The molecular basis of structural plasticity of dendritic spines
by
Jackie Liu
A thesis submitted in conformity with the requirements for the degree of Master of Science
Physiology University of Toronto
© Copyright by Jackie Jia Liu 2014
ii
Abstract
The molecular basis of structural plasticity of dendritic spines
By Jackie Jia Liu, Master of Science, 2014
Graduate Department of Physiology, University of Toronto
The dynamics of actin, the major cytoskeletal component in dendritic spines, is responsible for
the regulation of spine morphogenesis, formation, and elimination. Furthermore, activity-induced
input-specific actin polymerization at dendritic spines have been suggested as a plausible
mechanism for synaptic tagging, a process that involves the selective marking of potentiated
synapses that is necessary for the proper establishment of LTP. P21-activated kinases (PAKs) are
known regulators of actin dynamics and their mutations have been linked with cognitive deficits
and impaired synaptic plasticity. Despite the importance of PAKs in learning and memory, their
role in dendritic spines’ structural plasticity remains elusive. Here, we show that group I PAKs
are important for basal spine morphology, density, and maturation by using two-photon
microscopy and organotypic hippocampal slice cultures. In addition, PAK1/3 DKO also
exhibited impaired structural plasticity and input-specificity following glutamate-uncaging, but
normal basal spine dynamics, demonstrating the importance of PAKs in activity-dependent
structural modifications.
iii
Acknowledgements
I would like to thank my supervisor (Dr. Zhengping Jia) and our collaborator/committee member
(Dr. Keniche Okamoto) for their continuous support throughout the project. I would also like to
thank Dr. Okamoto for generously lending me his equipment, reagents, (two-photon microscope,
slice-culture preparation setups, incubator, gene-gun apparatus, pCAGGS-EGFP, etc.) and for
his insightful guidance throughout the project. I would like to thank my supervisory committee
member, Dr. Luyang Wang, for his insightful comments and suggestions in all of our meetings.
I greatly appreciate all the help and support that I have received from members of the Jia Lab and
Okamoto lab. In particular, I would like to thank Gurpreet Lakhanpal (Okamoto lab) and
Mustafa Khan (Okamoto lab) for training me extensively on slice culture preparation, imaging,
and analysis. In addition, I would like to thank Tyler Luyben (Okamoto lab) and Celeste Leung
(Jia Lab) for their help with imaging. I would like to thank my students, Ruhan Wei, Mengyuan
Zhu, and Su Jin Lee, as well as our previous lab technician, Shouping Zhang, for their help with
genotyping.
Furthermore, I would like to thank my friends and fellow lab members, Yelena Borovac
(Okamoto lab), Laurence David (Hospital for Sick Children), Helena Kim (Department of
Pharmacology and Toxcicology), and Tyler Luyben (Okamoto lab) for their very helpful
comments and suggestions on my thesis.
iv
Table of contents
Chapter 1 – Introduction I.………………………………………………………………………...1
1.1 Synaptic Transmission and Plasticity.…………………………………………….1
1.2 LTP Persistence and Specificity – The Synaptic Tagging and Capturing Model…2
1.3 Synaptic Tag Candidates …………………………………………………………6
1.4 Dendritic Spines and Structural Plasticity ………………………………………..9
1.4.1 Dendritic spines…………………………………………………………...9
1.4.2 Structural plasticity………………………………………………………10
1.5 Actin Dynamics and Regulation…………………………………………………13
1.5.1 Actin dynamics and their regulators……………………………………..13
1.5.2 Commonly used pharmacological agents that affect actin dynamics…....15
1.5.3 Actin dynamics during synaptic plasticity……………………………….15
1.5.4 Actin cytoskeleton regulation……………………………………………18
1.5.4.1 CaMKII…………………………………………………………..20
1.5.4.2 Rho-GTPase signaling in synaptic plasticity and actin
regulation………………………………………………………...............21
1.5.4.3 Cofilin……………………………………………………………25
1.5.4.4 Other ABPs………………………………………………………27
1.5.4.5 Current model of actin regulation during plasticity……………...28
1.6 The Role of p-21 Activated Kinases in Synaptic Properties ……........................29
1.6.1 The two families of PAKs: structures and activation mechanisms……...29
1.6.2 The regulation of PAK1 and PAK3……………………………………..31
v
1.6.3 The regulation of actin dynamics and synaptic properties by PAK1 and
PAK3……………………………………………………………………33
1.6.4 Other PAKs………………………………………………………………36
Chapter 2 – Rationale and Hypothesis …………………………………………………………..38
Chapter 3 – Materials and Methods ……………………………………………………………..45
3.1 Genotyping……………………………………………………………………….45
3.1.1 DNA extraction…………………………………………………………..45
3.1.2 PCR and gel-electrophoresis……………………………………………..46
3.2 Material Preparation for Culture and Imaging...…………………………………47
3.2.1 Slice culture media……………………………………………………….47
3.2.2 Gene-gun bullets…………………………………………………………48
3.3 Organotypic Culture……………………………………………………………..49
3.4 Transfection……………………………………………………………………...50
3.5 Imaging…………………………………………………………………………..50
3.5.1 Basal spine properties……………………………………………………50
3.5.2 Basal spine dynamics and structural plasticity following glutamate
uncaging………………………………………………………………….51
3.6 Image Processing: 3D Deconvolution…………………………………………...52
3.7 Analysis…………………………………………………………………………..52
Chapter 4 – Results I: p-21 Activated Kinases are Important Regulators of Basal Spine Density
and Morphology …………………………………………………………………....56
Chapter 5 – Results II: PAK1/3 DKO Neurons Exhibit Impaired and Non-specific LTP-
Induced Structural Plasticity………………………………………………………..62
vi
5.1 PAK1/3 DKO Stimulated Spines Exhibited Faster Decay of the
Transient Spine-Enlargement Phase……………………………………………..62
5.2 p-21 Activated Kinases are Important for Input-Specificity Associated
With LTP………………………………………………………………………...68
5.3 PAK1/3 DKO Exhibited Normal Basal Structural Dynamics of Dendritic
Spines…………………………………………………………………………….69
Chapter 6 – Discussion and Conclusion………………………………………………………....76
6.1 PAK1/3 DKO Exhibited Normal Basal Structural Dynamics of Dendritic
Spines…………………………………………………………………………….76
6.2 Group I PAKs are Important for Input-Specific Structural Plasticity during
LTP………………………………………………………………………………77
6.3 Potential mechanisms underlying input-specific LTP…………………………...79
6.4 Limitations and Technical Considerations………………………………………82
6.4.1 Other PAK functions…………………………………………………….84
6.4.2 Other technical considerations…………………………………………...84
vii
List of Tables
Table 1. Spinogenesis over time in PAK1/3 DKO………………………………………………74
viii
List of Figures
Figure 1. Electrophysiological experiments for examining synaptic tagging and capturing
hypothesis……………………………………………………………………………….4
Figure 2. Bidirectional regulation of actin cytoskeleton during synaptic plasticity …………….19
Figure 3. Activity-dependent Rho-GTPases-PAK-cofilin regulation in dendritic spines……….43
Figure 4. Measurement of spine head width …………………………………………………….55
Figure 5. PAK1 and PAK3 are important regulators of basal spine density and morphology…..59
Figure 6. Glutamate-uncaging induced spine enlargement in both WT/HET and
PAK1/3 DKO ……………………………………………………………………….....65
Figure 7. PAK1/3 DKO spines exhibited more dynamic enlargements ………………………...67
Figure 8. Glutamate-uncaging induced enlargement of neighboring spines in PAK1/3 DKO but
not in WT/HET …………………………………………………………………..........71
Figure 9. PAK1/3 DKO did not exhibit abnormal spontaneous changes in spine size ………....75
Figure SI1. Basal spine morphology at DIV 7-8………………………………………………...96
Figure SI2. PAK1/3 DKO exhibited more elongated spines at DIV 15-16……………………...98
ix
List of Abbreviations
AMPA: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
AMPAR: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors
ca: constitutively active
CaMKII: Ca2+/calmodulin-dependent protein kinase II
Cyto D: cytochalasin D
DIV: days in vitro
DKO: double knockout
dn: dominant negative
E-LTP: early-phase long-term potentiation
GAP: GTPase-activating proteins
GEF: guanine nucleotide exchange factor
HFS: high-frequency stimulation
KD: knockdown
KO: knockout
Lat A: latrunculin A
LTD: long-term depression
LTP: long-term potentiation
L-LTP: late-phase long-term potentiation
NMDA: N-methyl-D-aspartate
NMDAR: N-methyl-D-aspartate receptors
PAK: p21-activated kinase
PRP: plasticity related protein
x
STET: strong tetanization
TBS: theta-burst stimulation
WTET: weak tetanization
1
Chapter 1
Introduction
1.1 Synaptic Transmission and Plasticity
The efficiency of transmission of information from one neuron to the other neuron relies on the
structural and functional integrity of the pre- and postsynaptic elements of the synapse
(Hotulainen and Hoogenraad, 2010). Upon the arrival of action potentials at a presynaptic axon
terminal, voltage-gated calcium (Ca2+) channels open, allowing an influx of Ca2+ which binds to
the components of synaptic-release machinery such as synaptotagmin on the synaptic vesicle,
allowing the release of neurotransmitters into the synaptic cleft. The released neurotransmitters
(e.g. glutamate) travel across synaptic cleft to bind to their receptors located at the post-synaptic
membrane. The binding of these neurotransmitters to their receptors can subsequently facilitate
the opening of postsynaptic channels and initiates a cascade of signalling events.
Synaptic plasticity is the experience-dependent alteration in synaptic efficacy and is widely
accepted as the basis for learning and memory (Hotulainen and Hoogenraad, 2010; Ho et al.,
2011). N-methyl-D-aspartate receptors (NMDAR) are specific type of ionotropic glutamate
receptors, which are blocked by magnesium (Mg2+) at resting potential but allow influx of Ca2+
upon activation (Ho et al., 2011). When the postsynaptic membrane is sufficiently depolarized,
the Mg2+ is expelled from the channel pore, allowing Ca2+ entry into the postsynaptic sites,
2
triggering cascade of cytoplasmic events, including the activation of the serine/threonine protein
kinase, Ca2+/calmodulin-dependent protein kinase II (CaMKII) (Okamoto et al., 2007; Okamoto
et al., 2009; Ho et al., 2011). The activation of CaMKII promotes its dissociation from F-actin, a
allowing the actin cytoskeleton to reorganize and expand during synaptic activity (Okamoto et al.,
2007; Okamoto et al., 2009). In addition, the mobile CaMKII can also trigger many other events
including phosphorylation of the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
(AMPA) glutamate receptors (AMPAR), a significant event necessary for increasing synaptic
efficacy during long term potentiation (LTP) (Okamoto et al., 2007; Okamoto et al., 2009; Ho et
al., 2011). During LTP, AMPAR are trafficked to and inserted at extrasynaptic sites via
exocytosis, which are then moved to synaptic sites via lateral diffusion (Lee et al., 2003;
Lamprecht and LeDoux, 2004; Ho et al., 2011). This increased surface AMPAR expression at
postsynaptic membrane is believed to underlie the enhanced synaptic strength during NMDAR-
dependent LTP in the hippocampus (Lamprecht and LeDoux, 2004; Ho et al., 2011). In contrast,
during long-term depression (LTD), which is a period of reduced synaptic efficacy, AMPA
receptors are believed to be removed from postsynaptic sites, resulting in an decreased response
to glutamate (Lee et al., 2003; Ho et al., 2011).
1.2 LTP Persistence and Specificity – the Synaptic
Tagging and Capturing Model
The allocation of memory is believed to be governed by specific synapses that display LTP, a
process thought to generate the physical trace of memory, and as a consequence, impairments in
3
such memory allocation mechanisms may lead to cognitive deficits (Bliss and Collingridge, 1993;
Rogerson et al., 2014).
LTP has two distinct phases, the early protein-synthesis independent phase (E-LTP), and the late
protein-synthesis dependent maintenance phase (L-LTP) (Frey and Morris, 1997; Ramachandran
and Frey, 2009). The maintenance phase of LTP requires the synthesis of plasticity related
proteins (PRPs), which are mostly synthesized at dendrites and soma (Frey and Morris, 1997;
Ramachandran, 2009; Ho et al., 2011). However, how these PRPs are trafficked to the activated
synapses and selectively interact with these synapses to enable persistent changes in their
synaptic efficacy remains unclear (Frey and Morris, 1997; Ramachandran, 2009; Ho et al., 2011).
To explain this phenomenon, Frey and Morris (1997) proposed the synaptic tag and capture
hypothesis, which states that during E-LTP, a protein-synthesis independent “synaptic tag” is
generated to transiently mark these potentiated synapses after synaptic activity. This transient tag
could subsequently sequester other proteins that are required for LTP maintenance to these sites,
enabling the expression of L-LTP at these selective synapses (Frey and Morris, 1997, 1998).
This idea was investigated by performing series of experiments by applying tetanus stimulation
of varying frequencies and pharmacological manipulations (Figure 1). More specifically, the
authors proposed that instead of trafficking the PRPs to the activated sites, PRPs could travel
along the dendrite but only get sequestered at the activated synapses due to their interactions
with the tags. This proposed process renders the proper setting of synaptic tags especially critical
for the establishment of proper synaptic connections.
4
Figure 1. Electrophysiological experiments for examining synaptic tagging and capturing
hypothesis (adapted from Frey and Morris, 1997)
Two regions as shown, S1 and S2, were stimulated independently. The following sets of
electrophysiological experiments were carried out based on the knowledge that 1) weak
tetanization (WTET, 1 train, 100Hz, duration 0.2ms) can activate E-LTP, but not L-LTP as it
could not activate the synthesis of plasticity related proteins (PRPs), 2) repeated strong
tetanization (STET, 3 repeated trains of tetanus, 100Hz, 0.2ms) can induce both E-LTP and L-
LTP, and 3) STET is only capable of inducing E-LTP if it was delivered in the presence of
protein synthesis inhibitors (Frey and Morris, 1997, 1998). In Frey and Morris’ (1997, 1998)
study, they demonstrated that the pathway stimulated by WTET can also express L-LTP if the
WTET was delivered 1 hour prior or after a STET was delivered in the same neuronal population.
The authors suggested that WTET stimulation may be capable of producing a tag that marks the
activated synapses, enabling their capture of PRPs whose synthesis was triggered by the stronger
stimulus delivered at the same region (Frey and Morris, 1997, 1998). To test if the PRPs
synthesized following a strong stimulus is responsible for the expression of L-LTP induced by
WTET, anisomycin, a protein synthesis inhibitor, was applied during the delivery of the stronger
5
stimulus. As expected, L-LTPs were abolished in both pathways if protein synthesis triggered by
the strong stimulus was inhibited (Frey and Morris, 1998). To further demonstrate whether these
capturing events are region specific, a STET was delivered in the absence of anisomycin before
two separate regions were stimulated in the presence of anisomycin: one was a region that was
stimulated by stimulus 1, and the other was a region that had not received prior stimulation (Frey
and Morris, 1997). As expected, the neuronal population that was previously stimulated
exhibited L-LTP, but not the other region. Furthermore, it was demonstrated that although
protein synthesis is necessary for L-LTP expression, protein tag-setting is protein synthesis
independent, as anisomycin treatment was unable to block L-LTP expression at S1 and S2 if it
was applied only during the first stimulus (WTET for synaptic tag setting). Lastly, Frey and
Morris (1997) also showed that this protein synthesis independent tag generation is transient. If
STET+anisomycin were applied at 2.5 hours prior to another STET being delivered in the
absence of anisomycin, L-LTP was only induced in the pathway stimulated by the second
stimulus, but not at the pathway stimulated by the first stimulus, suggesting that synaptic tag
production is transient and lasts less than 2.5 hrs as it is no longer able to sequester the required
proteins to the targeted site after this period. Consistent with this idea, when Frey and Morris
(1998) applied the WTET more than 2 hours prior to the STET, the STET did not transform the
WTET induced E-LTP into L-LTP.
6
1.3 Synaptic Tag Candidates
Although the specificity of late-phase synaptic plasticity depends on the selective sequestering
of PRPs at the activated synapses, the mechanism underlying this input-specific synaptic tag
setting during E-LTP remains unclear. To be a synaptic tag, the candidate needs to fulfill the
following requirements: exhibit spatially restricted activation, time-limited and reversible
activation that need to be able to persist long enough to allow the PRPs to identify them, and
ability to interact with other factors that enable persistent and local alteration in synaptic efficacy
(Martin and Kosik, 2002):
To date, several molecules have been proposed to be involved in the tagging process, including
actin and CaMKII (Okamoto et al., 2009; Ramachandran and Frey, 2009; Rogerson et al., 2014).
Actin is the major cytoskeletal component in dendritic spines (Okamoto et al., 2009). It was
shown to be involved in protein trafficking in dendritic spines, and was proposed to regulate LTP
by functioning as a synaptic tag (Langford and Molyneaux, 1998; Ramachandran and Frey,
2009). Similar to Frey and Morris’ (1997 and 1998) experiments, Ramachandran and Frey (2009)
showed that STET at S2 enabled the expression of L-LTP at S1 which was previously stimulated
by a WTET. However, the administration of actin polymerization inhibitor, latrunulin A (Lat A)
and cytochalasin D (Cyto D) between 30 mins prior to and 30 mins after the STET blocked L-
LTP at both S1 and S2, suggesting that an intact actin polymerization mechanism is essential for
the interaction of PRPs with synaptic tags (Ramachandran and Frey, 2009). More specifically,
these results revealed that actin dynamics may be important for synaptic tag setting, synthesis of
7
PRPs, or both. To investigate the specific role of actin dynamics in L-LTP expression, these
authors first induced L-LTP by STET at S1, then 1 hour later, a WTET was applied at S2 either
with Lat A (delivered 30 minutes after STET and washed out 1 hour after) or without. In the
absence of Lat A, L-LTP was shown to be expressed in the pathway stimulated by a WTET as
the tags set by WTET hijack the PRPs produced by STET in S1. In contrast, in the presence of
the actin polymerization inhibitor, this L-LTP was absent in S2 but present in S1, suggesting that
actin polymerization is important for setting the tags. However, when the second stimulus (STET)
was applied in the presence of actin polymerization inhibitor following WTET, L-LTP was
absent at S2 but present at S1 after this stimulation, suggesting that actin polymerization is not
important for PRP synthesis nor its capture by pre-existing tags. On the other hand, co-
application of anisomycin with Lat A or Cyto D during STET following WTET blocked L-LTP
at both S1 and S2. This suggests that new proteins need to be synthesized as well as sequestered
by the synaptic tags, which may require actin polymerization in order to establish L-LTP. In the
same study, Lat A or Cyto D had no effect on baseline recordings, suggesting that actin
polymerization is not involved in the regulation of basal synaptic transmission. On a structural
level, the expression of functional LTP corresponds with an increase in dendritic spine volume as
well as F-actin accumulation (Chen et al., 2007; Honkura et al., 2008). Glutamate-uncaging,
unlike electrophysiological techniques, enables the stimulation of a single spine and was used to
study the expression of structural LTP at the single synapse level. In 2008, Honkura et al. used a
combination of glutamate-uncaging and fluorescence live imaging techniques to demonstrate
actin dynamics in spines following LTP induction. They demonstrated that the persistence of
spine enlargement is determined by the confinement of a CaMKII-dependent activity-induced F-
actin accumulation at the stimulated spines (Honkura et al., 2008). In addition, this activity-
8
induced F-actin increase is restricted to the stimulated spines and persists for at least 30 minutes
(Okamoto et al., 2009). Lastly, F-actin can function as a scaffolding component and sequester
various signalling proteins such as CaMKII and AMPA receptors, fulfilling all the proposed
features of a synaptic tag (Martin and Kosik, 2002; Okamoto et al., 2009).
CaMKII is a serine/threonine protein kinase known to be involved in the regulation of many
cellular processes and is both necessary and sufficient to induce LTP (Okamoto et al., 2007;
Okamoto et al., 2009). Previous studies have shown that NMDA or glutamate application to
hippocampal neurons resulted in an accumulation of CaMKII in spines, suggesting that CaMKII
may be responsible for LTP synapse-specificity (Zhang et al., 2008; Rogerson et al., 2014). To
test if CaMKII could function to selectively tag potentiated synapses, Zhang et al. (2008) showed
that CaMKIIα expression was upregulated in stimulated spines that showed persistent
enlargement (more than 30mins), but not neighboring spines. On the other hand, other
investigators have shown that CaMKIIβ, another abundant form of CaMKII, was demonstrated
to be critical for F-actin bundling and the confinement of activity-induced F-actin accumulation
to the stimulated spines (Okamoto et al., 2007; Honkura et al., 2008).
Similar to the activation pattern of CaMKIIα, the activation of Cdc42, a Rho-family small
GTPases involved in the regulation of actin dynamics, was shown to be increased for a
prolonged time in the activated spine and the small fraction of Cdc42 that spread into the
dendrite was shown to decay sharply (length constant ~1.9um), signifying rapid inactivation of
9
Cdc42 in the dendrite and revealing spatial restricted activation (Murakoshi et al., 2011). Thus,
suggesting their involvement in input-specificity (Murakoshi et al., 2011).
1.4 Dendritic Spines and Structural Plasticity
1.4.1 Dendritic spines
Dendritic spines are actin-rich protrusions where most excitatory synapses are formed. They
serve to compartmentalize postsynaptic signalling molecules such as Ca2+, and their aberrations
are associated with many neuropathologies such as Down syndrome, mental retardation, and
Alzheimer’s disease (Fiala et al., 1998; Hayashi and Majewska, 2005; Shrestha et al., 2006).
Dendritic spines are highly heterogeneous in their size and shape (Hotulainen and Hoogenraad,
2010; Bosch and Hayashi, 2012). Newly formed spines are usually characterized by thin
elongated morphology known as filopodia, which tend be transient in nature (Ziv and Smith,
1996). In fact, study showed that these structures tend to last approximately 10 minutes (Ziv and
Smith, 1996). As they mature, they either gradually develop into mushroom-like protrusions or
retract into the dendrites (Fiala et al., 1998; Bourne and Harris, 2008; Bosch and Hayashi, 2012).
Studies have demonstrated that the size of spine head could be regulated by activity and is
correlated with the size of postsynaptic density, which is an electron dense region filled with
postsynaptic molecules (e.g. neurotransmitter receptors, scaffolding proteins, ion channels, and
cytoskeletal components), and the strengths of NMDA and AMPA receptor-mediated synaptic
transmissions (Hayashi and Majewska, 2005; Cingolani and Goda, 2008; Ho et al., 2011; Bosch
and Hayashi, 2012). These observations suggest that spine head size is strongly correlated with
10
synaptic efficacy. It is postulated that only NMDA receptors are expressed in very small spines
thus AMPA receptor-mediated transmission appears to be functionally “silent” (Matus, 2000;
Bourne and Harris, 2008). However, these spines can be activated by NMDA receptor-dependent
LTP induction, which would subsequently lead to an enlargement of their spine heads and the
exocytic insertion of AMPA receptors (Bourne and Harris, 2008). In contrast to smaller spines,
these larger spines have higher densities of both NMDA and AMPA glutamate receptors,
resulting in their elevated responsiveness to glutamate (Bourne and Harris, 2008).
1.4.2 Structural plasticity
Dendritic spines are highly dynamic structures that undergo constant remodeling and their
morphology and density were shown to be altered by stimulation in vitro and experience in vivo
(Engert and Bonhoeffer, 1999; Hayashi and Majewska, 2005; Hotulainen and Hoogenraad, 2010).
In time-lapse imaging studies, spontaneous formation, elimination and morphogenesis of
dendritic spines were observed in hippocampal slice cultures (Dubos et al., 2012). More
importantly, spines were demonstrated to show activity dependent dynamics.
Following LTP induction and learning tasks, an increase in spine density has been observed and
this increase was shown to be NMDAR activation dependent as it was completely blocked by the
administration of APV, an NMDAR antagonist (Matus, 2000; Lamprecht and LeDoux, 2004; Ho
et al., 2011). In another study, tetanic stimulation (2 trains, 100 pulse at 100Hz) was shown to
induce NMDAR-dependent generation of filopodia ~20 mins post stimulation. Moreover, as time
persists, 27% of these newly formed filopodia developed into mature spines (Maletic-Savatic et
11
al., 1999). This activity-induced spinogenesis was also seen in several other studies (Engert and
Bonhoeffer, 1999).
To further examine the specificity and morphological alterations associated with LTP expression,
numerous imaging studies utilized two-photon MNI-glutamate uncaging, which enables the
selective release of glutamate 1 µm in range to dendritic spines, to induce single-synapse E-LTP.
These studies consistently demonstrated immediate increases in spine size as well as AMPA
receptor mediated transmission of the stimulated spines (Matsuzaki et al., 2004; Hayashi and
Majewska, 2005; Tanaka et al., 2008; Okamoto et al., 2009; Bosch et al., 2014; Meyer et al.,
2014). In addition, neither spine size nor synaptic transmission of neighboring spines were
altered by glutamate uncaging, suggesting that both structural and functional plasticity are highly
input-specific (Matsuzaki et al., 2004; Hayashi and Majewska, 2005; Tanaka et al., 2008;
Okamoto et al., 2009; Meyer et al., 2014).
Further, this increase in spine size and strengthening of AMPA receptor mediated transmission
were completely abolished by NMDAR antagonist APV (Matsuzaki et al., 2004). Unlike APV,
which blocked both transient (5 mins), CaMKII
inhibitor, KN62, was only able to block sustained spine enlargements and not transient
enlargements (Matsuzaki et al., 2004). Together, these results suggest that while both the early
and sustained phases of glutamate-induced changes in spine size and synaptic efficacy require
NMDA receptor activation, CaMKII activation may only be necessary for sustained spine
enlargement (Matsuzaki et al., 2004). This sustained phase requirement for CaMKII is consistent
12
with its delayed accumulation peak time in dendritic spines, which was found to peak at ~10
minutes post stimulation (Zhang et al., 2008). Similar to the protein synthesis independent
expression of E-LTP in electrophysiological recordings, protein synthesis blockers, anisomycin
and cycloheximide, had no effect on the early phase of spine enlargement induced by glutamate
uncaging (Tanaka et al., 2008).
In addition to spine enlargement, an increase in PSD size was also observed in the spines
immediately after potentiation (Bosch et al., 2014; Meyer et al., 2014). However, PSD markers,
PSD-95 and Homer1c, were not increased until much later (Bosch et al., 2014; Meyer et al.,
2014). Furthermore, Meyer et al., (2014) demonstrated that the persistence of spine enlargement
is positively correlated with the increase in PSD-95. In their study, they showed that following
spine enlargement, there was a gradual increase in Homer1c, which was later followed by an
increase in PSD-95 exclusively in the spines with sustained enlargements (Meyer et al., 2014).
At the time when PSD-95 started to show significant increase in some potentiated spines, the
spines that lacked this PSD-95 increase began to reduce in size, and this was then followed by
the reversal of Homer1c increase (Meyer et al., 2014). Their observations suggested that
although the increase in PSD size does not appear to be necessary for the early phase of
structural plasticity, it may be critical for the maintenance of activity-induced structural
enlargement (Meyer et al., 2014). These increases in spine size and PSD sizes were postulated to
facilitate LTP by providing more anchorage space for postsynaptic complexes and receptors
(Hayashi and Majewska, 2005). Conversely, reduction in spine size, elimination of existing
spines, and synaptic transmission depression were observed following LTD induction (Okamoto
et al., 2004; Zhou et al., 2004; Hayashi and Majewska, 2005; Bosch and Hayashi, 2012).
13
Overall, studies showed that the strengthening and weakening of synaptic connections is
accompanied by pre-existing spine enlargement and spinogenesis during LTP, and spine size
reduction and elimination during LTD (Engert and Bonhoeffer, 1999; Meyer et al., 2014). On the
other hand, using glutamate uncaging in combination with tetanic stimulation experiments,
Matsuzaki et al (2004) and Maletic-Savatic et al (1999) have shown that formation new of spines
occurs during L-LTP but not E-LTP, suggesting that morphological alterations of dendritic
spines may be important for early LTP (Matsuzaki et al., 2004).
1.5 Actin Dynamics and Regulation
1.5.1 Actin dynamics and their regulators
Filamentous actin (F-actin) is the major cytoskeletal component in dendritic spines and their
regulation has been shown to regulate spine morphology, motility, formation, and elimination
(Hotulainen and Hoogenraad, 2010). Actin cytoskeleton functions as an anchorage site for many
postsynaptic receptors (e.g. AMPAR and NMDAR at excitatory synapses and glycine receptors
at inhibitory synapses) by interacting with various scaffolding proteins, and its disassembly
would subsequently result in the loss of those receptors at the synapses (Cingolani and Goda,
2008; Hotulainen and Hoogenraad, 2010; Bosch and Hayashi, 2012). In addition to postsynaptic
receptor anchorage, actin also regulates receptor trafficking (Cingolani and Goda, 2008;
Hotulainen and Hoogenraad, 2010).
14
Actin in dendritic spines is highly dynamic. In fact, approximately 85% of actin are turned over
within 2 mins (Cingolani and Goda, 2008). At equilibrium, while actin-GTPs are continuously
being added to the filamentous actin (F-actin) at the barbed end (apical end), these F-actins are
constantly being depolymerized into their globular monomers (G-actin) at the pointed end (basal
end). This process is known as actin treadmilling, which enables a constant turnover of the actin
filaments while maintaining their lengths (reviewed in Cingolani and Goda, 2008; Honkura et al.,
2008).
At rest, two distinct pools of actin were observed in dendritic spines, which include a dynamic
pool with rapid actin turnover rate distributed throughout dendritic spines (turnover time constant
40+/-13 s) and a more stable pool localized towards the basal end of spines (time constant 17
mins) (Honkura et al., 2008). The size of the latter stable pool is strongly correlated with the
spine volume (Honkura et al., 2008). The dynamics of actin cytoskeleton were demonstrated to
be critical for spine morphogenesis as well as synaptic expression of postsynaptic proteins. For
an example, latrunculin A, a drug that is capable of increasing the actin depolymerization to
polymerization ratio, promotes AMPAR endocytosis (Cingolani and Goda, 2008). Furthermore,
cytochalasin-induced enhancement of actin depolymerization was demonstrated to reduce
NMDA receptor activity as well (Rosenmund and Westbrook, 1993). Conversely, jasplaskinolide,
a drug that promotes actin polymerization and stability, prevented AMPAR endocytosis
(Cingolani and Goda, 2008). In addition, Lat A application also reduced spine head size at rest,
suggesting that spine head size is maintained by continuous actin polymerization (Honkura et al.,
2008).
15
1.5.2 Commonly used pharmacological agents that affect actin
dynamics
A few pharmacological agents are used frequently to examine the role of actin dynamics in
various aspects of cellular activities. These include jasplakinolide, latrunculin A, and
cytochalasin D:
Jasplakinolide promotes actin stability and polymerization (Cingolani and Goda, 2008).
Latrunculin A sequesters G-actin to inhibit actin polymerization (Krucker et al., 2000).
Cytochalasin D competitively binds to the barbed end to prevent the addition of G-actin at this
end (Krucker et al., 2000).
1.5.3 Actin dynamics during synaptic plasticity
In addition to the dynamic and stable pools of actin at rest, glutamate uncaging induces the
formation of a third F-actin pool throughout the stimulated spine heads (τ = 2-15min) (Honkura
et al., 2008). It appears that the formation of this relatively stable pool is due to an increase in
actin polymerization as its formation is abolished by Lat A (Honkura et al., 2008). More
importantly, the persistence of glutamate-induced spine enlargement was demonstrated to
depend on CaMKII-activation dependent confinement of this newly formed F-actin pool
(Honkura et al., 2008). The contribution of CaMKII in the confinement of this F-actin pool is
likely due to its bundling activity, which would result in increased stability of the newly
reorganized F-actin cytoskeleton (Okamoto et al., 2007; Honkura et al., 2008; Okamoto et al.,
2009). Consistent with the idea that the enlargement of F-actin cytoskeleton is critical for
16
sustained structural plasticity, the size of the stable F-actin pool is proportional to the square of
the spine-head volume (Honkura et al., 2008). Furthermore, a similar increase in F-actin was also
demonstrated following high-frequency electrical stimulation. In free moving animals, high-
frequency stimulation (HFS) at the hippocampal medial perforant pathway induced both L-LTP
and increased F-actin expression at its postsynaptic region (medial one-third of the molecular
layer of the hippocampus) without altering total actin levels (F+G actin) (Fukazawa et al., 2003).
This selective increase in F-actin was observed immediately at 20 mins and persisted for weeks
after the HFS (Fukazawa et al., 2003). Furthermore, consistent with these observations of
increased F-actin following glutamate uncaging and HFS, a FRET study showed that the F/G
actin equilibrium in dendritic spines is activity dependent. More specifically, it was observed that
LTP induction shifts the actin equilibrium towards F-actin (increase began within ~20s and
peaked at ~1min) in an NMDAR-activation dependent manner (Okamoto et al., 2004).
Conversely, this actin equilibrium was demonstrated to shift toward G-actin during LTD
(Okamoto et al., 2004). This LTP induction protocol induced increase in F-actin is only observed
in the dendritic spines and not the surrounding dendritic shafts (Kramár et al., 2006; Chen et al.,
2007), and is associated with increased inactivation of cofilin, an actin depolymerization factor,
suggesting that a reduction in F-actin depolymerization may contribute to the spine-specific
increase in F-actin (Fukazawa et al., 2003; Chen et al., 2007).
In addition, this NMDAR-activation induced increase in actin polymerization to
depolymerization ratio is believed to provide the driving force for spine enlargement as well as
providing an increasing number of binding sites for postsynaptic proteins such as CaMKII and
glutamate receptors that may be necessary for maintaining the reorganized spine structures
17
(Okamoto et al., 2004; Hayashi and Majewska, 2005; Bosch and Hayashi, 2012). Honkura et al.
(2008) observed a synchronized ruffling of the spine and F-actin pool enlargement during spine
enlargement, supporting the hypothesis that an increase in F-actin may provide the driving force
for spine expansion. In contrast to the F-actin increase following LTP induction, the reduction in
F-actin promotes the disassembly of postsynaptic protein complexes, shrinkage of spine, and the
loss of anchorage for glutamate receptors following LTD induction (Hayashi and Majewska,
2005).
Pharmacological studies have consistently demonstrated that this actin rearrangement is crucial
for both LTP and LTD expression, and the administration of actin polymerization and
depolymerization disrupting agents interfere with the expression of both structural and functional
synaptic plasticity (Cingolani and Goda, 2008). For example, pharmacological interference (e.g.
latrucunlin A and cytochalasins) of actin polymerization and depolymerization were
demonstrated to block functional LTP (Kim and Lisman, 1999; Krucker et al., 2000; Fukazawa
et al., 2003; Lisman et al., 2012), activity-dependent AMPAR insertion (Gu et al., 2010), and
spine enlargement following uncaging (Matsuzaki et al., 2004). For an instance, at 20 nm, while
Lat A blocked the sustained phase (>5mins) of spine enlargement without alternating the
transient phase of spine enlargement (
18
of spine heads (Hayashi and Majewska, 2005; Gu et al., 2010; Lisman et al., 2012). Conversely,
following LTD induction, the reduction in F-actin promotes the disassembly of the postsynaptic
protein complexes, shrinkage of spine, and the loss of anchorage for glutamate receptors
(Okamoto et al., 2004; Zhou et al., 2004; Hayashi and Majewska, 2005). Moreover, similar to the
expression of E-LTP and glutamate uncaging induced spine enlargement, this increase in F-actin
is independent of protein synthesis (Fukazawa et al., 2003) (Figure 2).
1.5.4 Actin cytoskeleton regulation
Actin network is regulated by a variety of actin binding proteins (ABPs) such as cross-linking
proteins (α-actinin and CaMKII) that bundle the F-actin, capping proteins that restrict actin
filament elongation, and proteins that regulate actin polymerization and depolymerization rate
(gelsolin and cofilin) (reviewed by Cingolani and Goda, 2008). Similar to activity dependent
actin dynamics, expression of these ABPs at the synapses is also regulated by synaptic activity.
In a recent study, Bosch et al. (2014) demonstrated that glutamate uncaging induced spine
enlargement and synaptic transmission is accompanied by elevated synaptic expressions of actin
and many ABPs, including cofilin, Aip1, Arp2/3, GluA1, profilin, drebin, and CaMKII.
Consistent with the immediate increase in spine size and actin accumulation, the increase of
some ABPs occurred immediately after stimulation and persisted for at least 30 mins.
In the next few pages, the role of CaMKII-Rho family GTPases-cofilin signaling pathway in
actin regulation will be addressed in detail and a few of the other aforementioned ABPs will also
be briefly discussed.
19
Figure 2. Bidirectional regulation of actin cytoskeleton during synaptic plasticity (adapted
from Hayashi and Majewska, 2005)
Following HFS or glutamate uncaging (LTP induction protocols), there is an increase in F-actin
in dendritic spines due to an elevated actin polymerization to depolymerization rate. This
increase in F-actin promotes the insertion of AMPA receptors, an increase in actin binding
proteins (e.g. CaMKII), and spine enlargement. Conversely, following a low-frequency
stimulation (LTD induction protocol), the F/G actin equilibrium shifts towards the G-actin side
due to an increase in the ratio of actin depolymerization to polymerization. This elevated actin
depolymerization promotes the removal of AMPA receptors, the disassembly of postsynaptic
complexes, and spine size reduction.
20
1.5.4.1 CaMKII
At rest, CaMKII bundles F-actin via the β subunit, which stabilizes the actin cytoskeleton and
enables the maintenance of stable spine morphology (Okamoto et al., 2007). Once the synapse is
activated, calcium entrance induces an increase in Ca2+/calmodulin, which subsequently activates
CaMKII by binding to its regulatory domain to relieve it from its autoinhibitory domain that
normally masks its kinase domain (Okamoto et al., 2009). The initial activation of CaMKII
triggers its autophosphorylation, enabling sustained kinase activity (Okamoto et al., 2007;
Okamoto et al., 2009). The activation of CaMKII promotes its dissociation from F-actin and
allows actin to reorganize. In addition, the mobile CaMKII can phosphorylate other proteins such
as AMPA receptors, which increases its synaptic transmission (Okamoto et al., 2007; Okamoto
et al., 2009). This increase in CaMKII activation is restricted to activated spines with a decay
time constant of ~10s (Lee et al., 2009), and is both necessary and sufficient for the induction of
LTP (Okamoto et al., 2009). The activation of CaMKII was shown to persist for a few minutes
following LTP induction, once deactivated, they could re-bundle the reorganized actin
cytoskeleton and stabilize the newly modified spine structure (Okamoto et al., 2009).
In addition to actin bundling, CaMKII can regulate several pathways that are involved in the
regulation of actin dynamics. These pathways mostly converge on the Rho family small GTPases
(e.g. RhoA, Rac1, Cdc42). (Okamoto et al., 2009).
21
1.5.4.2 Rho-GTPase signalling in synaptic plasticity and actin
regulation
Many NMDAR-CaMKII signalling cascades converge on the Rho-family small GTPase
signalling cascades, which are among the key regulatory pathways of actin dynamics
(Hotulainen and Hoogenraad, 2010; Tolias et al., 2011).
These small GTPases alternate between an inactive GDP bound state and an active GTP bound
state (Tolias et al., 2011). Precise regulation of Rho family small GTPases is achieved by
guanine nucleotide exchange factors (GEFs), which activate them by catalyzing GDP/GTP
exchange, and GTPase-activating proteins (GAPs), which deactivate them by facilitating their
GTPase activity (Tolias et al., 2011).
The three most extensively studied Rho GTPases in spine and synaptic regulation include RhoA,
Cdc42, and Rac1.
Following glutamate uncaging, RhoA is activated within 30s (Murakoshi et al., 2011). However,
its activation is abolished by the co-administration of the NMDAR antagonist, APV, suggesting
that glutamate uncaging activates RhoA through NMDAR-dependent pathways (Murakoshi et al.,
2011). In addition, RhoA has consistently been demonstrated as a critical regulator of spine
morphogenesis and synaptic plasticity. For example, while constitutively active RhoA (caRhoA)
reduced spine size and density, the opposite was observed following RhoA knockdown (Impey et
22
al., 2010). Also, the downregulation of RhoA was shown to suppress both transient and sustained
spine enlargements. However, a stronger effect was observed during the transient phase
(Murakoshi et al., 2011). Furthermore, RhoA mediated cofilin inactivation was also
demonstrated to be necessary for LTP in electrophysiological recordings (Rex et al., 2009).
Consistent with the idea that actin cytoskeleton remodelling may be essential for long term
memory formation, RhoA inhibition impaired long-term fear memory, but showed no effect on
short-term fear memory (Lamprecht et al., 2002).
Similar to RhoA, Cdc42 could be selectively activated in the stimulated spines following
glutamate uncaging within 30s, and this activation was abolished by APV as well as KN62
(CaMKII antagonist), suggesting that Cdc42 is downstream of both NMDAR and CaMKII
(Murakoshi et al., 2011). However, unlike RhoA, which was demonstrated to have a stronger
role during the initial transient phase of spine enlargement, Cdc42 activity appears to be critical
for sustained dendritic spine structural plasticity (Murakoshi et al., 2011). For example, Cdc42
knockdown via shRNA was able to block sustained spine enlargement, but the transient phase of
spine enlargement was not altered by this manipulation.
The simultaneous binding of G-protein-coupled receptor kinase-interacting protein 1 (GIT1) and
PAK interacting exchanging factor (PIX) to Rac1 facilitate the translocation of Rac1 into
dendritic spines (Zhang et al., 2005). This activation of Rac1 enables its binding to the p-21
binding domain (PBD) of its effectors, such as PAKs (Zhang et al., 2005), and this process is
believed to be critical for de novo spinogenesis (Zhang et al., 2003; Zhang et al., 2005). For
23
instance, a dominant negative Rac1 mutation is able reduce spine density at rest as well as block
activity-induced spinogenesis (Nakayama et al., 2000; Impey et al., 2010).
Some of the identified GEFs and GAPs for the above Rho-family GTPases include Tiam1,
kalirin 7, βPIX, Lfc, and oligophrenin1 (Tolias et al., 2011).
Tiam 1, a Rac1 GEF, is expressed at both dendrites and spines (Tolias et al., 2005; Tolias et al.,
2011). It can be activated through NMDAR-CaMKII dependent phosphorylation, which
subsequently targets it to the synapses (Fleming et al., 1999; Tolias et al., 2011). The
downregulation of this GEF is able to block NMDAR activation induced spine morphogenesis
(Tolias et al., 2005).
Kalirin 7, another Rac-GEF, is highly enriched at the PSDs (Tolias et al., 2011). Overexpression
of this GEF resulted in an increase in spine density (Tolias et al., 2011). Conversely, knocking
down of kalirin 7 resulted in reduced basal spine density as well as activity-induced increase in
spine density (Ma et al., 2008; Impey et al., 2010; Tolias et al., 2011). However, kalirin 7
knockout (KO) mice exhibited normal spine morphogenesis during early development despite
the observed spine aberrations in adulthood, suggesting that kalirin 7 may be important for spine
maturation and maintenance (Ma et al., 2008; Cahill et al., 2009; Tolias et al., 2011).
Alternatively, this phenotype may also result from the low level of endogenous kalirin 7
expression during early development (Ma et al., 2008; Cahill et al., 2009; Tolias et al., 2011).
24
βPIX is a GEF for both Rac1 and Cdc42 (Tolias et al., 2011). By binding simultaneously to
GIT1, PAK, and Cdc42/or Rac1, it coordinates PAK activation by Cdc42 and Rac1 (Tolias et al.,
2011). The down-regulation of βPIX results in the attenuation of Rac1 activity as well as spine
density reduction (Zhang et al., 2005). In addition, maintaining a proper level of βPIX/GIT1 was
shown to be important for spine specific targeting of Rac1 activation as the overexpression of
βPIX resulted in the spreading of active Rac1 into dendritic shafts (Zhang et al., 2005).
Lfc is a Rho specific GEF that is preferentially localized in dendrites at rest (Ryan et al., 2005).
However, upon NMDAR activation, it is rapidly translocated into the spines where it may
activate RhoA and its subsequent regulation of actin dynamics (Ryan et al., 2005; Tolias et al.,
2011). Consistent with the effect of constitutively active RhoA on spine density and morphology,
the overexpression of this GEF is associated with reduced spine size, while its downregulation is
linked with increased spine density (Ryan et al., 2005; Impey et al., 2010; Tolias et al., 2011).
On the other hand, Oligophrenin1 (OPHN1) is a RhoA GAP and its activation is believed to
promote synaptic maturation through the inhibition of RhoA (Govek et al., 2004; Tolias et al.,
2011). The downregulation of OPHN1 is associated with reduced spine size and synaptic
efficacy (Nadif Kasri et al., 2009). Conversely, its overexpression is linked with increased spine
size and AMPAR mediated transmission (Nadif Kasri et al., 2009).
25
Considering their importance in actin regulation, it is not surprising that mutations in these Rho-
GTPases pathway signalling molecules are linked with aberrations of both spine morphogenesis
and synaptic transmission (Hotulainen and Hoogenraad, 2005; Tolias et al., 2011). Moreover,
studies reported that many mutations in these Rho-family GTPases signalling pathways are
linked with neuropathologies such as Alzheimer’s disease and mental retardation (Hotulainen
and Hoogenraad, 2010; Ba et al., 2013; DeGeer and Lamarche-Vane, 2013).
1.5.4.3 Cofilin
The effect of cofilin, an actin depolymerization factor, on actin dynamics is rather complex. At
physiological conditions, cofilin binds to actin at a ratio of 1:4 to 1:25, and it has a
concentration-dependent role in the regulation of actin dynamics in vitro (Bamburg and
Bernstein, 2010). Although cofilin persistently severs F-actin at low concentrations, at higher
concentrations (cofilin to actin 1:10-1:2), cofilin molecules bind to actin filaments and severe it
in a coordinated fashion to form twisted bundles (Andrianantoandro and Pollard, 2006; Bamburg
and Bernstein, 2010). These twisted bundles can be further depolymerized with the aid of Aip1,
promote actin polymerization by facilitating actin nucleation, or form rod-like bundles of F-actin,
which could block transport and cause the loss of synapses (Andrianantoandro and Pollard, 2006;
Bamburg and Bernstein, 2010).
The activity of cofilin is regulated by the phosphorylation of its serine 3 (S3) position (Bosch et
al., 2014). Phosphorylation of cofilin S3 by LIMK is believed to promote its dissociation from F-
26
actin, which subsequently enables other actin binding proteins (ABPs) to interact with F-actin
(Agnew et al., 1995; Fukazawa et al., 2003). Both high frequency stimulation and learning were
demonstrated to increase cofilin phosphorylation in the hippocampus (Fukazawa et al., 2003;
Chen et al., 2007; Okamoto et al., 2009). In addition, following glutamate uncaging, cofilin is
immediately translocated into stimulated spines (~20s) in a NMDAR and CaMKII activation
dependent manner (Bosch et al., 2014). Moreover, this cofilin translocation is accompanied by
an elevated proportion of F-actin bound cofilin at these sites, resulting in a reduced cofilin efflux
rate from these spines (Bosch et al., 2014). This net increase in cofilin results in a high
stoichiometric ratio of cofilin to actin binding, which may subsequently promote actin nucleation
and spine expansion (Andrianantoandro and Pollard, 2006; Bosch et al., 2014). Considering that
LTP expression is also accompanied by increased cofilin phosphorylation or inactivation, this
observation of increased F-actin bound non-phosphorylated cofilin following glutamate uncaging
was rather unexpected (Fukazawa et al., 2003; Chen et al., 2007). Bosch et al. (2014) suggested
that this paradox may result from the slightly delayed and transient inactivation pattern of cofilin.
More specifically, although p-cofilin level was observed to be elevated between 2-7 minutes post
HFS, it was not significantly altered at 0.5 min or 15-30 mins post stimulation, during which
cofilin would be able to bind to F-actin and promote actin nucleation (Chen et al., 2007).
Furthermore, both the inhibition of cofilin activity and the inhibition of cofilin inactivation were
observed to alter LTP expression and impaired learning/memory, suggesting an intact bi-
directional cofilin activity regulation is necessary for proper synaptic plasticity (Meng et al.,
2002; Fukazawa et al., 2003).
27
1.5.4.4 Other ABPs
In addition to the Rho GTPases-cofilin pathway, which is the focus of this study, there are
numerous other actin binding proteins that also regulate actin cytoskeleton stability and many of
which exhibit activity dependent expression. For instance, Aip1, which can be upregulated by
glutamate uncaging, binds cofilin-actin and facilitates cofilin’s actin severing activity (Bamburg
and Bernstein, 2010; Okreglak and Drubin, 2010). Another APB, Arp2/3, can bind to the side of
existing actin-filaments and nucleate new filament growth at a 70o angle to promote branching
(Chan et al., 2009; Bamburg and Bernstein, 2010; Hotulainen and Hoogenraad, 2010). This
branching provides many barbed ends where G-actin can be added and is critical for spine head
formation (Bamburg and Berstein, 2010; Hotulainen and Hoogenraad, 2010). Moreover,
reduction of Arp2/3 activators (cortacin, Wave-1, N-Wasp) was shown to alter spine morphology
and density (Hering and Sheng, 2003; Soderling et al., 2007; Wegner et al., 2008; Hotulainen
and Hoogenraad, 2010). On the other hand, profilin binds G-actin to facilitate actin
polymerization, and is suggested to stabilize spine structures during plasticity and fear
conditioning (Ackermann and Matus, 2003; Lamprecht et al., 2006; Cingolani and Goda, 2008;
Hotulainen and Hoogenraad, 2010). Lastly, similar to CaMKII, α-actinin also promotes actin
bundling and stabilization (Bamburg and Berstein, 2010).
Overall, studies have demonstrated that the dynamics of actin cytoskeleton is regulated by an
array of actin binding proteins, and its integrity is critical for both structural and functional
synaptic plasticity.
28
1.5.4.5 Current model of actin regulation during plasticity
Among all the tested proteins in Bosch et al.’s (2014) study, cofilin increased rapidly in the
stimulated spines within 20s post-stimulation, and was the only one that had its concentration
(cofilin/spine volume) remain elevated for at least 30 mins. In addition, actin, Aip1, and Arp2/3
concentration (amount proportional to spine volume) increased initially but returned to baseline
level as their expressions adjusted to the newly increased spine volume. In contrast,
concentrations of some other ABPs, such as CaMKII and α-actinin, showed a rapid reduction
following stimulation and gradually returned to baseline. Interestingly, the proteins that showed
initial reduction in Bosch et al.’s (2014) study are all known to promote actin cytoskeleton
stability. The current model of activity-dependent actin remodeling suggests that during the
initial phase of spine enlargement, there is a rapid increase in ABPs that are involved in
modifying actin cytoskeleton, however, most of these return to basal level during the sustained
phase of enlargement except cofilin (Bosch et al., 2014). In contrast, a transient reduction in
actin cytoskeleton stabilizers such as CaMKII and α-actinin also occur during this time window
(Bosch et al., 2014). The simultaneous increase in actin modifiers and reduction in actin
stabilizers during the initial phase allow a window for rapid actin modification (Okamoto et al.,
2014). This process is followed by actin and spine stabilization as reflected by the increase in
actin stabilizers such as actin bundling proteins CaMKII and α-actinin during sustained spine
enlargement (Okamoto et al. 2009; Bosch et al., 2014).
In summary, studies showed that the proper regulation of actin dynamics is essential for spine
density, morphology, and synaptic efficacy (Matus, 2000; Cingolani and Goda, 2008; Hotulainen
29
and Hoogenraad, 2010). It is believed that spine enlargement during LTP expression is likely
induced by NMDAR activation and cofilin-dependent actin reorganization (Cinolani and Goda,
2008). Because actin is the major cytoskeletal component in dendritic spines and it serves as the
binding site for many postsynaptic molecules, it is postulated that the stable increase in F-actin is
responsible for the observed spine enlargement as well as elevated levels of ABPs. Consistent
with this theory, pharmacological facilitation of actin depolymerization and inhibition of actin
polymerization blocked LTP induced spine enlargement, increase in ABPs, and strengthening of
functional synaptic efficacy (Bosch et al., 2014).
1.6 The Role of p-21 Activated Kinases in Synaptic
Properties
1.6.1 The two families of PAKs: structures and activation
mechanisms
p-21 activated kinases (PAKs) are a family of serine/threonine protein kinases that are involved
in actin cytoskeleton regulation (Hofmann et al., 2004; Arias-Romero and Chernoff, 2008). Six
PAKs have been identified to date and all of which are expressed in the central nervous system
(Boda et al., 2006; Arias-Romero and Chernoff, 2008). Studies show that all 6 PAKs possess a p-
21 binding domain (PBD) near the N-terminus, which is capable of binding Rho-family GTPases,
and a catalytic kinase domain near the C-terminus (Boda et al., 2006). These 6 PAKs are further
separated into 2 groups based on their structural features and regulatory mechanisms (Boda et al.,
30
2006; Arias-Romero and Chernoff, 2008). They are known as group I PAKs (PAK1, 2, 3) and
group II PAKs (PAK4, 5, 6) (Boda et al., 2006; Arias-Romero and Chernoff, 2008).
Group I PAKs exhibit high sequence homology within the group and are composed of PAK1,
PAK2, and PAK3 (Boda et al., 2006; Arias-Romero and Chernoff, 2008). All group I PAKS
possess an autoinhibitory domain (AID) that partially overlaps with the p-21 binding domain
(PBD) (Boda et al., 2006; Combeau et al., 2012). At rest, Group I PAKs dimerize to enable the
overlap between the PBD/AID of one monomer with the catalytic domain of another PAK
monomer (Boda et al., 2006; Combeau et al., 2012). This interaction between the AID and the
catalytic domain alters the conformation of the kinase domain, rendering it inactive (Lei et al.,
2000; Wang et al., 2011). Following synaptic activation, the binding of Rho-family small
GTPases (namely, Cdc42 and Rac1) at PBD leads to conformational changes that disrupt this
dimerization and releases the AID from the catalytic domain (Boda et al., 2006). This
dissociation further triggers a series of autophosphorylation steps, involving the phosphorylation
of Thr423, Ser144, Ser199, and Ser204, which eventually result in the manifestation of its full kinase
activity (Thompson et al., 1998; Arias-Romero and Chernoff, 2008).
Unlike group I PAKs, group II PAKs do not have an AID that blocks its own catalytic activity
(Arias-Romero and Chernoff, 2008). Although Cdc42 and Rac1 binding was also observed with
this group, Rho-family GTPases binding does not consistently lead to group II PAK activation
(Boda et al., 2006; Arias-Romero and Chernoff, 2008). Instead they were shown to lead to
translocation of group II PAKs (Boda et al., 2006; Arias-Romero and Chernoff, 2008).
31
1.6.2 The regulation of PAK1 and PAK3 activity
PAK1 and 3 are members of the group I PAKs and they were postulated to have mutually-
inhibitory, distinct, as well as overlapping functions in the nervous system.
As mentioned previously, the catalytic activity of group I PAKs is regulated by the formation of
dimers and their dissociation (Arias-Romero and Chernoff, 2008). A recent study by Combeau et
al. (2012) demonstrated that PAK1 and 3 can form homodimers as well as heterodimers in vitro.
By utilizing co-immunoprecipitation, they showed that while PAK3 prefers to form heterodimers
(7 fold greater prevalence) with PAK1 over homodimerization with itself, PAK1 displays equal
binding preference for PAK1 monomer and PAK3 monomer during dimerization (Combeau et
al., 2012). Furthermore, PAK1 and 3 were demonstrated to co-localize in spines and are both
expressed in PSDs (Combeau et al., 2012). Considering their tendency for heterodimerization in
solution and similar distribution pattern, endogenous PAK1 and 3 may be capable of forming
both hetero- and homo-dimers in vivo, which would subsequently result in their ability to inhibit
each other’s kinase activity at rest (Combeau et al., 2012).
In addition to their expression in dendritic spines, group I PAKs are also highly expressed in
dendrites (Meng et al., 2005; Kreis et al., 2007). In fact, dendritic shafts were shown to have
higher group I PAK expressions than spines at rest (Meng et al., 2005; Kreis et al., 2007) and
several mental retardation related mutations of PAKs were not shown to influence their basal
subcellular localization (Boda et al., 2006).
32
Furthermore, the activity and subcellular localization of PAKs have been demonstrated to be
regulated by synaptic plasticity. For instance, Chen et al. (2007) showed that LTP induction
increases p-PAK (the antibody identifies the phosphorylated form of all group I PAKs), the
active PAK, exclusively in spines without altering total PAK level (Chen et al., 2007). In
addition, p-PAK punctas were colocalized with PSD-95 markers suggesting that active PAKs are
specifically targeted to synaptic sites (Zhang et al., 2005). Consistent with the idea that LTP is
input-specific and only a small population of synapses would be responsive after stimulation, the
increase in p-PAK was only observed at 1.3% of the synapses following theta-burst stimulation
(TBS) (Chen et al., 2007). Consistent with this observation, glutamate uncaging selectively
recruits EGFP-PAK3 to activated spines while decreasing EGFP-PAK3 intensity in adjacent
dendrites, suggesting that synaptic activity facilitates the translocation of active PAKs into spines
from adjacent dendrites (Dubos et al., 2012). Similar to the NMDAR dependency of cofilin-
inactivation and dendritic spine F-actin accumulation following synaptic activity, APV is able to
block synaptic activity-induced PAK activation, suggesting that PAK activation requires
NMDAR just like cofilin-inactivation and F-actin accumulation (Chen et al., 2007).
1.6.3 The regulation of actin dynamics and synaptic properties
by PAK1 and PAK3
Despite their high sequence homology and overlapping expression, PAK 1 and 3 may be
differentially regulated (Arias-Romero and Chernoff, 2008; Kreis et al., 2007). More specifically,
a study showed that while PAK3 is preferentially activated by Cdc42 over Rac1, PAK1 is
equally activated by both Cdc42 and Rac1 (Kreis et al., 2007). Once activated, PAK1 and PAK3
33
are believed to phosphorylate LIM-kinase 1 (LIMK1), which subsequently phosphorylates
cofilin, promoting its dissociation from F-actin, slowing F-actin turn-over rate in dendritic spines
(Arber et al., 1998; Yang et al., 1998; Meng et al., 2002; Meng et al., 2005; Asrar et al., 2009;
Huang et al., 2011).
Group I PAKs have been demonstrated to be important for learning and memory as well as
synaptic plasticity. For instance, knocking out PAK1 and PAK3 separately was able to
selectively impair LTP, without altering basal synaptic transmission (e.g. NMDAR EPSCs,
AMPAR EPSCs, or NMDAR/AMPAR ratio), presynaptic function (paired-pulse-ratio), or LTD
(Meng et al., 2005; Asrar et al., 2009).
Because the proper regulation of dendritic spines is critical for both basal synaptic efficacy and
activity-dependent synaptic plasticity, numerous studies have utilized imaging techniques to
examine the role of PAKs in spine properties. Interestingly, the effects of PAK manipulations on
basal spine properties appear to be dependent on the methods of manipulation. For instance, both
PAK1 KO and PAK3 KO cultures display normal spine density and spine size at rest, consistent
with the observed normal basal synaptic transmission in electrophysiological recordings (Meng
et al., 2005; Asrar et al., 2009). In addition, another study demonstrated that knocking down
PAK1 and PAK3 by siRNA also did not alter spine density (Boda et al., 2006). However, these
findings are different from the reports based on dominant negative (dn) and kinase dead PAK
mutants. For example, studies showed that while dominant negative mutation of group I PAKs
reduced spine density as well as PSD-95 clusters, kinase dead mutations resulted in reduced
34
spine density, reduced PSD size, and exhibited elongated immature spine morphology,
suggesting that PAKs may be important regulators for either spine formation or maintenance as
well as excitatory synapse formation (Boda et al., 2004; Zhang et al., 2005; Boda et al., 2006;
Impey et al., 2010). In another study, dnPAK reduced spine density but increased the proportion
of larger spines on cortical neurons which was accompanied by the strengthening of synaptic
efficacy in the cortex (Hayashi et al., 2004). However, in the same study, dnPAK did not affect
spine density or basal synaptic properties in the hippocampus (Hayashi et al., 2004). These
discrepancies may result from several reasons: 1) there might be some intrinsic differences
between cortex and hippocampus, 2) dnPAK manipulation also suppresses the expression of
other group I PAKs, and 3) the differences observed in kinase dead and KO mutants may arise
from the fact that PAKs have both kinase dependent and independent functions which are both
suppressed in PAK KO but only kinase function is blocked in kinase dead mutants (Kreis et al.,
2007). Conversely, constitutively active PAK1 and PAK3 mutations resulted in increased spine
and protrusion density as well as synaptic PSD-95 expression, which are consistent of the effects
of constitutively active Rac1 mutation. (Zhang et al., 2005; Boda et al., 2006)
As mentioned previously, F-actin is preferentially localized at dendritic spines, and activity-
induced F-actin accumulation accompanied by an increase in p-cofilin/cofilin ratio in dendritic
spines is critical for both structural and functional synaptic plasticity. In PAK1 KO, Asrar et al.
(2009) observed a reduction of basal F-actin accumulation in spines (reduced ratio of
spine/dendrite phallodin intensity, an F-actin marker) as well as an impaired activity-induced
cofilin inactivation. In fact, consistent with the attenuated LTP in PAK1 KO hippocampal slices,
the HFS induced increase in p-cofilin/cofilin ratio was completely abolished in PAK1KO (Asrar
35
et al., 2009). In comparison, a normal p-cofilin/cofilin ratio was observed at basal conditions in
these mutants, suggesting that deletion of PAK1 selectively impaired activity-induced cofilin
inactivation (Asrar et al., 2009). In contrast, neither F-actin distribution nor p-cofilin/cofilin ratio
were altered by PAK3 KO, however, it is unclear if activity dependent cofilin regulation is
altered by this mutation (Meng et al., 2005). Similar to PAKs, their downstream target LIMK1
also exhibited reduced basal p-cofiln/cofilin ratio and activity-induced cofilin inactivation which
is associated with the reduced F-actin accumulation in dendritic spines (Meng et al., 2002). In
addition to cofilin, PAK1 is also known to regulate actin dynamics by regulating myosin
regulatory light chin kinase (MLCK), however, no difference in MLCK was observed in
PAK1KO (Zhang et al., 2005; Asrar et al., 2009). Furthermore, other PAK downstream effectors
that were previously demonstrated to be involved in synaptic plasticity, such as molecules in the
MAPK signaling cascades, were also not altered by PAK1 and PAK3 KO mutations (Meng et al.,
2005; Asrar et al., 2009).
Although neither PAK1 nor PAK3 KO exhibited deficits in hippocampus-dependent fear
memory, PAK1/3 double KO (DKO) exhibited complete abolishment of fear memory and
profound impairment in spatial memory, suggesting that PAK1 and 3 may be functionally
redundant, at least partially, in hippocampus-dependent memory formation (Meng et al., 2005;
Huang et al., 2011). In addition, PAK1/3 DKO also have reduced proportion of mushroom-like
spines and displayed attenuated LTP and LTD (Huang et al., 2011). Furthermore, unlike PAK1
KO and PAK3 KO, PAK1/3 DKO reduced basal p-cofilin/cofilin ratio (no change in total
cofilin), again, suggesting that PAK1 and 3 may compensate each other in their regulation of
basal cofilin activity (Huang et al., 2011). Consistent with this idea, caPAK1 cotransfection was
36
able to rescue the abnormally elongated spine phenotype in a kinase dead PAK3 mutant, further
suggesting that PAK1 and 3 may have overlapping functions in spine morphogenesis (Boda et al.,
2008). In addition, the decrease in basal p-cofilin indicates an enhanced cofilin activity in
PAK1/3 DKO, which may result in increased basal spine dynamics. Consistent with the idea that
cofilin activity is crucial for spine morphogenesis, blockade of cofilin activity rescued the
observed spine phenotype in DKO at rest (Huang et al., 2011). However, it is unknown if
activity-dependent cofilin activity is altered in these DKO.
In summary, previous studies demonstrated that PAK1 and 3 are important regulators of learning
and memory (Meng et al., 2005; Asrar et al., 2009; Huang et al., 2011). PAK1 and PAK3 were
postulated to have distinct as well as overlapping functions in the regulation of synaptic plasticity
and basal spine properties.
1.6.4 Other PAKs
The other p21-activated kinases are not going to be the focus of this thesis because they are
either embryonically/developmentally lethal or do not exhibit significant neurological
impairments.
PAK2 is ubiquitously expressed. PAK2 KO animals are not viable and no spine abnormalities
have been observed in these animals (Hofmann et al., 2004; Boda et al., 2006).
37
PAK4, on the other hand, is known to interact with Cdc42 and is essential for cytoskeletal
reorganization (Abo et al., 1998; Arias-Romero and Chernoff, 2008). The null mutation is
embryonically lethal due to an unknown mechanism involving both cardiac and neuronal defects
(Qu et al., 2003).
PAK5 null mutants are viable and fertile, however, no profound adverse effects in the nervous
system have been observed in vivo (Boda et al., 2006).
PAK6, on the other hand, has few known functions (Boda et al., 2006).
38
Chapter 2
Rationale and Hypothesis
Synaptic plasticity, which is the activity-dependent alteration of synaptic efficacy between
neurons, is believed to be necessary for learning and memory (Hotulainen and Hoogenraad, 2010;
Ho et al., 2011). Long-term potentiation (LTP) is a highly input-specific process, meaning that
the potentiation of synaptic efficacy only occur at the activated synapses (Frey and Morris, 1997;
Fukazawa et al., 2003). LTP can be classified into two types on the basis of their temporal
persistence and molecular mechanisms. These two phases are early protein synthesis independent
phase (E-LTP) that persists less than a few hours and late protein-synthesis dependent
maintenance phase (L-LTP) that could persists days, weeks, or even years (Frey and Morris,
1997, 1998; Ramachandran and Frey, 2009). During late phase LTP, synaptic plasticity related
proteins, which can be produced in the nearby dendrites as well as the distant soma, are required
to be captured by the stimulated synapses (Frey and Morris, 1997, 1998; Ramachandran and Frey,
2009). To achieve the proper long-term wiring of neuronal networks, a process that involves
input-specific and protein-synthesis-independent synaptic tag setting during the early LTP phase
is believed to be essential (Frey and Morris, 1997, 1998; Ramachandran and Frey, 2009).
However, it remains unclear how the input-specific early synaptic tagging is established. To
qualify as a synaptic tag, Martin and Kosik (2002) proposed that the candidate must display
highly input-specific local activation following stimulation and to persist long enough to enable
other PRPs to interact with it. Recent studies proposed several such candidates, which include
39
CaMKII and F-actin (Okamoto et al., 2007; Honkura et al., 2008; Zhang et al., 2008; Okamoto et
al., 2009; Ramachandran and Frey, 2009).
Dendritic spines are actin-rich protrusions where most excitatory synapses are formed (Fiala et
al., 1998; Hayashi and Majewska, 2005; Shrestha et al., 2006). They are believed to serve as
compartmentalization units for signalling molecules during synaptic activity and their aberrations
are associated with many neuropathologies such as Down syndrome, mental retardation, and
Alzheimer’s diseases (Fiala et al., 2002; Shrestha et al., 2006). Dendritic spines are highly
dynamic structures and their size, morphology, and density were demonstrated to be activity
dependent. For example, while LTP induction was shown to induce rapid spine head
enlargement (Honkura et al., 2008; Tanaka et al., 2009; Okamoto et al., 2009), enabling higher
expression of AMPAR receptors at the postsynaptic membrane (Matsuzaki et al., 2001; Bourne
and Harris, 2008; Okamoto et al., 2009), the opposite was observed following an LTD induction
(Zhou et al., 2004; Bourne and Harris, 2008; Okamoto et al., 2009).
Actin, the major cytoskeleton component in dendritic spines, was demonstrated to be responsible
for the regulation of spine formation, elimination, and morphogenesis (Matus, 2000; Okamoto et
al., 2004; Okamoto et al., 2007; Cingolani and Goda, 2008). Actin at the dendritic spine goes
through continuous treadmilling, meaning that while it is continuously being polymerized to
form filamentous (F-) actin at the apical end, it is continuously being depolymerized into their
globular (G-) monomers at the basal end (Honkura et al., 2008). In addition, previous studies
have demonstrated that shifts in this F/G actin equilibrium can occur in an activity dependent
40
manner and interference with these activity dependent shifts in actin equilibrium can result in the
alteration of both structural and functional synaptic plasticity. More specifically, studies
consistently demonstrated that while LTP shifts this equilibrium towards F-actin, LTD shifts this
equilibrium towards G-actin (Okamoto et al., 2004; Chen et al., 2007; Honkura et al., 2008). In
addition, this activity-induced F-actin accumulation was observed exclusively at stimulated
synapses (Okamoto et al., 2004; Okamoto et al., 2009; Honkura et al., 2008; Bosch et al., 2014).
In addition, an increase in F-actin persists long after the stimulation and it is capable of binding
various postsynaptic molecules, making it a good candidate for synaptic tag (Okamoto et al.,
2004; Okamoto et al., 2009; Honkura et al., 2008; Bosch et al., 2014). Furthermore, blockade of
this activity-induced F-actin accumulation is capable of blocking both glutamate induced spine
enlargement as well as functional LTP (Kim and Lisman, 1999; Krucker et al., 2000; Fukazawa
et al., 2003; Cingolani and Goda, 2008; Honkura et al., 2008).
Actin dynamics can be regulated by various molecules. One of the major regulators of this actin
dynamics is the actin depolymerization factor, cofilin (Cingolani and Goda, 2008). P-21
activated kinases are a family of serine/threonine kinases. PAK1 and PAK3, belonging to group I
PAKs, are downstream targets of the Rho-family small GTPases, Rac1 and Cdc42, and their
aberrations are linked with non-syndromic mental retardation in patients as well as severe
cognitive deficits in rodent models (Arias-Romero and Chernoff, 2008; Huang et al., 2011)
(Figure 3). Once activated, PAK1 and PAK3 are believed to phosphorylate LIM kinase 1
(LIMK1), which in turn transiently phosphorylate the actin depolymerization factor, cofilin,
promoting its dissociation from F-actin, slowing the F-actin turn-over rate at the spines (Arber et
al., 1998; Yang et al., 1998; Meng et al., 2002; Meng et al., 2005; Asrar et al., 2009) (Figure 3).
41
A transient increase in p-cofilin/cofilin ratio (cofilin inactivation) occurs at ~2 mins post-
stimulation and persists less than 15 mins post-stimulation (Chen et al., 2007). Furthermore,
activity-induced cofilin translocation into the spine and its transient inactivation are believed to
be critical for the expression of LTP (Chen et al., 2007; Bosch et al., 2014) (Figure 3). Previously,
our lab demonstrated impaired synaptic plasticity in mutants with PAK1 KO (Asrar et al., 2009),
PAK3 KO (Meng et al., 2005), as well as PAK1/3 DKO (Huang et al., 2011). In addition,
activity-induced cofilin inactivation was abolished in PAK deletion mutants (Asrar et al., 2009).
Our lab also demonstrated a complete abolishment of fear and spatial memory in PAK1/3 DKO
(Huang et al., 2011). Despite the importance of PAK signalling in synaptic plasticity and
learning and memory, its roles in the regulation of dendritic spine structural plasticity remain
unclear.
Research Objectives and Hypothesis:
Because PAKs are key regulators of activity-dependent cofilin inactivation and that synapse-
specific accumulation actin and cofilin are critical for spine morphogenesis and structural
plasticity, I hypothesized that PAKs are crucial for the regulation of basal spine morphology and
structural plasticity. More specifically, considering that F-actin and cofilin dynamics are
essential regulators of spinogenesis, PAK1/3 deletion(s) are expected to alter spine density in the
mutants. In addition, given that stable F-actin expression is strongly correlated with spine head
size, and that F-actin is reduced in PAK1 KO but not in PAK3 KO, I hypothesized that spine
head size would be selectively reduced in PAK1 KO. Also, because PAKs are believed to be
critical regulators of cofilin activity, PAK deletion mutants were expected to have impaired
structural plasticity. Lastly, because activity-induced synapse specific cofilin trafficking and
42
inactivation as well as F-actin accumulation were suggested to be important synaptic tagging
mechanisms, whether PAK deletion would alter input specificity was also examined.
43
44
Figure 3. Activity-dependent Rho-GTPases-PAK-cofilin regulation in dendritic spines
Immediately after synaptic activation, non-phosphorylated cofilin is translocated into the spines
from the dendrites (~20 s) (Bosch et al., 2014). Initially, this increased cofilin likely promotes
actin depolymerization in dendritic spines (Bamburg and Berstein, 2010). However, as the
cofilin to actin ratio continues to increase, cofilin switches its function to promote actin
nucleation (Bamburg and Bernstein, 2010). This increase in cofilin activity in the spine is
followed by a period of increased cofilin phosphorylation or inactivation (Chen et al., 2007).
This activity-dependent cofilin-actin dynamics is believed to be regulated by the NMDAR-
CaMKII-RhoGTPases pathway. Upon activation, an influx of Ca2+ promotes the phosphorylation
of CaMKII in the activated spines and releases CaMKII from F-actin (Okamoto et al., 2007;
Okamoto et al., 2009). This p-CaMKII activates GEFs, which subsequently activates Rac1 and
Cdc42. PAK1 and PAK3 are downstream effectors of Rac1 and Cdc42 (Kreis et al., 2007). Once
activated, PAK1 and PAK3 phosphorylates LIMK1, which in turn phosphorylates cofilin,
promoting its dissociation from F-actin (Arber et al., 1998; Yang et al., 1998; Meng et al., 2002;
Meng et al., 2005; Asrar et al., 2009). In addition, this phosphorylation of cofilin is necessary for
the retention of cofilin in the activated spines (Bosch et al., 2014).
45
Chapter 3
Materials and Methods
3.1 Genotyping
3.1.1 DNA extraction
Genotypes of mice utilized in this study were determined prior to experiments. Genomic DNA
was extracted by obtaining mouse tails ~0.5 cm. Tails were lysed in 500 µl proteinase K (PK)
tail buffer (50mM Tris-HCl pH 8, 100mM EDTA, 100mM NaCl, and 1%SDS and 10 µl PK)
andincubated overnight at 60oC.
The digested tails were retrieved from the incubator. Genomic DNA was then precipitated by
incubating the lysate with NaCl and chloroform. More specifically, 190 µl 5M NaCl and 600 µl
chloroform were added to each digested tail at room temperature. The mixture was centrifuged at
10,000 rpm for 10 mins and ~400 µl of the top layer containing genomic DNA, was transferred
to an Eppendorf tube. Equal volume of 100% ethanol was added into each tube that contains the
DNA solution. Each tube was mixed vigorously until DNA precipitation can be observed. The
mixture was then centrifuged at 14,000rpm for 5mins. Supernatant was discarded and pellet was
w