changes in gene expression associated with FTO overexpression in mice.
PLoS One 9, e97162 (2014).
Meyer, K.D., Saletore, Y., Zumbo, P., Elemento, O., Mason, C.E. & Jaffrey, S.R.
Comprehensive analysis of mRNA methylation reveals enrichment in 3' UTRs and near stop
codons. Cell 149, 1635-46 (2012).
Mignone, F., Gissi, C., Liuni, S. & Pesole, G. Untranslated regions of mRNAs. Genome Biol
3, REVIEWS0004 (2002).
Miller, C.A., Gavin, C.F., White, J.A., Parrish, R.R., Honasoge, A., Yancey, C.R., ... Sweatt,
J.D. Cortical DNA methylation maintains remote memory. Nat Neurosci 13, 664-6 (2010).
Miller, C.A. & Sweatt, J.D. Covalent modification of DNA regulates memory formation.
Neuron 53, 857-69 (2007).
Miller, S., Yasuda, M., Coats, J.K., Jones, Y., Martone, M.E. & Mayford, M. Disruption of
dendritic translation of CaMKIIα impairs stabilization of synaptic plasticity and memory
consolidation. Neuron 36, 507-19 (2002).
Milner, B., Corkin, S. & Teuber, H.L. Further analysis of the hippocampal amnesic
syndrome: 14-year follow-up study of H.M. Neuropsychologia 6, 215-234 (1968).
Misane, I., Tovote, P., Meyer, M., Spiess, J., Ogren, S.O. & Stiedl, O. Time-dependent
involvement of the dorsal hippocampus in trace fear conditioning in mice. Hippocampus 15,
418-26 (2005).
Miyashiro, K., Dichter, M. & Eberwine, J. On the nature and differential distribution of
mRNAs in hippocampal neurites: implications for neuronal functioning. Proc Natl Acad Sci
U S A 91, 10800-4 (1994).
85
Miyashiro, K.Y., Beckel-Mitchener, A., Purk, T.P., Becker, K.G., Barret, T., Liu, L., ...
Eberwine, J. RNA cargoes associating with FMRP reveal deficits in cellular functioning in
Fmr1 null mice. Neuron 37, 417-31 (2003).
Mojica, F.J., Diez-Villasenor, C., García-Martínez, J. & Soria, E. Intervening sequences of
regularly spaced prokaryotic repeats derive from foreign genetic elements. J Mol Evol 60,
174-82 (2005).
Moore, C.B., Guthrie, E.H., Huang, M.T. & Taxman, D.J. Short hairpin RNA (shRNA):
design, delivery, and assessment of gene knockdown. Methods Mol Biol 629, 141-158
(2010).
Moore, M.J. & Proudfoot, N.J. Pre-mRNA processing reaches back to transcription and
ahead to translation. Cell 136, 688-700 (2009).
Morris, R.G., Garrud, P., Rawlins, J.N. & O'Keefe, J. Place navigation impaired in rats with
hippocampal lesions. Nature 297, 681-3 (1982).
Moser, E.I., Kropff, E. & Moser, M.B. Place cells, grid cells, and the brain's spatial
representation system. Annu Rev Neurosci 31, 69-89 (2008).
Moser, E.I., Moser, M.B. & Andersen, P. Spatial learning impairment parallels the
magnitude of dorsal hippocampal lesions, but is hardly present following ventral lesions. J
Neurosci 13, 3916-25 (1993).
Mukherjee, N., Corcoran, D.L., Nusbaum, J.D., Reid, D.W., Georgiev, S., Hafner, M., ...
Keene, J.D. Integrative regulatory mapping indicates that the RNA-binding protein HuR
couples pre-mRNA processing and mRNA stability. Mol Cell 43, 327-39 (2011).
Neve, R.L., Neve, K.A., Nestler, E.J. & Carlezon, W.A. Use of herpes virus amplicon vectors
to study brain disorders. Biotechniques 39, 381-9 (2005).
O'Keefe, J. & Dostrovsky, J. The hippocampus as a spatial map. Preliminary evidence from
unit activity in the freely-moving rat. Brain Res 34, 171-175 (1971).
Oleynikov, Y. & Singer, R.H. Real-time visualization of ZBP1 association with beta-actin
mRNA during transcription and localization. Curr Biol 13, 199-207 (2003).
O'Malley, A., O'Connell, C., Murphy, K.J. & Regan, C.M. Transient spine density increases
in the mid-molecular layer of hippocampal dentate gyrus accompany consolidation of a
spatial learning task in the rodent. Neuroscience 99, 229-32 (2000).
Ostroff, L.E., Fiala, J.C., Allwardt, B. & Harris, K.M. Polyribosomes redistribute from
dendritic shafts into spines with enlarged synapses during LTP in developing rat
hippocampal slices. Neuron 35, 535-45 (2002).
Ouyang, Y., Rosenstein, A., Kreiman, G., Schuman, E.M. & Kennedy, M.B. Tetanic
stimulation leads to increased accumulation of Ca2+/calmodulin-dependent protein kinase II
via dendritic protein synthesis in hippocampal neurons. J Neurosci 19, 7823-33 (1999).
Paradies, M.A. & Steward, O. Multiple subcellular mRNA distribution patterns in neurons: a
nonisotopic in situ hybridization analysis. J Neurobiol 33, 473-93 (1997).
86
Ping, X.L., Sun, B.F., Wang, L., Xiao, W., Yang, X., Wang, W.J., ... Chen, Y.S. Mammalian
WTAP is a regulatory subunit of the RNA N6-methyladenosine methyltransferase. Cell Res
24, 177-89 (2014).
Pittenger, C., Huang, Y.Y., Paletzki, R.F., Bourtchouladze, R., Scanlin, H., Vronskaya, S. &
Kandel, E.R. Reversible inhibition of CREB/ATF transcription factors in region CA1 of the
dorsal hippocampus disrupts hippocampus-dependent spatial memory. Neuron 34, 447-62
(2002).
Pourcel, C., Salvignol, G. & Vergnaud, G. CRISPR elements in Yersinia pestis acquire new
repeats by preferential uptake of bacteriophage DNA, and provide additional tools for
evolutionary studies. Microbiology 151, 653-63 (2005).
Quevedo, J., Vianna, M.R.M., Roesler, R., de-Paris, F., Izquierdo, I. & Rose, S.P.R. Two
time windows of anisomycin-induced amnesia for inhibitory avoidance training in rats:
protection from amnesia by pretraining but not pre-exposure to the task apparatus. Learn
Mem 6, 600-7 (1999).
Ramírez-Amaya, V., Vazdarjanova, A., Mikhael, D., Rosi, S., Worley, P.F. & Barnes, C.A.
Spatial exploration-induced Arc mRNA and protein expression: evidence for selective,
network-specific reactivation. J Neurosci 25, 1761-8 (2005).
Ramón y Cajal, S. Estructura del asta de Ammon y fascia dentata. Anales Soc Esp Hist Nat 2,
53-125 (1893).
Ramón y Cajal, S. The Croonian Lecture: La fine structure des centres nerveux. Proc R Soc L
55, 444-68 (1894).
Ramón y Cajal, S. Histologie du système nerveux de l'homme et des vertébrés (Consejo
Superior de Investigaciones Cientificas, Madrid, 1954).
Ran, F.A., Cong, L., Yan, W.X., Scott, D.A., Gootenberg, J.S., Kriz, A.J., ... Zhang, F. In
vivo genome editing using Staphylococcus aureus Cas9. Nature 520, 186-91 (2015).
Ran, F.A., Hsu, P.D., Wright, J., Agarwala, V., Scott, D.A. & Zhang, F. Genome engineering
using the CRISPR-Cas9 system. Nature Protoc 8, 2281-308 (2013).
Redondo, R.L. & Morris, R.G. Making memories last: the synaptic tagging and capture
hypothesis. Nat Rev Neurosci 12, 17-30 (2011).
Rook, M.S., Lu, M. & Kosik, K.S. CaMKII-alpha 3′ untranslated region-directed mRNA
translocation in living neurons: visualization by GFP linkage. J Neurosci 20, 6385-93 (2000).
Rudy, J.W., Barrientos, R.M. & O'Reilly, R.C. Hippocampal formation supports conditioning
to memory of a context. Behav Neurosci 116, 530-8 (2002).
Sajikumar, S. & Frey, J.U. Anisomycin inhibits the late maintenance of long-term depression
in rat hippocampal slices in vitro. Neurosci Lett 338, 147-50 (2003).
Saxe, M.D., Battaglia, F., Wang, J.W., Malleret, G., David, D.J., Monckton, J.E., ... Drew,
M.R. Ablation of hippocampal neurogenesis impairs contextual fear conditioning and
synaptic plasticity in the dentate gyrus. Proc Natl Acad Sci U S A 103, 17501-6 (2006).
87
Schmittgen, T.D. & Livak, K.J. Analyzing real-time PCR data by the comparative CT
method. Nature Protoc 3, 1101-8 (2008).
Schratt, G.M., Tuebing, F., Nigh, E.A., Kane, C.G., Sabatini, M.E., Kiebler, M. &
Greenberg, M.E. A brain-specific microRNA regulates dendritic spine development. Nature
439, 283-9 (2006).
Scott, L.J., Mohlke, K.L., Bonnycastle, L.L., Willer, C.J., Li, Y., Duren, W.L., ... Boehnke,
M. A genome-wide association study of type 2 diabetes in Finns detects multiple
susceptibility variants. Science 316, 1341-5 (2007).
Scoville, W.B. & Milner, B. Loss of recent memory after bilateral hippocampal lesions. J
Neurol Neurosurg Psychiatry 20, 11-21 (1957).
Sekeres, M.J., Mercaldo, V., Richards, B., Sargin, D., Mahadevan, V., Woodin, M.A., ...
Josselyn, S.A. Increasing CRTC1 function in the dentate gyrus during memory formation or
reactivation increases memory strength without compromising memory quality. J Neurosci
32, 17857-68 (2012).
Sekeres, M.J., Neve, R.L., Frankland, P.W. & Josselyn, S.A. Dorsal hippocampal CREB is
both necessary and sufficient for spatial memory. Learn Mem 17, 280-3 (2010).
Senís, E., Fatouros, C., Große, S., Wiedtke, E., Niopek, D., Mueller, A.K., ... Grimm, D.
CRISPR/Cas9-mediated genome engineering: an adeno-associated viral (AAV) vector
toolbox. Biotechnol J 9, 1402-12 (2014).
Shevtsova, Z., Malik, J.M., Michel, U., Bähr, M. & Kügler, S. Promoters and serotypes:
targeting of adeno-associated virus vectors for gene transfer in the rat central nervous system
in vitro and in vivo. Exp Physiol 90, 53-9 (2005).
Shin, C.Y., Kundel, M. & Wells, D.G. Rapid, activity-induced increase in tissue plasminogen
activator is mediated by metabotropic glutamate receptor-dependent mRNA translation. J
Neurosci 24, 9425-33 (2004).
Simons, J.S. & Spiers, H.J. Prefrontal and medial temporal lobe interactions in long-term
memory. Nat Rev Neurosci 4, 637-48 (2003).
Solomon, P.R., Vander Schaaf, E.R., Thompson, R.F. & Weisz, D.J. Hippocampus and trace
conditioning of the rabbit's classically conditioned nictitating membrane response. Behav
Neurosci 100, 729-44 (1986).
Squire, L.R. Memory and the hippocampus: a synthesis from findings with rats, monkeys,
and humans. Psychol Rev 99, 195-231 (1992).
Squire, L.R. Memory systems of the brain: a brief history and current perspective. Neurobiol
Learn Mem 82, 171-7 (2004).
Squire, L.R., Amaral, D.G. & Press, G.A. Magnetic resonance imaging of the hippocampal
formation and mammillary nuclei distinguish medial temporal lobe and diencephalic
amnesia. J Neurosci 10, 3106-17 (1990).
88
Squire, L.R. & Zola-Morgan, S. The medial temporal lobe memory system. Science 253,
1380-6 (1991).
Squires, J.E., Patel, H.R., Nousch, M., Sibbritt, T., Humphreys, D.T., Parker, B.J., ... Preiss,
T. Widespread occurrence of 5-methylcytosine in human coding and non-coding RNA.
Nucleic Acids Res 40, 5023-33 (2012).
Srikantan, S., Tominaga, K. & Gorospe, M. Functional interplay between RNA-binding
protein HuR and microRNAs. Curr Protein Pept Sci 13, 372-9 (2012).
Stanton, P.K. & Sarvey, J.M. Blockade of long-term potentiation in rat hippocampal CA1
region by inhibitors of protein synthesis. J Neurosci 4, 3080-8 (1984).
Stebbins-Boaz, B., Cao, Q., de Moor, C.H., Mendez, R. & Richter, J.D. Maskin is a CPEB-
associated factor that transiently interacts with eIF-4E. Mol Cell 4, 1017-27 (1999).
Steward, O. & Levy, W.B. Preferential localization of polyribosomes under the base of
dendritic spines in granule cells of the dentate gyrus. J Neurosci 2, 284-91 (1982).
Steward, O. & Reeves, T.M. Protein-synthetic machinery beneath postsynaptic sites on CNS
neurons: association between polyribosomes and other organelles at the synaptic site. J
Neurosci 8, 176-84 (1988).
Steward, O. & Schuman, E.M. Protein synthesis at synaptic sites on dendrites. Ann Rev
Neurosci 24, 299-325 (2001).
Steward, O., Wallace, C.S., Lyford, G.L. & Worley, P.F. Synaptic activation causes the
mRNA for the IEG Arc to localize selectively near activated postsynaptic sites on dendrites.
Neuron 21, 741-51 (1998).
Steward, O. & Worley, P.F. Selective targeting of newly synthesized Arc mRNA to active
synapses requires NMDA receptor activation. Neuron 30, 227-40 (2001).
Stratigopoulos, G., Padilla, S.L., LeDuc, C.A., Watson, E., Hattersley, A.T., McCarthy, M.I.,
... Leibel, R.L. Regulation of Fto/Ftm gene expression in mice and humans. Am J Physiol
Regul Integr Comp Physiol 294, R1185-96 (2008).
Svec, D., Tichopad, A., Novosadova, V., Pfaffl, M.W. & Kubista, M. How good is a PCR
efficiency estimate: Recommendations for precise and robust qPCR efficiency assessments.
Biomol Detect Quantif 3, 9-16 (2015).
Swanson, L.W., Wyss, J.M. & Cowan, W.M. An autoradiographic study of the organization
of intrahippocampal association pathways in the rat. J Comp Neurol 181, 681-715 (1978).
Swiech, L., Heidenreich, M., Banerjee, A., Habib, N., Li, Y., Trombetta, J., ... Zhang, F. In
vivo interrogation of gene function in the mammalian brain using CRISPR-Cas9. Nature
Biotechnol 33, 102-6 (2015).
Thut, P.D. & Lindell, T.J. Alpha-amanitin inhibition of mouse brain form II ribonucleic acid
polymerase and passive avoidance retention. Mol Pharmacol 10, 146-54 (1974).
89
Tiedge, H. & Brosius, J. Translational machinery in dendrites of hippocampal neurons in
culture. J Neurosci 16, 7171-81 (1996).
Tiruchinapalli, D.M., Oleynikov, Y., Kelič, S., Shenoy, S.M., Hartley, A., Stanton, P.K., ...
Bassell, G.J. Activity-dependent trafficking and dynamic localization of zipcode binding
protein 1 and beta-actin mRNA in dendrites and spines of hippocampal neurons. J Neurosci
23, 3251-61 (2003).
Torre, E.R. & Steward, O. Demonstration of local protein synthesis within dendrites using a
new cell culture system that permits the isolation of living axons and dendrites from their cell
bodies. J Neurosci 12, 762-72 (1992).
Tsien, J.Z., Huerta, P.T. & Tonegawa, S. The essential role of hippocampal CA1 NMDA
receptor-dependent synaptic plasticity in spatial memory. Cell 87, 1327-38 (1996).
Tung, Y.C., Ayuso, E., Shan, X., Bosch, F., O'Rahilly, S., Coll, A.P. & Yeo, G.S.
Hypothalamic-specific manipulation of Fto, the ortholog of the human obesity gene FTO,
affects food intake in rats. PLoS One 5, e8771 (2010).
van den Pol, A.N., Ozduman, K., Wollmann, G., Ho, W.S., Simon, I., Yao, Y., ... Ghosh, P.
Viral strategies for studying the brain, including a replication-restricted self-amplifying delta-
G vesicular stomatis virus that rapidly expresses transgenes in brain and can generate a
multicolor golgi-like expression. J Comp Neurol 516, 456-81 (2009).
van Gemert, N.G., Carvalho, D.M.M., Karst, H., Van Der Laan, S., Zhang, M., Meijer, O.C.,
... Joels, M. Dissociation between rat hippocampal CA1 and dentate gyrus cells in their
response to corticosterone: effects on calcium channel protein and current. Endocrinology
150, 4615-24 (2009).
Van Hoesen, G.W., Augustinack, J.C. & Redman, S.J. Ventromedial temporal lobe
pathology in dementia, brain trauma, and schizophrenia. Ann N Y Acad Sci 877, 575-94
(1999).
Vasudevan, S., Tong, Y. & Steitz, J.A. Switching from repression to activation: microRNAs
can up-regulate translation. Science 318, 1931-4 (2007).
Vessey, J.P., Vaccani, A., Xie, Y., Dahm, R., Karra, D., Kiebler, M.A. & Macchi, P.
Dendritic localization of the translational repressor Pumilio 2 and its contribution to dendritic
stress granules. J Neurosci 26, 6496-508 (2006).
Volpe, B.T., Davis, H.P., Towle, A. & Dunlap, W.P. Loss of hippocampal CA1 pyramidal
neurons correlates with memory impairment in rats with ischemic or neurotoxin lesions.
Behav Neurosci 106, 457-64 (1992).
Walters, B.J., Azam, A.B., Gillon, C.J., Josselyn, S.A. & Zovkic, I.B. Advanced in vivo use
of CRISPR/Cas9 and anti-sense DNA inhibition for gene manipulation in the brain. Front
Genet 6 (2015).
Wang, X., Lu, Z., Gomez, A., Hon, G.C., Yue, Y., Han, D., ... He, C. N6-methyladenosine-
dependent regulation of messenger RNA stability. Nature 505, 117-20 (2014).
90
Wang, X., Zhao, B.S., Roundtree, I.A., Lu, Z., Han, D., Ma, H., ... He, C. N(6)-
methyladenosine Modulates Messenger RNA Translation Efficiency. Cell 161, 1388-99
(2015).
Wang, X., Zhu, L., Chen, J. & Wang, Y. mRNA m(6)A methylation downregulates
adipogenesis in porcine adipocytes. Biochem Biophys Res Commun 459, 201-7 (2015).
Wang, Y., Li, Y., Toth, J.I., Petroski, M.D., Zhang, Z. & Zhao, J.C. N6-methyladenosine
modification destabilizes developmental regulators in embryonic stem cells. Nat Cell Biol 16,
191-8 (2014).
Wei, C.M., Gershowitz, A. & Moss, B. Methylated nucleotides block 5' terminus of HeLa
cell messenger RNA. Cell 4, 379-86 (1975).
Weiler, I.J., Irwin, S.A., Klintsova, A.Y., Spencer, C.M., Brazelton, A.D., Miyashiro, K., ...
Greenough, W.T. Fragile X mental retardation protein is translated near synapses in response
to neurotransmitter activation. Proc Natl Acad Sci U S A 94, 5395-400 (1997).
Wells, D.G., Dong, X., Quinlan, E.M., Huang, Y.S., Bear, M.F., Richter, J.D. & Fallon, J.R.
A role for the cytoplasmic polyadenylation element in NMDA receptor-regulated mRNA
translation in neurons. J Neurosci 21, 9541-8 (2001).
Wheeler, A.L., Teixeira, C.M., Wang, A.H., Xiong, X., Kovacevic, N., Lerch, J.P., ...
Frankland, P.W. Identification of a functional connectome for long-term fear memory in
mice. PLoS Comput Biol 9, e1002853 (2013).
Wilkie, G.S., Dickson, K.S. & Gray, N.K. Regulation of mRNA translation by 5'- and 3'-
UTR-binding factors. Trends Biochem Sci 28, 182-8 (2003).
Wilson, M.A. & McNaughton, B.L. Reactivation of hippocampal ensemble memories during
sleep. Science 265, 676-9 (1994).
Wilson, R.C. & Doudna, J.A. Molecular mechanisms of RNA interference. Annu Rev
Biophys 42, 217-39 (2013).
Wiltgen, B.J., Sanders, M.J., Anagnostaras, S.G., Sage, J.R. & Fanselow, M.S. Context fear
learning in the absence of the hippocampus. J Neurosci 26, 5484-91 (2006).
Wimsatt, W.C. in Re-Engineering Philosophy for Limited Beings (ed. Press, H.U.) 449
(2007).
Wu, L., Wells, D.G., Tay, J., Mendis, D., Abbott, M.A., Barnitt, A., ... Richter, J.D. CPEB-
mediated cytoplasmic polyadenylation and the regulation of experience-dependent translation
of α-CaMKII mRNA at synapses. Neuron 21, 1129-1139 (1998).
Yiu, A.P., Mercaldo, V., Yan, C., Richards, B., Rashid, A.J., Hsiang, H.-L.L., ... Kushner,
S.A. Neurons are recruited to a memory trace based on relative neuronal excitability
immediately before training. Neuron 83, 722-35 (2014).
Yiu, A.P., Rashid, A.J. & Josselyn, S.A. Increasing CREB function in the CA1 region of
dorsal hippocampus rescues the spatial memory deficits in a mouse model of Alzheimer's
disease. Neuropsychopharmacology 36, 2169-86 (2011).
91
Zalfa, F., Giorgi, M., Primerano, B., Moro, A., Di Penta, A., Reis, S., ... Bagni, C. The fragile
X syndrome protein FMRP associates with BC1 RNA and regulates the translation of
specific mRNAs at synapses. Cell 112, 317-27 (2003).
Zetterström, R.H., Williams, R., Perlmann, T. & Olson, L. Cellular expression of the
immediate early transcription factors Nurr1 and NGFI-B suggests a gene regulatory role in
several brain regions including the nigrostriatal dopamine system. Mol Brain Res 41, 111-20
(1996).
Zhang, M., Zhang, Y., Ma, J., Guo, F., Cao, Q., Zhang, Y., ... Zhao, R. The demethylase
activity of FTO (fat mass and obesity associated protein) is required for preadipocyte
differentiation. PLoS One 10, e0133788 (2015).
Zhang, Y.Q., Bailey, A.M., Matthies, H.J., Renden, R.B., Smith, M.A., Speese, S.D., ...
Broadie, K. Drosophila fragile X-related gene regulates the MAP1B homolog Futsch to
control synaptic structure and function. Cell 107, 591-603 (2001).
Zhang, Z., Theler, D., Kaminska, K.H., Hiller, M., de la Grange, P., Pudimat, R., ... Allain,
F.H.T. The YTH domain is a novel RNA binding domain. J Biol Chem 285, 14701-10
(2010).
Zhao, X., Yang, Y., Sun, B.F., Shi, Y., Yang, X., Xiao, W., ... Wang, W.J. FTO-dependent
demethylation of N6-methyladenosine regulates mRNA splicing and is required for
adipogenesis. Cell Res 24, 1403-19 (2014).
Zheng, G., Dahl, J.A., Niu, Y., Fedorcsak, P., Huang, C.M., Li, C.J., ... Song, S.H. ALKBH5
is a mammalian RNA demethylase that impacts RNA metabolism and mouse fertility. Mol
Cell 49, 18-29 (2013).
Zhong, S., Li, H., Bodi, Z., Button, J., Vespa, L., Herzog, M. & Fray, R.G. MTA is an
Arabidopsis messenger RNA adenosine methylase and interacts with a homolog of a sex-
specific splicing factor. Plant Cell 20, 1278-88 (2008).
Zhou, J., Wan, J., Gao, X., Zhang, X., Jaffrey, S.R. & Qian, S.B. Dynamic m6A mRNA
methylation directs translational control of heat shock response. Nature 526, 591-4 (2015).
Zola-Morgan, S., Squire, L.R. & Amaral, D.G. Human amnesia and the medial temporal
region: enduring memory impairment following a bilateral lesion limited to field CA1 of the
hippocampus. J Neurosci 6, 2950-67 (1986).
Zovkic, I.B., Guzman-Karlsson, M.C. & Sweatt, J.D. Epigenetic regulation of memory
formation and maintenance. Learn Mem 20, 61-74 (2013).
92
Appendix
Figure S1. m6A pathway components levels in context-only or shock-only controls. None of the pathway
components showed changes at different time points following context-only exposure or shock-only exposure.
A. Mettl3 context (F3,34 = 0.87, p < 0.05), shock (F3,33 = 0.82, p < 0.05). B. Mettl14 context (F3,27 = 1.51, p < 0.05),
shock (F3,26 = 1.11, p < 0.05). C. Alkbh5 context (F3,14 = 0.10, p < 0.05), shock (F3,14 = 3.156, p < 0.05). D. Fto
context (F3,28 = 1.86, p < 0.05), shock (F3,28 = 2.09, p < 0.05). E. Elavl1 context (F3,30 = 0.11, p < 0.05), shock
(F3,27 = 0.70, p < 0.05). F. Ythdf2 context (F3,24 = 1.59, p < 0.05), shock (F3,23 = 1.33, p < 0.05). All time points
normalized to naïve levels. Time points for context-only or shock-only conditions were collapsed into single groups
for each gene for analysis in Fig. 3.2. n = 5-11 (30 min, context-only), 5-8 (1h, context-only), 4-10 (2h context-
only), 4-8 (4h context-only), 5-13 (30 min, shock-only), 5-8 (1h, shock-only), 4-8 (2h shock-only), 4-8 (4h shock-
only). *p < 0.05.