University of Southern California Dissertations and Theses

Sensory learning and anatomical plasticity in barrel cortex

(USC Thesis Other)

Sensory learning and anatomical plasticity in barrel cortex

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

SENSORY LEARNING AND ANATOMICAL PLASTICITY IN BARREL CORTEX

by

Jennifer Park

______________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(NEUROSCIENCE)

May 2016

DEDICATION

Dedicated to my family and friends, who provided endless encouragement

and support throughout my graduate studies.

ii
TABLE OF CONTENTS

DEDICATION....................................................................................................................ii

LIST OF FIGURES........................................................................................................... v

CHAPTER ONE: Introduction......................................................................................1-11

Introduction to barrel cortex................................................................................... 1
The role of sensory experience in circuit refinement.............................................. 4
Introduction to the gap crossing task..................................................................... 6
Introduction to NgR1.............................................................................................. 9
Presented contributions........................................................................................11

CHAPTER TWO: Nogo receptor 1 limits tactile tactile task performance
independent of basal anatomical plasticity................................................................12-36

Abstract................................................................................................................ 12
Introduction.......................................................................................................... 13
Materials and methods......................................................................................... 15
Results................................................................................................................. 22
Discussion............................................................................................................ 33

CHAPTER THREE: Dendritic spine stability, but not turnover, is enhanced in
barrel cortex during acquisition of a perceptual learning task....................................37-62

Abstract................................................................................................................ 37
Introduction.......................................................................................................... 38
Materials and methods......................................................................................... 40
Results................................................................................................................. 47
Discussion............................................................................................................ 59

CHAPTER FOUR: Conclusions.................................................................................63-64

Conclusions..........................................................................................................63

iii

REFERENCES..........................................................................................................65-70

APPENDIX: Autophosphorylation at the aCaMKII Threonin-286 site is essential for
sensory-based learning............................................................................................71-84

Abstract................................................................................................................ 71
Introduction...........................................................................................................72
Materials and methods......................................................................................... 74
Results................................................................................................................. 78
Discussion............................................................................................................ 81
References........................................................................................................... 83

iv
LIST OF FIGURES

1.1. Barrel cortex is a specialized subfield of primary somatosensory cortex 3

1.2. The gap cross assay serves as a tactile learning paradigm 8

2.1 Mice lacking ngr1 perform better on the gap cross assay but display
normal tactile learning across sessions 24

2.2 Cranial windows allow for long-term, in vivo imaging 26

2.3 Dendritic spines remain dynamic throughout life 27

2.4 Dendritic spine turnover and stability are normal in ngr1-/- mice 30

2.5 The acute deletion of ngr1 does not affect basal cortical spine dynamics 32

3.1 Mice display improved performance on the gap crossing task with
increased number of trials 48

3.2 Learning a whisker-dependent task increases the stabilization of newly
formed spines in barrel cortex 50

3.3 Mice with better performance on the gap crossing task display higher
rates of new spine survival 52

3.4 Although most dendritic spines are stable in adulthood, a small
population of spines continues to appear and disappear 54

3.5 Learning on the gap crossing task does not dramatically impact rates
of spine formation or elimination in sensory cortex 55

3.6 Learning on the gap crossing task does not increase the formation of
dendritic spine clusters in barrel cortex 58

A.1 T286A mutant mice display impaired performance on the gap crossing
task 80

v

CHAPTER ONE
Introduction

Introduction to barrel cortex

The barrel cortex in rodents is a specialized region of primary somatosensory
cortex that receives and processes haptic information gathered by large whiskers found
on the snout (Fig. 1.1A). This area receives its name from the distinctive pattern of
discrete barrels located in cortical layer 4 (Fig. 1.1A). Large facial whiskers are
organized into rows (A-E) and columns (1-5) and this pattern is replicated in barrel
cortex (Fig. 1.1B). Each so-called barrel corresponds to an individual whisker found on
the snout and maintains its neighbor relations, resulting in a topographic representation
(Woosley and Van Der Loos, 1970). The barrel field occupies a remarkable portion of
cortical real estate, representing ~13% of the total cortical surface area and close to
70% of the primary somatosensory cortex (Lee and Erzurumlu, 2005). This highlights
both the high level of innervation at the whisker follicle as well as the importance of this
sensory information.
Rodents utilize their whiskers to gather tactile information about their
environment, much in the way that humans use touch receptors in their fingers.
Manipulating sensory experience to examine sensory perception and measure
corresponding anatomical or physiological changes will help us better understand
cortical function. Using sensory systems is especially advantageous because
1

experimenters can control the specific sensory information that the animal receives. For
example, studying experience dependent changes in a spared whisker protocol, where
all but a single whisker is removed, provides insight on the neocortical changes that
occur with digit amputation (Fox 2008). Taking advantage of the topographic map in
barrel cortex, one can examine changes in the barrel corresponding to the remaining
whisker as well as changes occurring in surrounding barrels that are now devoid of
sensory input. Rodents, monkeys, and humans all display an expansion of the spared
sensory organ’s representation into the deprived region (Hand, 1982; Kaas et al., 1983;
Fox 1992), suggesting that plasticity mechanisms are conserved across mammals.
Thus the study of sensory systems in rodents provides insight into sensory processing
and plasticity in humans.

2

A
B

Figure 1.1. Barrel cortex is a
specialized subfield of primary
somatosensory cortex. A, Sensory
cortices V1 (primary visual cortex), A1
(primary auditory cortex), S1 (primary
somatosensory cortex), and S2
(secondary somatosensory cortex) can
be visualized in flattened mouse cortex
sectioned through layer 4. S1 shows
representations for the HL (hindlimb),
FL (forelimb), T (trunk), and whiskers
on the LL (lower lip) and the SW (small
whiskers). The large facial whiskers
are seen in barrel cortex (boxed
region). B, Barrels are organized into
rows (A-E) and columns (1-5) and
maintain a topographic relationship
with the whiskers.

3

The role of sensory experience in circuit refinement

The neocortex has the ability to change with sensory experience, which plays
a fundamental role in learning and memory. It is now widely accepted that plasticity
occurs into adulthood, although experience-dependent changes are much more robust
at younger ages, especially during critical periods (Keuroghlian and Knudsen, 2007;
Feldman, 2009; Hofer et al., 2006). During development, the sensory cortices use
environmental stimuli to help fine-tune and sculpt neural circuits. Although adults
maintain the ability to undergo plastic changes with alterations in their sensory
environment, these changes are often smaller, more restricted, and harder to induce
(Keuroghlian and Knudsen, 2007; Feldman, 2009; Hofer et al., 2006).
Structural changes though the formation and elimination of synapse provide a
mechanism for neural circuit plasticity. Abnormal sensory experience during
development can result in large-scale reorganization of axonal projections and
associated synaptic connectivity within the brain (Antonini and Stryker, 1993; Tailby et
al., 2005). In the adult, however, sensory experience impacts the formation and stability
of synaptic contacts formed between otherwise static dendritic arbors and nearby axons
(Trachtenberg et al, 2002; Mizrahi and Katz, 2003). New dendritic spines appear and
disappear over days (Grutzendler et al., 2002; Trachtenberg et al., 2002). New
persistent spines lasting more than 8 days always bear synapses (Knott et al., 2006)
and are likely to be maintained for a month or longer (Holtmaat et al., 2005). New
synapses are established by a dendritic spine growing out to form a connection with an
4

existing axon (Knott et al., 2006). These new synapses become functional within days
(Knott et al., 2006; Zito et al., 2009). Therefore, the formation of new synapses involves
the growth of new, persistent spines that form functional, synaptic connections, allowing
for cortical rewiring. How synapses are eliminated is not fully understood, but it does
require NMDA receptor activation (Zuo et al., 2005b).
Dendritic spines are the postsynaptic partner for the vast majority of fast,
excitatory synapses in cortex. Glutamate serves as the major excitatory
neurotransmitter in the central nervous system and its receptors are found throughout
the mammalian brain (Hollman and Heinemann, 1994). Ionotropic glutamate receptors
mediate rapid excitatory neurotransmission in the central nervous system. This family of
receptors is further classified into NMDA (Nmethyl-D-aspartate), AMPA (alpha-amino-3-
hydroxy-5-methyl-4-isoxazole-propionic acid), and kainate receptors, named after their
most selective agonist (Wisden and Seeburg, 1993).
Though particularly motile during development, dendritic spines remain
dynamic throughout adulthood. Repeated in vivo imaging of dendritic spines in mice
expressing genetically encoded fluorescent proteins has revealed that most dendritic
spines are stable and persist for months (Zuo et al., 2005; Grutzendler et al., 2002;
Holtmaat et al., 2005). However, a subpopulation of dendritic spines appear and
disappear throughout life. Importantly, the rate of spine turnover, density, and
distribution responds to a number of factors including sensory perturbation
(Trachtenberg et al., 2002; Holtmaat et al., 2006; Yang et al., 2009; Jung and Herms,
2014) and learning (Yang et al., 2009; Xu et al., 2009; Fu et al., 2012; Akbik et al., 2013).
5

Thus, the formation and elimination of dendritic spines has been proposed to contribute
to the alterations in brain circuitry subserving learning and other forms of plasticity.

Introduction to the gap crossing task

The whisker system of rodents is an excellent model for studying anatomical
plasticity, behavior, and the interplay between the two. Because whiskers are easily
accessible for sensory manipulation and each whisker is mapped onto cortex in a
topographic manner, this system has been a favored model for studying experience-
dependent plasticity. Studies examining experience-dependent changes have shed light
on the plasticity mechanisms involved; because sensory processing is required for and
leads to sensory learning, such mechanism may also be crucial to learning and memory
(Fox 2008; Gilbert et al., 2001).
Although manipulating sensory input is fairly straightforward, modeling
sensory learning in the tactile system is more difficult. To address this, we utilize the gap
crossing paradigm, a whisker input- and barrel cortex-dependent learning task (Hutson
and Masterton, 1986). In this learning paradigm, a mouse is placed on an elevated
home platform and sweeps its whiskers across an experimenter-defined gap distance to
detect the presence of a target platform (Fig. 1.2A). It serves as both a distance
detection and object localization task (Hutson and Masterton, 1986; Jenkinson and
Glickstein, 2000; Harris et al., 1999; Celikel et al., 2007; Voigts et al., 2008) in which
6

mice are able to freely explore a controlled environment while performing an unbaited
task. The mouse’s movements are tracked using motion sensors (Fig. 1.2A), and the
pattern of movements determines whether the mouse successfully crossed the gap or
not in a given trial. Performance at each presented gap distance is displayed in real
time; gap distances are presented in a pseudo-random fashion to encourage learning
while preventing any chance at predicting the next distance based on prior performance
(Fig 1.2B).
Presented gap distances are categorized into two distinct groups: nose
distances and whisker distances. At shorter nose distances (3.0, 3.5, and 4.0 cm), mice
are able to use haptic sensory input from both their whiskers and touch receptors in
their nose to explore the target platform (Fig. 1.2C). At longer whisker distances (5.0,
5.5, and 6.0 cm), mice must rely solely on tactile information gathered by their whiskers,
as the target platform is too far away for their noses to reach (Fig. 1.2C). Importantly,
the successful crossing of mice at whisker distances improves with increased
experience.
This task is performed on a custom-built, automated robot providing a closed-
loop environment (D.H. Herman, manuscript in preparation). The automated nature of
this behavioral task reduces handler-related effects on performance. Contrary to other
sensory manipulations, such as enriched environment, this task is desirable because of
its quantifiable nature.

7

#1 #2 #3 #4
Motion sensors
Home platform Target platform
3
4
5
6
7
x
x
x
x
x
x
x
x x x
10 20
Gap distance (cm)
Trial
x
success
failure
x
4cm
5cm
6cm
Figure 1.2. The gap cross
assay serves as a tactile
learning paradigm. (A) A
schematic of the gap cross
apparatus. The mouse is placed
on the home platform and must
cross a variable gap distance to
reach the target platform. (B)
An example session, where
successful crosses are marked
by a green dot and failed
crosses are notated by a red X.
(C) Mice are presented with
both nose (<4 cm, dashed line)
and whisker distances (>5 cm,
solid lines).
A
B
C
8

Introduction to NgR1

Nogo-66 receptor 1 is a glycophosphatidylinosital (GPI)-linked membrane protein
that is abundantly expressed in projection neurons in the adult cortex and hippocampus
(Barrette et al., 2007). It was first discovered to have high affinity binding for Nogo-66, a
growth cone collapse domain of Nogo-A (Fournier et al., 2001). Since then, NgR1 has
been shown to bind myelin associated glycoprotein (MAG) and oligodendrocyte myelin
glycoprotein (OMgp) (Liu et al., 2002; Wang et al., 2002). All three ligands act as
myelin-associated inhibitors (MAIs), limiting axon regeneration and sprouting after injury.
Because NgR1 is localized to axonal membranes (especially growth cones) (Wang et al.,
2002; Venkatesh et al., 2005), dendrites, and post-synaptic compartments (Lee et al.,
2008), it has been suggested that NgR1 further acts to restrict synaptic plasticity in the
mature central nervous system (Borrie et al., 2012; Schwab and Strittmatter 2014).
The downregulation of NgR1 in an activity-dependent manner suggests it plays a
role in plasticity, possibly by allowing for changes in connectivity. For example, rats on a
running wheel exhibit a transient decrease in NgR1 expression in hippocampus and
cortex (Josophson et al., 2003). Overexpression of NgR1 impairs the formation of
lasting or long-term memories (Karlen et al., 2009), which implies that activity-induced
downregulation is required for memory formation. Knocking out NgR1 extends ocular
dominance critical period plasticity into adulthood (McGee et al., 2005) and enhances
improvement on the rotarod in adult mice (Akbik et al., 2013), further suggesting that
NgR1 limits plasticity in the mature nervous system.
9

Structural analysis indicates that loss of NgR1 induces morphological changes in
dendritic spines. Various staining and imaging methodologies reveal no deformities at
the gross anatomical level (Kim et al., 2004), with no changes in dendritic orientation or
gross neuronal architecture in hippocampus or neocortex (Lee et al., 2008). However,
NgR1 mutants have more stubby spines at the expense of mushroom spines and (to a
lesser degree) thin spines in hippocampus (Lee et al., 2008; Zagrebelsky et al., 2010).
Their dendritic profile resembles that of more immature rodents (Harris et al., 1992) and
ngr1-/- mice display increased cortical spine dynamics (Akbik et al., 2013), suggesting
that NgR1 plays a role in the maturation or stabilization of spines.

10

Presented contributions

The formation, elimination, and stabilization of synaptic structures provide a
mechanism for functional rewiring of neural circuitry. However, the genes regulating
these changes in synaptic connectivity are not well understood. One gene of interest is
ngr1 as it is thought to limit anatomical plasticity in the mature nervous system. As
NgR1 has been reported to restrict motor learning and dendritic spine turnover, I
examined whether the loss of ngr1 influenced improvement on a sensory-based
learning paradigm or basal spine dynamics in sensory cortex (Chapter 2). Studies
investigating the anatomical correlates of motor learning demonstrate substantial
changes in dendritic spine dynamics in corresponding motor cortex. In order to
investigate the structural changes that occur during tactile learning in wildtype mice, I
imaged the same population of dendritic spines before, during, and after training on the
gap crossing task (Chapter 3). To date, this is the first report examining structural
plasticity of layer 5 dendritic spines throughout the acquisition of a tactile task.

11
CHAPTER TWO
Nogo receptor 1 limits tactile task performance independent of basal anatomical
plasticity

Abstract

The genes that govern how experience refines neural circuitry and alters synaptic
structural plasticity are poorly understood. The nogo-66 receptor 1 gene (ngr1) is one
candidate that may restrict the rate of learning as well as basal anatomical plasticity in
adult cerebral cortex. To investigate if ngr1 limits the rate of learning, we tested adult
ngr1 null mice on a tactile learning task. Ngr1 mutants display greater overall
performance despite a normal rate of improvement on the gap-cross assay, a whisker-
dependent learning paradigm. To determine if ngr1 restricts basal anatomical plasticity
in the associated sensory cortex, we repeatedly imaged dendritic spines of both
constitutive and conditional adult ngr1 mutant mice in somatosensory barrel cortex for
two weeks through cranial windows with two-photon chronic in vivo imaging. Neither
constant nor acute deletion of ngr1 affected turnover or stability of dendritic spines.
Thus, ngr1 alters tactile task performance but does not appear to limit the rate of tactile
learning nor determine the low set point for synaptic turnover in sensory cortex.

12
Introduction

Experience plays a crucial role in refining neural circuitry and synaptic
connectivity, yet both functional and anatomical plasticity diminish as development
concludes (Holtmaat et al., 2005; Zuo et al., 2005a). In adult mice, sensory adaptation
and motor learning are associated with elevated cortical spine dynamics (Hofer et al.,
2009; Holtmaat et al., 2006; Trachtenberg et al., 2002; Xu et al., 2009; Wilbrecht et al.,
2010; Yang et al., 2010). However, the genes and mechanisms that govern experience-
dependent anatomical cortical plasticity in the developing and mature brain are not well
characterized.
The nogo-66 receptor 1 gene (ngr1) is one candidate that may restrict anatomical
plasticity within adult cortical circuitry (Akbik et al., 2013). NgR1 is a neuronal receptor
for a number of disparate ligands that inhibit neurite outgrowth in vitro (Dickendesher et
al., 2012; He and Koprivica, 2004). Mice lacking ngr1 display critical period visual
plasticity into adulthood, suggesting NgR1 is required to close this critical period
(McGee et al., 2005). Furthermore, ngr1 mutant mice display a faster rate of
improvement on the rotarod, an assay of motor coordination, as well as dramatically
increased basal spine dynamics and stability in both sensory and motor cortex (Akbik et
al., 2013). Thus, NgR1 has been proposed to be a critical molecular determinant gating
the transition from rapid anatomical plasticity in adolescence to lower dendritic spine
dynamics in adulthood that restricts the effects of experience on cortical anatomy.
The gap crossing learning paradigm is a prime example of a distance detection
13
and object localization task (Hutson and Masterton, 1986; Jenkinson and Glickstein,
2000; Harris et al., 1999; Celikel et al., 2007; Voigts et al., 2008). In this task, animals
are placed on an elevated starting home platform in a light-tight enclosure and explore
the dark environment using their whiskers to locate a target platform placed at user-
defined distance from the home platform. At short distances, mice perform this task by
contacting the target platform with their whiskers and nose, activating whiskers as well
as touch receptors in the skin around the nose. At longer distances they must solely rely
on their whiskers for tactile information (Hutson and Masterton, 1986). Successful task
acquisition requires intact somatosensory ‘barrel’ cortex (Hutson and Masterton, 1986).
Mice improve their performance on this task with experience; this learning yields a
greater percentage of successful crossings of a given distance in successive sessions
of trials.
Here we tested the role of NgR1 as a critical gate to both experience-dependent
learning and anatomical plasticity in barrel cortex. We compared the overall
performance and rate of learning across sessions with this whisker-dependent learning
task and basal anatomical plasticity in barrel cortex with chronic two-photon in vivo
imaging by adult ngr1−/− mice and wild-type (WT) controls. Mice lacking ngr1 displayed
typical improvement across sessions despite better initial performance. We observed
that the basal dynamics of dendritic spines were indistinguishable between ngr1−/−
mice and controls. Thus, we conclude that ngr1 contributes to performance on a
sensory learning task, but does not restrict either the rate of learning or basal synaptic
turnover in corresponding sensory cortex.
14
Materials and methods

Mice
The constitutive ngr1−/− and conditional ngr1flx/flx strains have been described
previously (Kim et al., 2004; Wang et al., 2011). The ngr1−/− strain was F8 and the
ngr1flx/flx strain was F6 when these mice were re-derived (The Jackson Laboratory).
Subsequently, the ngr1flx/flx strain was backcrossed onto the C57/Bl6 background to
F8+. Each line was then backcrossed against C57/Bl6 Thy1-EGFP-M transgenic mice
(Feng et al., 2000) obtained from a commercial vendor (The Jackson Laboratory). Mice
were group housed with same-sex littermates and food and water were available ad
libitum.
Mice were maintained and all experiments conducted according to protocols
approved by the Children's Hospital Los Angeles Institutional Animal Care and Use
Committee. Mice were anesthetized by isoflurane inhalation and euthanized by carbon
dioxide asphyxiation in accordance with approved protocols.

The gap cross assay
The gap cross assay system is a closed-loop robotic environment with motor
controlled units and sensing elements. The mouse behaves upon raised platforms
driven by independent linear actuators. The platforms are equipped with servo-motor
doors and positional sensors. Data acquisition and control algorithms are both executed
online for real-time dynamic control and offline for more advanced analysis.
15
Independent linear actuators move the Plexiglass platforms to generate a range
of gap distances. To monitor the rodent motion four IR motion sensors are at the back
and edge of each platform. Near the edge of each platform are servo-controlled doors
that prevent exploratory behavior during repositioning of the platforms. The linear
motors, servos, and motion sensors are USB controlled through microcontroller boards
(Arduino Mega 2560 and the Quadstepper Motor Driver) that communicate with a quad-
core CPU.
Motor positions are processed on a quad-core CPU using the Arduino and
Matlab programming environments. Platform position, door status (open/closed) and
feeders are real-time controlled using the Arduino C-based development environment
(ADE). Motion sensor data are continuously acquired and pre-processed within ADE
and are visualized and stored in real time within Matlab via serial communication.
Specifically, sensor activity are encoded as behavioral performance metrics:
success/failed crossing events, 1) successful: animal approaches the gap and crosses
to back of target platform; or 2) failed: animal approaches the gap and then retreats to
back of home platform. This information is computed in real-time.
To control the positional and door motors, the GCS employs a closed-loop finite
state machine algorithm (D.H. Herman, unpublished observations). Animal behaviors
are segmented into interactive events at the gap. Consequently, the system is
structured as a two state machine: Exploration and Adjustment. During Exploration, the
motors are disabled and the system continuously acquires behavioral data via motion
sensors. During Adjustment, the doors close to halt exploration, and the motors
16
reposition the platforms for the next exploration phase as determined by the
programmed protocol. Transitions between the two states are triggered by behavioral
events (i.e. successful/failed gap-crossing).
Male littermates were group housed with food and water available ad libitum. The
mice were 10–12 weeks old at the start of the task. Animals were handled for 10
minutes a day for one week prior to beginning the task. The day before training began,
mice were habituated to the gap cross apparatus by placing each mouse in the chamber
with background white noise for 20 minutes in white light immediately followed by 20
minutes in the dark. A bridge was placed over the gap to prevent exploration of the gap
and gap crossing behavior during habituation.
Mice were subsequently trained once per day. Each session lasted for 20
minutes or 20 successful crosses, whichever came first. All GC training took place in the
dark with background white noise to mask any visual or auditory cues. All sessions
began with a trial at 3.0 cm, the shortest distance tested. Throughout the session, mice
were presented with nose distances (3-4 cm) and whisker distances (5-6 cm) in 0.5 cm
increments. Position of the mouse was tracked with motion sensors placed at the back
and near the edge of each platform. As a mouse traversed the platform, these sensors
record its progressive position. A successful trial was identified as any trial in which the
mouse successfully crossed the gap between the home and target platforms and
activated the motion sensor at the back of the target platform. A failed attempt was
defined as an attempt in which the mouse explored the edge of the home platform and
returned to the back of the platform. Regardless of whether the mouse succeeded or
17
failed, the gap distance would chance for the next trial in order to prevent the mouse
from having any previous knowledge of the gap distance. The next distance was
determined with a learning algorithm that randomly chose the distance from a Gaussian
distribution centered a gap distance 0.5 cm longer than the previous distance if the
preceding trial were successful, and a gap distance 0.5 cm shorter if the preceding trial
were a failure (D.H. Herman, manuscript in preparation). This approach decreases the
predictability of the subsequent gap distance relative to a ‘laddering’ learning paradigm
in which the next distance increased or decreased by a set distance depending on the
success or failure in the preceding trial.

Cranial windows
Male and female c57/Bl6 EGFP-M transgenic mice expressing green fluorescent
protein in a sparse population of cortical layer 5 pyramidal neurons (transgenic line M;
Jackson Laboratories) were used. WT, ngr1−/−, and ngr1flx/flx;Cre-ER mice received
cranial windows after P60 and imaging was initiated 4 or more weeks later. Mice were
anaesthetized with isoflurane and administered dexamethasone (4 µg/g body weight)
subcutaneously. Body temperature was maintained with a biofeedback heatpad
(Physitemp). Cranial windows were implanted as previously described, with minor
modifications (Holtmaat et al., 2009). A circular region of the skull over barrel cortex (1.2
mm caudal and 3.5 mm lateral from Bregma) was removed without perturbing the
underlying dura. A 2.5 mm diameter #0 thickness cover glass (Bellco Glass Inc.) was
placed on the dura, affixed with cyanoacrylate (Krazyglue), and sealed with dental
18
acrylic (Lang Dental). A small aluminum bar with tapped screw holes was embedded
into the acrylic to stabilize the animal for subsequent imaging sessions. Animals
received buprenorphine (0.1 µg/g body weight) and baytril (1∶1000) in water post-
surgery. Their water was also supplemented with carprofen (1∶2000) throughout the
imaging series. Mice were imaged over 4-day intervals beginning 4 weeks after
implanting cranial windows.

Optical imaging of intrinsic signals
Imaging was performed as described previously (Kalatsky and Stryker, 2003).
Mice were administered chlorprothixene (1 µg/g body weight) and isoflurane anesthesia
was maintained near 1%. To visualize whisker-evoked changes in intrinsic signals in S1
barrel cortex, a single whisker (e.g. C2) contralateral to the cranial window was
deflected approximately 15 degrees with a 3 Hz sinusoidal pulse train for 3 s every 20 s
using a piezoelectric actuator controlled by a function generator (GW Instek). This was
repeated 35 times per run.
Green light (530 nm±30 nm) was used to visualize cerebral vasculature and red
light (620 nm±20 nm) to image intrinsic signals. The imaging plane was focused ∼200–
400 µm below the cortical surface. Images were acquired at 10 Hz at 1024×1024 pixels
per image at 12-bit depth with a high-speed camera (Dalsa 1M60) and custom
acquisition and analysis software (C++ and Matlab). Collected images were spatially
binned before the response at the stimulus frequency was extracted from a complete
time series for each pixel by Fourier analysis.
19
Chronic in vivo two-photon imaging
All imaging was conducted blind to the genotype. Mice were imaged at least 4
weeks after implanting the cranial window and the average age of mice was similar
between genotypes at start of imaging (WT, P113–150, average P130; ngr1−/−, P130–
173, average P140; ngr1flx/flx;Cre-ER P104). Animals were anaesthetized with
isoflurane and body temperature was maintained with a biofeedback heat pad
(Physitemp). Images were acquired with a modified Movable Objective Microscope
(MOM) (Sutter Instruments) and 40× 1.0 NA water immersion objective (Zeiss) using
scanimage software (MatLab) (Pologruto et al., 2003). The light source is a Ti:sapphire
tunable laser (Chameleon Ultra II, Coherent) operating at 910 nm. Imaging typically
required less than 50 mW of power. The identity of L5 neurons was confirmed by
measuring the depth at which the cell body resided. Image stacks consisted of sections
(512×512 pixels) collected in 1 µm steps. Low-magnification images (0.56 µm/pixel)
were taken to visualize dendritic arbors and branching patterns. These guided the high-
magnification images (0.14 µm/pixel) collected for spine analysis. Care was taken to
maintain the same level of fluorescence across imaging intervals. Animals were imaged
every 4 days. Imaging sessions lasted no more than 2 hours.

Image analysis for dendritic spines
Image analysis was done following published guidelines (Holtmaat et al., 2009).
All analysis was done blind to genotype using ImageJ (NIH, Bethesda, MD). Dendritic
spines were not grouped into sub-categories; “spine” refers to any post-synaptic
20
protrusion that complies with previously established guidelines, and includes filopodia,
thin, mushroom, and stubby spines. Clearly defined protrusions from the dendrite
present in the first imaging interval were labeled. In image stacks from subsequent
imaging sessions, experimenters determined whether a labeled spine was still present
or not, and checked for the appearance of new spines. Newly added spines were also
tracked throughout subsequent imaging sessions. Series of image stacks for each field
were analyzed by two experimenters independently. In the case of a discrepancy
between the two sets of analysis, a third experimenter repeated the analysis. All
statistical analyses were carried out using Prism software (GraphPad).

Image presentation
Images in synaptic structures are ‘best projections’, a montage of the best focal
plane for each spine or region within a stack of images in the z-plane. This montage has
received only linear contrast adjustment.

Tamoxifen injections
Tamoxifen treatment was performed as previously described (Akbik et al., 2013;
Wang et al., 2011). In brief, tamoxifen (Sigma, T5648) was solubilized in corn oil at 10
mg/ml. Once a day for three consecutive days, mice received an intraperitoneal injection
of tamoxifen at concentration of 100 mg/kg (1 mg/10 g body mass).

21
Results

Improvement on the gap crossing task is indistinguishable between ngr-/- and wildtype
mice
To investigate if ngr1 limits the rate of sensory learning, we examined the
improvement in performance of ngr1−/− and WT mice on a sensory-based learning task.
We tested ngr1−/− and WT mice on a validated test of tactile learning, the gap cross
assay. In these experiments, we relied on the natural exploratory behavior of mice to
perform the task. To minimize variation associated with different responses to
motivation, performance was not rewarded by delivery of an appetitive food pellet with
or without food restriction. Mice were acclimated to the device and then tested at a
range of distances spanning 3.0 to 6.0 cm at 0.5 cm increments for 20 successful trials
or a maximum of 20 minutes per day for 8 consecutive days.
WT and ngr1−/− mice displayed a similar high probability to cross at distances
less than 4.5 cm (Fig. 2.1A). At these nose distances, mice are able to detect the target
platform by touching it with their nose as well as whiskers (nose + whisker, Fig. 2.1A). In
contrast, mice acquire tactile information exclusively with their whiskers at longer
distances: 5.0, 5.5 and 6.0 cm. At these ‘whisker distances’, overall performance
declines with increasing gap distance (whisker only, Fig. 2.1A). However, overall
performance was significantly higher for ngr1−/− mice at whisker distances but not the
closer nose distances (Fig. 2.1A). To assess improvement in performance with
experience, we compared the probability of successful crossing between the first four
22
sessions (1–4) and the subsequent four sessions (5–8) at all gap distances tested. At
the shorter nose distances, sensory information is more definitive and mice exhibit little
improvement in performance between the first four sessions and subsequent four
sessions. However, at the longer whisker distances, both WT and ngr1−/− mice improve
with experience (Fig. 2.1B,C). The percentage of improvement by WT and ngr1−/− mice
was greatest at the shortest of the three whisker distances and the two genotypes
exhibited similar improvement across these three distances (WT n = 19; ngr1−/− n = 14,
two-way ANOVA, P > 0.75; for each corrected pairwise comparison for distances 5.0,
5.5, and 6.0 cm, P > 0.97; Fig. 2.1B,C). In addition, although ngr1−/− mice are reported
to be slightly hypoactive in an open field test (Kim et al., 2004), WT and ngr1−/− mice
performed a similar number of total trials (WT, 142.1 ± 9.9 trials; ngr1−/−, 165.0 ± 10.5
trials; P > 0.18). Thus, ngr1−/− mice display greater overall performance but no evident
difference from WT mice in the rate of improvement is seen at any given gap distance
across sessions.

23

A
B C
Figure 2.1. Mice lacking ngr1 perform better on the gap cross assay but display
normal tactile learning across sessions. A, ng1-/- mice cross whisker distances at
significantly higher success rate (WT n = 19, ngr1-/- n = 14, two-way ANOVA
followed by Bonferroni’s multiple comparisons test, **P < 0.01 at 5.5 and 6.0 cm, P
> 0.32 at 3.5 and 4.0 cm). B, WT mice improve with experience at whisker
distances from the first 4 session (Early, grey line) to the second 4 session (Late,
black line) (WT n = 19, two-way RM ANOVA, *P < 0.05). C, Ngr1 mutant mice
improve with experience at whisker distances from the first 4 sessions (Early, pink
line) to the second 4 session (Late, red line) (ngr1-/- n = 14, two-way RM ANOVA,
*P < 0.05). Data presented as mean ± s.e.m.
24
Cranial windows placed over barrel cortex allow for repeating imaging of synaptic
structures
To determine if the better performance on the gap cross performance correlated
with greater basal synaptic structural plasticity, we imaged dendritic spines in layer I of
somatosensory (S1) barrel cortex with chronic in vivo two-photon microscopy. WT and
ngr1−/− mice harboring the EGFP-M transgene were implanted with cranial windows
over barrel cortex (Fig. 2.2A,B). To check if the cranial windows were properly
positioned, we determined the functional representation of the C2 barrel with intrinsic
signal imaging in a subset of mice (Kalatsky and Stryker, 2003) (Fig. 2.2C). The C2
barrel was near the center of the window, consistent with location of the dendritic spines
imaged in our study in S1 barrel cortex. We repeatedly imaged the apical dendrite of
individual layer V neurons over 12 days (Fig. 2.3 A,B). As the EGFP-M transgene
provides sparse expression of GFP in S1 (Fig. 2.3A), we were able to obtain high quality
images across multiple 4-day intervals (Fig. 2.3B). These images stacks were sufficient
to discern not only large mushroom spines but also thinner and often transient thin
spines on these dendrites.

25

A B
C
Figure 2.2. Cranial windows allow for long-term, in vivo imaging.
A, A cranial window immediately after implantation. Note the
healthy, undisturbed cortex. B, The same cranial window 1 week
after implantation. Vasculature remains unchanged and the window
is optically clear. C, An example of optical imaging of intrinsic
signals reveals the cortical region responsive to stimulation of the
C2 whisker.
26

10um
2um
A
B
Figure 2.3. Dendritic spines
remain dynamic throughout
life. A, The same dendritic
branch imaged at low-
magnification 12 days apart.
Dendritic branches remain
stable in adulthood. B,
Higher magnification
images of a portion (yellow
box) of the dendrite seen
above imaged every 4 days
for 12 days. Arrowheads
show examples of stable
spines (blue), lost spines
(red), transient spines
(yellow), and a new
persistent spine (green).
27
Dendritic spine dynamics and stability are indistinguishable between WT and ngr1-/-
mice
To examine baseline cortical spine dynamics in ngr1-/- versus WT mice, we
imaged the same population of spines in 4 sessions over 12 days (WT, 1512 spines, 10
neurons, n = 5 mice; ngr1−/−, 1106 spines, 8 neurons, n = 4 mice). The average
turnover ratio was almost identical between genotypes (WT 16.3% ± 1.6% versus
ngr1−/− 17.1% ± 1.5%, P > 0.96; Fig. 2.4A). Turnover was not significantly different
between 4-day intervals for WT or ngr1−/− mice (P > 0.9) nor between genotypes for a
specified interval (P > 0.4) (Fig. 2.4A). The average percentage of spines gained and
lost every 4 days was also similar (gained: WT 15.2% ± 1.5% versus ngr1−/− 18.0% ±
1.1%, P > 0.2; lost: WT 16.9% ± 1.2% versus ngr1−/− 16.9% ± 1.5%, P > 0.99; Fig.
2.4B).
Next, we measured the survival fraction of spines present at the first imaging
session (Day 0) at three subsequent 4-day intervals to determine the stability of the
overall population of spines imaged. The percentage of spines imaged on Day 0 that
were present on Day 4, Day 8, and Day 12, is identical between WT and ngr1−/− mice
(P > 0.82; Fig. 2.4C). These rates for spine turnover and stability are consistent with
previous measurements from S1 for WT mice 3–4 months of age, the same ages of
mice examined here (Holtmaat et al., 2005; Wilbrecht et al., 2010; Holtmaat and
Svoboda, 2009).
A subpopulation of transient spines are continuously appearing and disappearing
on excitatory neurons. These spines have a survival time constant of approximately 8
28
days (Holtmaat et al., 2005). We tracked the survival of spines that appeared during
Day 4 (imaging session 2) over the following two imaging sessions on Day 8 and Day
12. The percentage of transient spines surviving less than 4 days and spines present for
less than 8 days did not differ between genotypes (transient, WT 65.0% ± 1.5% versus
ngr1−/− 59.5% ± 4.4%, P > 0.39; 4 Day, WT 13.4% ± 5.6% versus ngr1−/− 13.3% ±
2.2%, P > 0.87; Fig. 2.4D). However, there was a trend towards greater persistence for
new spines in ngr1−/− mice, but this difference did not reach statistical significance (8+
Day, WT 20.9% ± 2.8% versus ngr1−/− 32.9% ± 5.2%, P > 0.11; Fig. 2.4D). We observe
that ngr1−/− mice display basal spine formation, spine retraction, and new spine stability
in barrel cortex that are indistinguishable from WT mice.

29

A
0
0.05
0.1
0.15
0.2
0.25
0.3
Turnover ratio
Day 0-4 Day 4-8 Day 8-12 Average
WT
ngr1-/-
B
Losses
0
5
10
15
20
25
30
4D Spine Change (%)
WT
ngr1-/-
Gains
0
0.2
0.4
0.6
0.8
1
0 12 8 4
Survival fraction
C
Days
WT
ngr1-/-
Transient
0
0.2
0.4
0.6
0.8
1
Fraction of new spines
D
4
Day
8+
Day
WT
ngr1-/-
A B
C D
Figure 2.4. Dendritic spine turnover and stability are normal in ngr1-/- mice. A, The
turnover of dendritic spines every 4 days in WT and ngr1-/- mice is similar across 4-
day intervals as well as the average across all sessions (WT n = 5, 1512 spines,
ngr1-/- n = 4, 1106 spines, Mann-Whitney). B, The average percent of spines
gained and lost is similar between WT and ngr1-/- mice (WT n = 5, ngr1-/- n = 4,
Mann-Whitney). C, The survival fraction of spines present on day 0 re-examined at
days 4, 8, and 12 is nearly between genotypes (WT n = 5, ngr1-/- n = 4, Mann-
Whitney). D, The fraction of new spines appearing on day 4 that are transient
(surviving less than 4 days), those lasting more than 4 days but less than 8 days,
and persistent spines surviving more than 8 days are similar between WT and ngr1-
/- mice (WT n = 5, ngr1-/- n = 4, Mann-Whitney). Data presented as mean + s.e.m.
30
Deleting ngr1 in adulthood does not impact basal cortical spine dynamics
Next, as suppressing NgR1 expression in primary cultures of hippocampal
neurons alters spine dynamics in vitro (Will et al., 2012), we employed a conditionally
targeted allele of ngr1, ngr1flx, to test if acute deletion of ngr1 affects cortical spine
dynamics in vivo. We examined ngr1flx/flx mice that also harbored a transgene that
ubiquitously expresses a fusion protein of Cre recombinase and a mutant version of the
estrogen receptor (Cre-ER) that is activated by the estrogen analog tamoxifen (Wang et
al., 2011; Hayashi and McMahon, 2002). We repeatedly imaged spines in S1 barrel
cortex of ngr1flx/flx; Cre-ER mice immediately prior to tamoxifen treatment and then for
another 4 sessions over 16 days (1083 spines, 6 neurons, 3 mice) (Fig. 2.4A).
Densitometry of immunoblots for NgR1 confirms that tamoxifen treatment reduces NgR1
protein to less than a quarter of normal levels within a week and that NgR1 protein is not
detectable after 2 weeks (Wang et al., 2011). Spine turnover was nearly identical in
ngr1flx/flx; Cre-ER mice across four 4-day intervals spanning the decline and absence
of NgR1 (P > 0.9; Fig. 2.4B). Overall, we observe that cortical spine dynamics were not
affected by the acute loss of NgR1.

31

0
0.05
0.1
0.15
0.2
0.25
0.3
Turnover ratio
0-4 4-8 8-12 12-16
Days
ngr1 flx/flx
Cre-ER
Cranial
Window
14+
days
Imaging (Days)
Tmx
0 4 8 12 16
A
B
Figure 2.5. The acute deletion of ngr1 does not affect basal
cortical spine dynamics. A, Timeline of acute deletion of ngr1 in
ngr1flx/flx;Cre-ER mice following tamoxifen injection and
subsequent imaging schedule. B, The turnover ratio does not
change with the decline or absence of NgR1 protein (n = 3, 6
neurons, 1083 spines, one-way RM ANOVA). Data presented as
mean + s.e.m.

32
Discussion

NgR1 has been implicated in restricting plasticity in the adult central nervous
system. In the visual system, ngr1-/- mice exhibit critical period plasticity into adulthood
(McGee et al., 2005). These mice have also been reported to limit the rate of motor
learning and determine the set point for synaptic turnover in adult motor and sensory
cortex (Akbik et al., 2013). To evaluate if ngr1 restricts tactile performance or learning,
we examined the performance and learning of ngr1−/− and WT mice with the gap cross
assay. We observe that ngr1−/− mice perform significantly better on this task. However,
this greater overall performance by ngr1−/− mice was not associated with a greater rate
of learning at any gap distance across sessions. The percent improvement with
experience averaged between genotypes was similar at both nose and whisker
distances, despite significantly better initial performance by ngr1−/− mice.
Whether the behavioral phenotypes of ngr1 mutant mice are a consequence of
deletion of the gene within neocortex is not yet known. NgR1 is also expressed in
thalamus, hippocampus, and amygdala (Barrette et al., 2007). Loss of ngr1 may affect
the circuitry or plasticity in these brain structures to alter the performance of mice on
various sensory and motor learning paradigms. Future experiments combining the
ngr1flx allele with transgenic expression of cre recombinase restricted to cortical
neurons or subcortical neurons may be able to answer this question (Gong et al., 2007;
Stephany et al., 2014).

33
Although it has been previously reported that ngr1 mutant mice display
dramatically elevated spine turnover in barrel cortex (Akbik et al., 2013), we observe
basal spine dynamics that are indistinguishable from WT mice. The reasons for the
differences between our results are unclear. There are important and extensive
similarities between the experiments presented here and those published by Akbik et al.
(2013). Both studies employ the same strains of ngr1 mutant mice. The two studies
measure dendritic spine turnover in the same regions of sensory cortex in mice of
similar ages. The two studies employ the same EGFP-M transgene to sparsely express
GFP in cortical neurons in similar experiments. Whether ngr1 mutant mice display
greater synaptic structural plasticity during sensory adaptation or learning remains to be
determined.
However, there are some differences in the imaging experiments between the
two studies. We employed cranial windows to image dendrites and axons repeatedly at
multiple consecutive 4-day intervals in barrel cortex. We confirmed that the windows
were properly positioned in a subset of mice with optical imaging of intrinsic signals.
Akbik et al. (2013) predominantly used the ‘thinned-skull’ transcranial technique to
image dendrites and axons from YFP-H transgenic mice at either a 2-day interval or a
14-day interval near specified stereotaxic coordinates. However, they also imaged the
identical strain of ngr1−/− mice expressing GFP from the same EGFP-M transgene
through cranial windows as we employed. They concluded that these two preparations
yield similar results. Thus, a core set of imaging experiments were performed under
conditions that are directly comparable.
34
Transcranial imaging and cranial window imaging have distinct advantages and
disadvantage (Holtmaat et al., 2009; Xu et al., 2007; Yang e al., 2010). The surgical
preparation for ‘thin skull’ imaging is faster, more reliable, and unlikely to damage the
underlying cortex. However, the small cortical region accessible for imaging often
requires employing YFP-H transgenic mice that express fluorescent protein in many
more neurons relative to the comparable EGFP-M transgenic mice (Feng et al., 2000;
Holtmaat et al., 2005). This broader expression often results in a dense lattice of labeled
dendritic arbors in layer I, especially in older mice. With this approach, typically 100–150
spines are imaged from many distinct segments of dendrite originating from a number of
neurons across often a single 2-day or 14-day interval. By comparison, the surgical
preparation for cranial windows is more elaborate, the rate of successful surgery lower,
and the consequence of inflammation or damage to the region of interest are a concern.
However, the imaging region is significantly larger and compatible with the more sparse
EGFP-M transgenic line that offers greater imaging contrast. In addition, this stable
preparation permits repeated imaging of the same dendritic structures over extended
periods of time. Thus, transcranial imaging may provide a more general sampling of
cortical synaptic structural dynamics because spines are distributed among multiple
neurons, whereas cranial window imaging permits a more specific analysis of individual
neurons. On average, Akbik et al. (2013) imaged approximately 100 spines per mouse
for one time interval for both transcranial and cranial window imaging. By comparison,
we imaged on average 300 spines per mouse over three consecutive time intervals.
Repeated imaging improves the statistical power and precision of estimates of
35
spine turnover and stability relative to imaging a single time interval. Although some
controversy persists regarding the relative magnitude of dendritic spine turnover
observed by transcranial imaging (Xu et al., 2007; Yang et al., 2010) versus cranial
window imaging (Trachtenberg et al., 2002; Holtmaat et al., 2009), several groups have
both independently and collectively verified that cortical spine dynamics imaged through
cranial windows are consistent for weeks to months (Holtmaat et al., 2005; Hofer et al.,
2009; Wilbrecht et al., 2010; Holtmaat et al., 2009). Notably, a recent study monitored
the spine stability and density by imaging the same dendrites of layer 5 neurons in S1
through cranial windows in adult mice for more than a year (Mostany et al., 2013).
Overall, we conclude that while ngr1 regulates tactile performance, it does not
limit the rate of tactile learning nor determine the low set point for synaptic turnover in
adult sensory cortex.

36
CHAPTER THREE
Dendritic spine stability, but not turnover, is enhanced in barrel cortex during acquisition
of a perceptual learning task

Abstract

Motor learning has been shown to induce drastic changes in the neural circuitry
of layer 5 neurons in motor cortex. Concurrent with motor learning, layer 5 neurons
exhibit increases in dendritic spine turnover, stabilization, and clustering. This
rearrangement and selective reinforcement in synaptic connectivity is thought to allow
for learning and maintenance of motor skills. However, it remains to be determined
whether such structural changes occur in other cortical areas. To determine whether
such changes also occur in sensory cortex, we repeatedly visualized dendritic spines in
barrel cortex throughout the acquisition of a tactile-based learning task. We found an
increase in learning dependent stabilization of newly formed spines while turnover and
clustering remained steady. Thus, stabilization of new spines rather than an overall
increase in turnover contributes to lasting changes to neural circuitry associated with
sensory learning.

37
Introduction

Experience-dependent modifications to synaptic structures are most pronounced
during development, but continue into adulthood (Holtmaat and Svoboda, 2009).
Dendritic spines remain sensitive to altered sensory experience as seen by changes in
dendritic spine dynamics in adult sensory cortex (Trachtenberg et al., 2002; Holtmaat et
al., 2006; Hofer et al., 2009; Yang et al., 2010). This formation and elimination of
dendritic spines may serve as a substrate for neural circuit plasticity and may provide an
anatomical basis for long-term information storage.
Recent studies have demonstrated extensive changes to turnover, stabilization,
and clustering of new spines during learning. Most of these studies have been carried
out using motor learning paradigms and examining fluorescently labeled dendritic
spines of layer V pyramidal neurons in motor cortex (Xu et al., 2009; Yang et al., 2009;
Fu et al., 2012). Motor learning increases new spine formation and elimination as well
as increases the fraction of new spines that are stabilized (Xu et al., 2009; Yang et al.,
2009; Fu et al., 2012). Furthermore, spines that are formed during learning are more
likely to appear in clusters, and spines forming in clusters are more likely to be
maintained (Fu et al., 2012).
In sensory systems, experience sculpts anatomical and functional modifications
to neural circuitry and drives perceptual learning. It remains unknown whether the rules
governing learning-dependent changes in motor cortex are applicable to sensory cortex.
To address this question, we examined spine dynamics in barrel cortex throughout the
38
acquisition of a tactile learning task. Mice were repeatedly imaged using two-photon
laser scanning microscopy throughout baseline and learning conditions. No change in
dendritic spine turnover or clustering was detected during learning compared to baseline
conditions. However, learning induced an increase in new spine stabilization, which was
more pronounced in better performing mice. Thus, we conclude that learning a tactile-
based task does not result in drastic changes in dendritic spine turnover or clustering,
but does enhance the maintenance of newly formed spines.

39
Materials and methods

Mice
8 male adult mice were used for this study. C57/Bl6 Thy1-EGFP-M transgenic
mice (Feng et al., 2000) were obtained from a commercial vendor (The Jackson
Laboratory). These mice express green fluorescent protein in a small population of
cortical layer 5 pyramidal neurons. Male littermates were group housed with food and
water available ad libitum. Mice were maintained and all experiments conducted
according to protocols approved by the Children’s Hospital Los Angeles Institutional
Animal Care and Use Committee. Mice were anesthetized by isoflurane inhalation and
euthanized by carbon dioxide asphyxiation in accordance with approved protocols.

Cranial windows surgeries
Male c57/Bl16 EGFP-M transgenic mice received cranial windows over barrel
cortex at 6-12 weeks of age as previously described (Holtmaat et al., 2009) with minor
modifications. Mice were anaesthetized with 1.5-2% isoflurane and administered
dexamethasone (4 ug/g body weight) subcutaneously. A round portion of skull over
barrel cortex (1.2 mm caudal and 3.5 mm lateral from Bregma) was removed without
disturbing the underlying dura and replaced by a 2.5 mm diameter #0 thickness glass
coverslip (Bellco Glass Inc.). This was held in place with cyanoacrylate (Krazyglue) and
sealed with dental acrylic (Lang Dental). A small aluminum bar with tapped screw holes
(custom-made) was embedded in the acrylic in order to stabilize the head during
40
subsequent imaging sessions. Body temperature was maintained with a biofeedback
heatpad (Physitemp). Post-surgery, animals received buprenorphine (0.1 ug/g body
weight) and their water was supplemented with baytril (1:1000) and carprofen (1:2000).
Mice were housed in pairs to reduce incidences of fighting. Mice were given at least 4
weeks to recover before starting experiments.

Chronic in vivo two-photon imaging
Mice were 12-16 weeks old at the start of imaging and at least 4 weeks post
cranial window surgery. Mice were anaesthetized with isoflurane and body temperature
was maintained with a biofeedback heat pad (Physitemp). Images were acquired with a
modified Movable Objective Microscope (MOM) (Sutter Instruments) and 40× 1.0 NA
water immersion objective (Zeiss) using scanimage software (MatLab) (Pologruto et al.,
2003). The light source is a Ti:sapphire tunable laser (Chameleon Ultra II, Coherent)
operating at 910 nm. Imaging typically required less than 50 mW of power. The identity
of L5 neurons was confirmed by measuring the depth at which the cell body resided.
Image stacks consisted of sections (512×512 pixels) collected in 1 µm steps. Low-
magnification images (0.56 µm/pixel) were taken to visualize dendritic arbors and
branching patterns. These guided the high-magnification images (0.14 µm/pixel)
collected for spine analysis. Care was taken to maintain the same level of fluorescence
across imaging intervals. Apical dendrites that were imaged were traced back to the cell
body to confirm layer V cell identity. Animals were imaged every 4 days for 28 days with
each imaging session lasting less than 2 hours. Mice were not imaged on days where
41
they were trained on the gap crossing task. All imaging was conducted blind to the
animal’s performance on the learning task.

Gap cross paradigm
The gap cross assay system is a closed-loop robotic environment with motor
controlled units and sensing elements. The mouse behaves upon raised platforms
driven by independent linear actuators. The platforms are equipped with servo-motor
doors and positional sensors. Data acquisition and control algorithms are both executed
online for real-time dynamic control and offline for more advanced analysis.
Independent linear actuators move the Plexiglass platforms to generate a range
of gap distances. To monitor the rodent motion four IR motion sensors are at the back
and edge of each platform. Near the edge of each platform are servo-controlled doors
that prevent exploratory behavior during repositioning of the platforms. The linear
motors, servos, and motion sensors are USB controlled through microcontroller boards
(Arduino Mega 2560 and the Quadstepper Motor Driver) that communicate with a quad-
core CPU.
Motor positions are processed on a quad-core CPU using the Arduino and
Matlab programming environments. Platform position, door status (open/closed) and
feeders are real-time controlled using the Arduino C-based development environment
(ADE). Motion sensor data are continuously acquired and pre-processed within ADE
and are visualized and stored in real time within Matlab via serial communication.
Specifically, sensor activity are encoded as behavioral performance metrics:
42
success/failed crossing events, 1) successful: animal approaches the gap and crosses
to back of target platform; or 2) failed: animal approaches the gap and then retreats to
back of home platform. This information is computed in real-time.
To control the positional and door motors, the GCS employs a closed-loop finite
state machine algorithm (D.H. Herman, unpublished observations). Animal behaviors
are segmented into interactive events at the gap. Consequently, the system is
structured as a two state machine: Exploration and Adjustment. During Exploration, the
motors are disabled and the system continuously acquires behavioral data via motion
sensors. During Adjustment, the doors close to halt exploration, and the motors
reposition the platforms for the next exploration phase as determined by the
programmed protocol. Transitions between the two states are triggered by behavioral
events (i.e. successful/failed gap-crossing).
Mice were habituated to the apparatus (20 mins/day) between imaging sessions
3 and 4 (3 days total). A bridge was placed over the gap to preclude exploration of the
gap and gap crossing behavior. The first day of habituation was presented with white
light and background white noise. The following two days of habituation were presented
in the dark with background white noise. This block of habituation served two purposes:
the first was to habituate mice to the GC apparatus, and the second was to control for
any changes arising from increased exploratory behavior or exposure to a novel
environment. Mice were then trained once a day between imaging sessions 4 and 7 (9
days total). Mice were not trained on imaging days. Each session lasted for 20 minutes
or 20 successful crosses, whichever came first. Mice were presented with nose
43
distances (3-4 cm) and whisker distances (5-6 cm) in 0.5 cm increments. All GC training
took place in the dark with background white noise to mask any visual or auditory cues.
All sessions began with a trial at 3.0 cm, the shortest distance tested. Position of
the mouse was tracked with motion sensors placed at the back and near the edge of
each platform. As a mouse traversed the platform, these sensors record its progressive
position. A successful trial was identified as any trial in which the mouse successfully
crossed the gap between the home and target platforms and activated the motion
sensor at the back of the target platform. A failed attempt was defined as an attempt in
which the mouse explored the edge of the home platform and returned to the back of
the platform. Regardless of whether the mouse succeeded or failed, the gap distance
would chance for the next trial in order to prevent the mouse from having any previous
knowledge of the gap distance. The next distance was determined with a learning
algorithm that randomly chose the distance from a Gaussian distribution centered a gap
distance 0.5 cm longer than the previous distance if the preceding trial were successful,
and a gap distance 0.5 cm shorter if the preceding trial were a failure (D.H. Herman,
manuscript in preparation). This approach decreases the predictability of the
subsequent gap distance relative to a ‘laddering’ learning paradigm in which the next
distance increased or decreased by a set distance depending on the success or failure
in the preceding trial.
Behavioral and performance data was not analyzed until the image analysis was
complete. Poor learners were defined as those that crossed at a maximum gap distance
of 5.0cm (n = 3 mice, 6 neurons). Good learners were defined as those that crossed at
44
a maximum gap distance of 5.5 or 6.0cm with at least a 50% success rate (n = 5 mice, 8
neurons).

Image analysis
Dendritic spine identification and analysis was performed following previously
established guidelines (Holtmaat et al., 2009) using Image J (NIH, Bethesda, MD).
Dendritic spines were not grouped into sub-categories; “spine” refers to any post-
synaptic protrusion that complies with previously established guidelines, and includes
filopodia, thin, mushroom, and stubby spines. To calculate turnover ratios, each neuron
was required to have a minimum of 50 spines on the first imaging day along dendritic
segments that maintained high image quality throughout all 8 imaging sessions in order
to be included in the analysis. For turnover analysis, n = 8 mice, 15 neurons, 163 ± 67
spines (mean ± S.D.) per neuron. To calculate new spine survival, each neuron was
required to have a minimum of 90 spines on the first imaging day. For stability analysis,
n = 8 mice, 14 neurons, 171 ± 61 spines (mean ± S.D.) at day 0 per neuron.
Neighborhood analysis was performed to determine whether synapses appear in
clusters during learning (Fu et al., 2012). The formation of 2+ new spines in the same
session was considered to be a cluster if the spines appeared without any previously
existing spine interspersed. For cluster analysis, n = 8 mice (1 neuron from each). All
statistical analyses were carried out using Prism software (GraphPad).

45
Image presentation
Images in synaptic structures are ‘best projections’, a montage of the best focal
plane for each spine or region within a stack of images in the z-plane. This montage
received linear contrast adjustment and was filtered with a 1.5-pixel radius Gaussian
blur for presentation.

46
Results

Performance on the gap crossing task improves with increased exposure
In order to examine anatomical changes associated with sensory learning, we
trained mice on the gap crossing task. The gap crossing task is a whisker- and barrel
cortex-dependent learning paradigm that serves as an ideal distance detection and
object localization task (Hutson and Masterton, 1986; Jenkinson and Glickstein, 2000;
Harris et al., 1999; Celikel et al., 2007; Voigts et al., 2008). Mice performed this task
without any food or water reward, and hence were not food or water restricted at any
time. After being habituated to the gap cross apparatus, mice received gap cross
training once per day for 9 days.
All mice used for this study (n=8) learned to cross at whisker distances and
improvement is seen over a number of days (Fig. 3.1A,B). “Early” refers to the first half
of training sessions while “late” refers to the second half of training sessions.
Performance improved at whisker distances in the second half of sessions, but
remained the same at nose distances, at which performance started and remained high
(two-way ANOVA, nose vs whisker (P < 0.0001) and early vs late (P < 0.002), Fig. 3.1B).
At smaller nose distance (3.0, 3.5, and 4.0 cm), mice are able to utilize both their
whiskers and touch receptors in their nose to gather information about the target
platform. At larger whisker distances (5.0, 5.5, and 6.0 cm), mice must rely solely on
tactile information relayed by their whiskers and their performance improves with
experience.
47

3 3.5 4 4.5 5 5.5 6
0
20
40
60
80
100
Gap distance (cm)
Percent success
Early
Late
whisker only nose + whisker
Early
Late
0
20
40
60
80
100
Percent success
Nose Whisker
Distance
n.s.
**
A
B
Figure 3.1. Mice display improved
performance on the gap crossing
task with increased number of
trials. A, Mice are presented with
nose and whisker distances. Per-
formance starts and remains high
at nose distances. B, Performance
improves at whisker distances in
the later half of sessions (n=8,
two-way ANOVA followed by
Bonferroni’s multiple comparisons
test, **P < 0.01, n.s. indicates not
significant, P = 0.4924). Data pre-
sented as mean + s.e.m.
48
Tactile learning enhances the stabilization of newly formed spines
In order to examine the structural changes that occur with tactile learning, we
repeatedly imaged the same population of dendritic spines before, during, and after
learning over a month long period. Mice carrying the EGFP-M transgene were implanted
with a cranial window over barrel cortex. Utilizing chronic in vivo two-photon microscopy,
we imaged the apical dendrites of layer 5 pyramidal neurons found in layer 1 every 4
days.
A total of 19,773 spine observations were made over 8 evenly spaced imaging
sessions across 28 days (2,472 ± 24 (mean ± S.D.) per session) throughout baseline
and learning conditions (Fig. 3.2A; Table 3.1). In order to determine whether new spines
formed during task acquisition established lasting connections, we looked at new spine
survival during baseline compared to learning (n = 8 mice, 14 neurons, 171 ± 61 spines
(mean ± S.D.) per neuron). We found that the 4 day stability for new spines formed
during gap cross training is greater compared to those formed during baseline (Fig.
3.2B). A larger fraction of new spines is maintained throughout the duration of learning,
suggesting that the stabilization of experience-dependent formed spines plays an
important role in learning.

49

Cranial
window
Day
-28 0 4 8 12 16 20 24 28
Imaging session
1 2 3 4 5 6 7 8
GC task
0
20
40
60
4-8 8-12 12-16 16-20 24-28 20-24
4 day survival (%)
Day
*
Baseline
Learning
A
B
Figure 3.2. Learning a whisker-dependent task increases the stabilization of
newly formed spines in barrel cortex. A, Timeline of the experiment. Imaging
begins 1 month after cranial window implantation and takes place every 4
days. Mice are trained on the gap crossing task between Day 12 and Day
24. B, The percentage of newly formed spines lasting 4+ days is greater
during learning than baseline conditions (n = 14, two-tailed t-test, *P < 0.02).
Data presented as mean + s.e.m.

50
Better performance on the gap crossing task correlates with increased new spine
stability
In order to investigate the possible structural changes that could contribute to the
differences in learning rate among mice, we grouped mice as ‘poor’ or ‘good’ learners.
Poor learners were defined as those who crossed at a maximum gap distance of 5.0 cm,
the smallest whisker distance (n = 3 mice, 6 neurons). Good learners were defined as
those who crossed at a maximum of 5.5 or 6.0 cm (n = 5 mice, 8 neurons). We then
examined the stabilization of new spines formed in the imaging session prior to a
mouse’s maximal performance across 12 days (spines formed on day 12 or 16). We
also compared each group’s respective baseline stability, to account for any pre-existing
differences between the groups.
We found that the new spine survival fraction was greater during learning in good
learners compared to poor learners at every imaging interval for 12 days. Survival
fraction in good learners during learning was greater compared to that of bad learners
during learning, bad learners during baseline, and good learners during baseline at 4, 8,
and 12 days (effect of time (P < 0.0001) and group (P < 0.01)), Fig. 3.3A,B). Baseline
values for each group were indistinguishable at every time point, suggesting that the
good learners’ greater improvement in performance and resultant increase in new spine
stability is not the result of a difference in basal spine dynamics. This reinforces the idea
that better learning drives the increased maintenance of newly formed spines. These
results further suggest that the timing of new spine stabilization corresponds to maximal
performance.
51

0 4 8 12
0.00
0.25
0.50
0.75
1.00
Day
Survival fraction
Poor learning
Good learning
Poor baseline
Good baseline
4 8 12
0.0
0.1
0.2
0.3
0.4
0.5
Day
Survival fraction
Bad Learning
Good Learning
Bad Baseline
Good Baseline
*
**
**
A
B
Figure 3.3. Mice with better performance on the gap crossing task
display higher rates of new spine survival. A, 12 day survival fraction of
newly formed spines in good and bad learners during baseline and
learning conditions. B, Good learners show enhanced survival of spines
formed during learning at all intervals over 12 days (n=14, RM two-way
ANOVA followed by Holm-Sidak multiple comparisons test, **P < 0.01,
*P < 0.05). Data presented as mean + s.e.m.
52
Learning on the gap crossing task does not increase spine formation and elimination
Next, we examined other aspects of spine dynamics to determine whether
learning resulted in additional structural changes (n = 8 mice, 15 neurons, 163 ± 67
spines (mean ± S.D.) per neuron; Table 3.1). Dendritic spines turn over under baseline
conditions (Fig. 3.4A), and it has been previously reported that this increases
significantly during learning (Xu et al., 2009; Yang et al., 2009; Fu et al., 2012). When
we looked at spine turnover, we found that formation and elimination were not
significantly altered during learning. Although the turnover ratio showed a modest
decrease at day 20 and day 24 (P < 0.01; Fig. 3.5A), this likely resulted alongside
increased new spine stability, as there was no change in total spine number per neuron
(P = 0.84; Fig. 3.5B). If a greater fraction of new spines are being maintained without a
change in total spine number, there should be an accompanying decrease in spine
formation and elimination. This further suggests homeostatic mechanisms are in place
to maintain spine density. We observe no change in the percent of new spines added
across all imaging intervals (P = 0.10). The percent of spine losses showed an overall
change throughout the imaging timeline (P < 0.02), but multiple comparisons did not
reveal any significant pairs. Again, this slight decrease in spine loss appears to be a by-
product of increased spine stability. Learning did not induce dramatic changes in spine
turnover, suggesting that an increase in spine formation and elimination is not critical to
sensory learning.

53

d0 d4 d8 d12 d16 d20 d24 d28
q
q
q
q
q
q
q
q
q
q
q
q
* *
*
2 um
d0 d4 d8 d12 d16 d20 d24 d28
q
q
q
q
q
q
q
q
q
q
q
q
* *
*
2 um
A
Figure 3.4. Although most dendritic spines are stable in
adulthood, a small population of spines continues to appear
and disappear. A, Example stretch of dendrite across all
imaging sessions. Arrowheads show examples of stable spines
(blue), lost spines (red) and transient spines (yellow). Green
star notates a new persistent spine formed during learning.
54

GC training
Baseline 12 16 20 24 28
0.0
0.1
0.2
0.3
Day
Turnover ratio
* *
GC training
Baseline 12 16 20 24 28
0
100
200
300
Day
Total spine number
GC training
Baseline 12 16 20 24 28
0
10
20
30
Day
Spine gain (%)
GC training
Baseline 12 16 20 24 28
0
10
20
30
Day
Spine loss (%)
A B
C D
Figure 3.5. Learning on the gap crossing task does not dramatically impact rates of
spine formation or elimination in sensory cortex. A, Turnover ratio remains fairly
stable, with slight decrease at day 20 and day 24 (n=15, RM one-way ANOVA
followed by Holm-Sidak multiple comparisons test, *P < 0.05). B, Total spine
number remains stable across all imaging sessions (n=15, RM one-way ANOVA).
C, Percent spine gain remains stable across all imaging sessions (n=15, RM one-
way ANOVA). D, Percent spine loss shows an overall change but no significant
pairs (n=15, RM one-way ANOVA). Data presented as mean + s.e.m. Lines indicate
individual neurons.

55

Day$0
Mouse Neuron Total T.O. Total T.O. Total T.O. Total T.O. Total T.O. Total T.O. Total T.O. Total
+45 +35 +46 +38 +27 +42 +52
;41 ;41 ;43 ;39 ;37 ;28 ;55
+20 +28 +17 +24 +13 +24 +34
;14 ;24 ;26 ;17 ;20 ;14 ;24
+30 +36 +29 +23 +18 +22 +14
;28 ;31 ;31 ;32 ;20 ;14 ;35
+17 +16 +21 +16 +17 +10 +15
;19 ;24 ;14 ;23 ;16 ;13 ;13
+23 +29 +29 +17 +20 +27 +25
;19 ;22 ;23 ;25 ;18 ;17 ;33
+60 +51 +53 +45 +48 +49 +52
;47 ;61 ;49 ;55 ;32 ;52 ;53
+18 +22 +14 +23 +13 +12 +16
;16 ;19 ;25 ;13 ;19 ;11 ;16
+10 +9 +9 +13 +10 +7 +9
;12 ;12 ;10 ;8 ;12 ;9 ;10
+55 +46 +50 +48 +53 +41 +47
;42 ;50 ;49 ;47 ;54 ;54 ;52
+33 +39 +21 +35 +32 +23 +34
;36 ;31 ;38 ;22 ;35 ;32 ;23
+44 +45 +39 +38 +37 +44 +45
;36 ;50 ;49 ;44 ;37 ;38 ;44
+67 +71 +69 +65 +65 +66 +69
;55 ;69 ;67 ;73 ;53 ;73 ;69
+37 +31 +32 +36 +31 +43 +34
;34 ;33 ;38 ;36 ;35 ;31 ;39
+12 +9 +10 +10 +9 +10 +11
;8 ;11 ;9 ;9 ;10 ;11 ;11
+36 +38 +33 +47 +28 +23 +27
;31 ;37 ;38 ;28 ;37 ;31 ;32
244 A
B
A 124
114
248
120
91
174
51
177
261
170
220 A
A
B
C
A
B
A
B
C
A
A
B
118
277
113
108
218
110
111
264
114
225
173
275
178
53
180 179
242
124
131
81
125
267
116
105
227 231
217
178
273
180
55
126
89
175
244
122
120
81
105
104
228
208
163
277
245
115
129
88
131
271
194
234
115
118
82
125 123
261
115
109
229
221
185
248
125
126
79
135
274 277
109
107
228
218
157
172
Day$4 Day$8 Day$12 Day$16 Day$20 Day$24 Day$28
104
210
220
164
274
175 180
53
177
245
135
105
81
127
273
110
7
8
1
2
3
4
5
6
53
110
105
215
209
163
274 281
168
54
157
269
172
55
172
54
Table 3.1. A total of 19,773 spine observations were made throughout this study.
15 neurons in 8 mice were imaged (1-3 neurons per mouse). All spines present on
Day 0 were tracked across 28 days. The fates of all new spines formed after Day 0
were also tracked during all remaining imaging sessions.
Table 3.1
56
Sensory learning does not induce the formation of new spines in clusters
Next, we examined whether learning induced the arrangement of new spines into
clusters. We used neighborhood analysis to determine whether learning promoted the
emergence of new spines in a temporally and spatially restricted manner (Fu et al.,
2012). A cluster was defined as the formation of 2 or more spines next to each other
(without an interspersed spine) within the same imaging interval. We examined where
spines were formed along the dendritic shaft before and during gap cross training (Fig.
3.6A). Clustering already occurs under baseline conditions, but we observed no
increase in the formation of clustered spines in any imaging interval during learning (P =
0.58; Fig. 3.6B). We also examined whether being part of a cluster improved the
survival rate of a new spine. The fraction of newly established clustered spines that
were maintained for longer than 4 days remained unchanged throughout baseline and
learning conditions (P = 0.55; Fig. 3c). These results suggest that learning neither
contributes to the clustered formation of synapses nor improves the survival of newly
formed spines that appear in clusters.

57

4 8 16 20 24
0
20
40
60
80
100
Day
Clustered spines (%)
4 8 16 20 24
0
20
40
60
80
100
Day
Clustered spines 4+ day (%)
A
B C
Figure 3.6. Learning on the gap crossing task does not increase the formation of
dendritic spine clusters in barrel cortex. A, Example stretch of dendrite from imaging
session 1. Colored dots indicate the locations of spines formed during baseline
(imaging sessions 2 (blue) and 3 (green)) and learning (imaging sessions 5 (yellow),
6 (orange), and 7 (red)). B, There is no increase in the percentage of spines that
appear in clusters during learning (n=8, RM one-way ANOVA). C, There is no change
in the fraction of clustered spines lasting 4+ days (n=8, RM one-way ANOVA). Data
presented as mean + s.e.m.
58
Discussion

In motor cortex, apical dendritic spines of layer 5 pyramidal neurons exhibit
extensive remodeling during acquisition of a motor task (Xu et al., 2009; Yang et al.,
2009; Fu et al., 2012). In order to investigate whether such changes also occur in
sensory cortex with the acquisition of a sensory based learning task, we imaged the
same population of dendritic spines over a month long period before, during, and after
gap cross training. Our results suggest that contrary to motor cortex, layer 5 neurons in
barrel cortex do not undergo extensive dendritic spine remodeling during tactile learning.
Learning on the gap crossing task did not induce dramatic changes in spine formation,
elimination, or total number. No increase in the formation of clustered spines was
observed with learning, further suggesting that coordinated spinogenesis may be a
phenomenon restricted to motor cortex.
There are several possible reasons why the results of this study do not support
the results seen in those examining motor cortex. The first may be the type of
transgenic mouse used in this study, the GFP M-line, compared to that used in most
other imaging studies, the YFP H-line. Though both developed by Feng et al. (2000),
these different lines could target different cell populations within layer 5. Furthermore,
these different strains of transgenic mice also influence the timeline of the studies.
When using the GFP M-line, cranial windows are used because of the sparse
expression of fluorescently labeled neurons (Holtmaat et al., 2005; 2009). With the YFP
H-line, a higher density of neurons is fluorescently labeled and a smaller cortical region
59
may be imaged (Holtmaat et al., 2005). This allows for a transcranial, or “thin-skull”
imaging approach. The main advantage of a cranial window is that it is a permanent and
stable prep that allows repeated imaging; however it does come with a more difficult
surgery and lower success rate (Holtmaat et al., 2009; Xu et al., 2007). On the other
hand, the simpler transcianial approach requires repeated thinning of the skull which
results in the skull becoming opaque, and thus degrades image quality after a few
imaging sessions (Yang et al., 2010). For this study, the first imaging session is 3 days
after the introduction of task, compared to other studies where imaging occurs within
hours-2 days. It’s possible that rapidly occurring changes in spine dynamics are missed.
However, this study aimed to investigate the anatomical changes underlying learning a
task that requires a number of sessions to see improvement and to examine the long-
term progression and lasting changes to circuitry. This is more reflective of learning,
versus any experience-dependent plasticity that may be more rapidly or immediately
occurring upon changes to sensory experience.
Most studies examining learning-related spine dynamics utilize motor cortex,
which is functionally unique and differently organized compared to sensory cortices. The
primary role of barrel cortex is to receive and process tactile information. Thus, it has a
very functionally distinct role from motor cortex; in particular, layer 5 serves as a major
output layer to drive movement (Oswald et al., 2013). Furthermore, the motor tasks
used in these studies involve performing very stereotypic movements that were
repeated over and over again. This type of behavior may indeed induce large changes
in dendritic spine dynamics, particularly in a motor output layer. Conversely, the gap
60
crossing task is a much more dynamic learning paradigm.
Most studies that have examined learning in barrel cortex have exclusively used
a head-fixed, spared-whisker learning paradigm (Xu et al., 2012; Kuhlman et al., 2014;
Peron et al., 2015). A head-fixed approach is known to change whisking behavior
(Voigts et al. 2015), as it prevents exploratory head motion and the accompanying
modulation in whisking (Towal and Hartmann, 2006). Exploratory whisking (rhythmic
whisker protractions) does not follow a fixed pattern (Voigts et al., 2008) and is
modulated by haptic feedback. A spared whisker protocol changes the premise of
whisker use, as the removal of all but one or a few whiskers alters the input pattern of
tactile information in barrel cortex (Diamond et al., 1993; Fox 2002, 2008). It examines
sensory information encoded by a few sensory organs, versus input and integration
across multiple whiskers. The gap crossing task is unique in that mice are freely
behaving with all whiskers intact; this provides a more ethologically relevant task
compared to head-fixed learning paradigms where mice are deprived of all but a few
whiskers. All behavior was spontaneous (unbaited), relying on the mouse’s natural
exploratory instinct.
At publication, the only other study examining anatomical plasticity in barrel
cortex during sensory learning focuses on layer 2/3 neurons (Kuhlman et al., 2014).
Though they did see an increase in spine turnover during learning, this may be a layer
specific phenomenon. Layer 2/3 serves as the main columnar input from layer 4 and
acts as a different level of information processing than layer 5. This suggests that
cortical layers may undergo discrete changes during learning. As they also utilized a
61
head-fixed, spared-whisker learning paradigm, differences in our findings may be
influenced by the specific learning task presented.
Improvement in performance results in enhanced stabilization of new spines and
a corresponding, but minor, decrease in spine turnover. Further, learning-induced
stabilization of newly formed spines strongly influences performance. The differences in
plasticity mechanisms observed here compared to motor cortex suggest that
functionally distinct areas of cortex are subject to different rules of plasticity, even within
the same cortical layer. We propose that sensory cortex is built to receive and integrate
sensory information and doesn’t require extensive changes in spine formation and
elimination to cope with changing sensory input. Neural circuitry must remain flexible
enough to allow for plasticity while maintaining largely stable synaptic connections.
Therefore, we conclude that barrel cortex remains fairly stable when presented with
ethologically relevant sensory information; however, sensory experience that leads to
improved performance on a learning paradigm leaves a lasting trace on neural circuitry
through the stabilization of newly formed spines.

62
CHAPTER FOUR
Conclusions

Structural plasticity through de novo growth and retraction of dendritic spines is a
proposed mechanism for the remodeling of synaptic structures and functional rewiring of
neural circuitry. We investigated the role of ngr1, a candidate gene proposed to restrict
anatomical plasticity, in tactile learning and basal cortical spine dynamics. Furthermore,
we examined the structural changes that occur during acquisition of a whisker-based
learning paradigm.
We observed no difference in the rate of learning between WT and ngr1-/- mice,
though ngr1-/- mice exhibit enhanced initial performance. We then examined their
baseline dendritic spine dynamics and found no difference in spine turnover, stability, or
survival between the two genotypes, indicating that better initial performance by ngr1-/-
mice cannot be explained by a difference in basal spine dynamics. Furthermore, the
deletion of ngr1 in adulthood did not alter spine dynamics during the decrease and
complete loss of NgR1. Thus we conclude that although ngr1 alters performance on a
tactile task, it does not restrict tactile learning or baseline cortical spine dynamics
Our findings indicate that barrel cortex does not undergo dramatic dendritic spine
turnover when challenged with a sensory-based learning task. Consistent with findings
in motor cortex, we observed an increase in new spine stability; however, we saw no
increase in percent spine gain or loss. There was no change in the percentage of
clustered spines during learning, and spines formed within a cluster did not exhibit
63
increased survival. Interestingly, new spine stability was significantly greater in mice
with better performance on the gap crossing task, indicating a critical role for the
maintenance of new spines formed during learning on performance.
As the function and organization of motor cortex is unique compared to sensory
cortex, it is conceivable that these areas undergo experience-dependent structural
plasticity in distinct ways. A key function of sensory cortex is to process incoming
sensory stimuli, and thus should remain fairly stable when presented with ethologically
relevant information. However, when repeated exposure to sensory experience results
in learning, the circuitry displays capacity for change in connectivity, resulting in a
behavioral outcome.
Here we have demonstrated that ngr1 does not limit sensory learning or baseline
cortical spine dynamics and that tactile learning results in enhanced new spine
stabilization in barrel cortex. These findings provide insight on the behavioral and
anatomical role of ngr1 in barrel cortex, as well as advance our understanding of how
tactile learning influences anatomical plasticity in sensory cortex.
64
REFERENCES
Akbik FV, Bhagat SM, Patel PR, Cafferty WBJ, Strittmatter SM (2013) Anatomical
! Plasticity of Adult Brain Is Titrated by Nogo Receptor 1. Neuron 77: 859–866.
Antonini A, Stryker MP (1993) Development of individual geniculocortical arbors in cat
! striate cortex and effecs of binocular impulse blockade. J Neurosci.13(8):
! 3549-73.
Barrette B, Vallières N, Dubé M, Lacroix S (2007) Expression profile of receptors for
! myelin-associated inhibitors of axonal regeneration in the intact and injured
! mouse central nervous system. Molecular and Cellular Neuroscience 34:
! 519–538.
Borrie SC, Baeumer BE, Bandtlow CE (2012) The Nogo-66 receptor family in the intact
! and diseased CNS. Cell Tissue Res 349: 105-117
Celikel T, Sakmann B (2007) Sensory integration across space and in time for decision
! making in the somatosensory system of rodents. Proc Natl Acad Sci USA 104:
! 1395–1400
Diamond ME, Armstrong-James M, Ebner FF (1993) Experience-dependent plasticity in
! adult rat barrel cortex. Proc Natl Acad Sci 90: 2081-2086.
Dickendesher TL, Baldwin KT, Mironova YA, Koriyama Y, Raiker SJ, et al. (2012) NgR1
! and NgR3 are receptors for chondroitin sulfate proteoglycans. Nature Publishing
! Group 15: 703–712.
Feng G, Mellor RH, Bernstein M, Keller-Peck C, Nguyen QT, et al. (2000) Imaging
! neuronal subsets in transgenic mice expressing multiple spectral variants of GFP.
! Neuron 28: 41–51.
Feldman DE (2009) Synaptic mechanisms for plasticity in neocortex. Annu Rev
! Neurosci 32:33-55.
Fournier AE, GrandPre T, Strittmatter SM (2001) Identification of a receptor mediating
! Nogo-66 inhibition of axonal regeneration.Nature 409(6818):341–346.
Fox K (1992) A critical period for experience-dependent synaptic plasticity in rat barrel
! cortex. J Neurosci 12:1826–1838.
Fox K (2002) Anatomical pathways and molecular mechanisms for plasticity in barrel
! cortex. Neuroscience 111(4): 799-814.
65
Fox K (2008) Experience-dependent plasticity mechanisms for neural rehabilitation in
! somatosensory cortex. Philos Trans R Soc Lond B Biol Sci 364(1515): 369-381.
Fu M, Yu X, Lu J, Zuo Y (2012) Repetitive motor learning induces coordinated formation
! of clustered dendritic spines in vivo. Nature 483(7387):92-5.
Gong S, Doughty M, Harbaugh CR, Cummins A, Hatten ME, et al. (2007) Targeting Cre
! recombinase to specific neuron populations with bacterial artificial chromosome
! constructs. J Neurosci 27: 9817–9823.
Grutzendler J, Kasthuri N, Gan WB (2002) Long-term dendritic spine stability in the
! adult cortex. Nature 420(6917):812-6.
Hand PJ. Plasticity of the rat barrel system. In: Morrison A.R., Strick P.L., editors.
! Changing Concepts of the Nervous System.Academic Press; New York, NY:
! 1982. pp. 49–68.
Harris KM, Jensen FE, Tsao B. 1992. Three-dimensional structure of dendritic spines
! and synapses in rat hippocampus (CA1) at postnatal day 15 and adult ages:
! implications for the maturation of synaptic physiology and long-term potentiation.
! J Neurosci 12:2685–2705.
Harris JA, Petersen RS, Diamond ME (1999) Distribution of tactile learning and its
! neural basis. Proc Natl Acad Sci USA 96: 7587–7591.
Hayashi S, McMahon AP (2002) Efficient recombination in diverse tissues by a
! tamoxifen-inducible form of Cre: a tool for temporally regulated gene
! activation/inactivation in the mouse. Developmental Biology 244: 305–318.
He Zm, Koprivica V (2004) The Nogo signaling pathway for regeneration block. Annu
! Rev Neurosci 27: 341–368.
Hofer SB, Mrsic-Flogel TD, Bonhoeffer T, Hübener M (2009) Experience leaves a lasting
! structural trace in cortical circuits. Nature 457: 313–317.
Hofer SB, Mrsic-Flogel TD, Bonhoeffer T, Hübener M (2006) Lifelong learning: ocular
! dominance plasticity in mouse visual cortex. Curr Opin Neurobiol 16(4):451-9.
Hollmann M, Heinemann S (1994) Cloned glutamate receptors. Annu Rev Neurosci
! 17:31-108.
Holtmaat A, Bonhoeffer T, Chow DK, Chuckowree J, De Paola V, et al. (2009) Long-
! term, high-resolution imaging in the mouse neocortex through a chronic cranial
! window. Nat Protoc 4: 1128–1144.
66
Holtmaat A, Trachtenberg JT, Wilbrecht L, Shepherd GM, Zhang X, et al. (2005)
! Transient and persistent dendritic spines in the neocortex in vivo. Neuron 45:
! 279–291.
Holtmaat A, Wilbrecht L, Knott GW, Welker E, Svoboda K (2006) Experience-dependent
! and cell-type-specific spine growth in the neocortex. Nature 441: 979–983.
Holtmaat A, Svoboda K (2009) Experience-dependent structural synaptic plasticity in
! the mammalian brain. Nat Rev Neurosci 10: 647–658.
Hutson KA and Masterton RB (1986) The sensory contribution of a single vibrissa's
! cortical barrel. J Neurophysiol 56: 1196–1223.
Jenkinson EW, Glickstein M (2000) Whiskers, barrels, and cortical efferent pathways in
! gap crossing by rats. J Neurophysiol 84(4):1781-9.
Josephson A, Trifunovski A, Scheele C, Widenfalk J, Wahlestedt C, Brene S, Olson L,
! Spenger C (2003) Activity-induced and developmental downregulation of the
! Nogo receptor. Cell Tissue Res 311: 333-342.
Jung CK, Herms J (2014) Structural dynamics of dendritic spines are influenced by an
! environmental enrichment: an in vivo imaging study. Cereb Cortex 24(2):377-84.
Kaas JH, Merzenich MM, Killackey HP (1983) The reorganization of somatosensory
! cortex following peripheral nerve damage in adult and developing mammals.
! Annu Rev Neurosci 6:325–356.
Kalatsky VA, Stryker MP (2003) New paradigm for optical imaging: temporally encoded
! maps of intrinsic signal. Neuron 38: 529–545.
Karlen A, Karlsson TE, Mattsson A, Lundstromer K, Codeluppi S, Pham TM, Backman
! CM, Ogren SO, Aberg E, Hoffman AF, Sherling MA, et al. Nogo receptor 1
! regulates formation of lasting memories. Proc Natl Acad Sci U S A.2009;106(48):
! 20476–20481.
Keuroghlian AS, Knudsen EI (2007) Adaptive auditory plasticity in developing and adult
! animals. Prog Neurobiol 82(3):109-21.
Kim JE, Liu BP, Park JH, Strittmatter SM (2004) Nogo-66 receptor prevents raphespinal
! and rubrospinal axon regeneration and limits functional recovery from spinal cord
! injury. Neuron 44: 439–451.
Knott GW, Holtmaat A, Wilbrecht L, Welker E, Svoboda K (2006) Spine growth
! precedes synapse formation in the adult neocortex in vivo. Nat Neurosci
! 9(9):1117-24.
67
Lee LJ, Erzurumlu RS (2005) Altered parcellation of neocortical somatosensory maps in
! NMDA receptor-deficient mice. J Comp Neurol 485: 57-63.
Lee H, Raiker SJ, Venkatesh K, Geary R, Robak LA, Zhang Y, Yen HH, Shrager P,
! Giger RJ (2008) Synaptic function for the Nogo-66 receptor NgR1: regulation of
! dendritic spine morphology and activity-dependent synaptic strength. J Neurosci
! 28(11):2753–2765.
Liu BP, Fournier A, GrandPre T, Strittmatter SM (2002) Myelin-associated glycoprotein
! as a functional ligand for the Nogo-66 receptor. Science 297(5584):1190–1193.
McGee AW, Yang Y, Fischer QS, Daw NW, Strittmatter SM (2005) Experience-driven
! plasticity of visual cortex limited by myelin and Nogo receptor. Science 309:
! 2222–2226.
Mizrahi A, Katz LC (2003) Dendritic stability in the adult olfactory bulb. Nat Neurosci
! 6(11):1201-7.
Mostany R, Anstey JE, Crump KL, Maco B, Knott G, et al. (2013) Altered Synaptic
! Dynamics during Normal Brain Aging. Journal of Neuroscience 33: 4094–4104.
Oswald MJ, Tantirigama MLS, Sonntag I, Hughes SM, Empson RM (2013) Diversity of
! layer 5 projection neurons in the mouse motor cortex. Front Cell Neurosci 7: 174.
Peron SP, Freeman J, Iyer V, Guo C, Svoboda K (2015) A cellular resolution map of
! barrel cortex activity during tactile behavior. Neuron 86(3): 783-99.
Pologruto TA, Sabatini BL, Svoboda K (2003) ScanImage: Flexible software for
! operating laser scanning microscopes. BioMed Eng OnLine 2: 13.
Schwab ME, Strittmatter SM (2014) Nogo limits neural plasticity and recovery from
! injury. Curr Opin Neurobiol 0: 53-60.
Stephany CE, Chan LLH, Parivash SN, Dorton HM, Piechowicz M, et al. (2014)
! Plasticity of Binocularity and Visual Acuity Are Differentially Limited by Nogo
! Receptor. Journal of Neuroscience 34: 11631–11640.
Tailby C, Wright LL, Metha AB, Calford MB (2005) Activity-dependent maintenance and
! growth of dendrites in adult cortex. Proc Natl Acad Sci USA 102:4631–4636.
Trachtenberg JT, Chen BE, Knott GW, Feng G, Sanes JR, et al. (2002) Long-term in
! vivo imaging of experience-dependent synaptic plasticity in adult cortex. Nature
! 420: 788–794.
68
Venkatesh K, Chivatakarn O, Lee H, Joshi PS, Kantor DB, Newman BA, Mage R, Rader
! C, Giger RJ (2005) The Nogo-66 receptor homolog NgR2 is a sialic acid-
! dependent receptor selective for myelin-associated glycoprotein. J Neurosci
! 25(4):808–822.
Voigts J, Herman DH, Celikel T (2015) Tactile object localization by anticipatory whisker
! motion. J Neurophysiol 113(2):620-32.
Voigts J, Sakmann B, Celikel T (2008) Unsupervised Whisker Tracking in Unrestrained
! Behaving Animals. J Neurophysiol 100: 504–515.
Wang KC, Koprivica V, Kim JA, Sivasankaran R, Guo Y, Neve RL, He Z (2002)
! Oligodendrocyte-myelin glycoprotein is a Nogo receptor ligand that inhibits
! neurite outgrowth. Nature 417(6892):941–944.
Wang X, Duffy P, McGee AW, Hasan O, Gould G, et al. (2011) Recovery from chronic
! spinal cord contusion after Nogo receptor intervention. Ann Neurol 70: 805–821.
Wilbrecht L, Holtmaat A, Wright N, Fox K, Svoboda K (2010) Structural Plasticity
! Underlies Experience-Dependent Functional Plasticity of Cortical Circuits.
! Journal of Neuroscience 30: 4927–4932.
Wills ZP, Mandel-Brehm C, Mardinly AR, McCord AE, Giger RJ, et al. (2012) The Nogo
! Receptor Family Restricts Synapse Number in the Developing Hippocampus.
! Neuron 73: 466–481.
Wisden W, Seeburg PH (1993) Mammalian ionotropic glutamate receptors. Curr Opin
! Neurobiol 3(3):291-8.
Woolsey TA, Van der Loos H (1970) The structural organization of layer IV in the
! somatosensory region (S1) of mouse cerebral cortex: The description of a
! cortical field composed of discrete cytoarchitectonic units. Brain Res 17(2)
! 205-42.
Xu NL, Harnett MT, Williams SR, Huber D, O’Connor DH, Svoboda K, Magee JC (2012)
! Nonlinear dendritic integration of sensory and motor input during an active
! sensing task. Nature 492(7428): 247-51.
Xu HT, Pan F, Yang G, Gan WB (2007) Choice of cranial window type for in vivo
! imaging affects dendritic spine turnover in the cortex. Nat Neurosci 10: 549–551.
Xu T, Yu X, Perlik AJ, Tobin WF, Zweig JA, et al. (2009) Rapid formation and selective
! stabilization of synapses for enduring motor memories. Nature 462: 915–919.
Yang G, Pan F, Gan W-B (2009) Stably maintained dendritic spines are associated with
! lifelong memories. Nature 462: 920–924.
69
Yang G, Pan F, Parkhurst CN, Grutzendler J, Gan W-B (2010) Thinned-skull cranial
! window technique for long-term imaging of the cortex in live mice. Nat Protoc 5:
! 213–220.
Zagrebelsky M, Schweigreiter R, Bandtlow CE, Schwab ME, Korte M. Nogo-a stabilizes
! the architecture of hippocampal neurons. J Neurosci. 2010;30(40):13220–13234.
Zito K, Scheuss V, Knott G, Hill T, Svoboda K (2009) Rapid functional maturation of
! nascent dendritic spines. Neuron 61(2):247-58.
Zuo Y, Lin A, Chang P, Gan W-B (2005a) Development of long-term dendritic spine
! stability in diverse regions of cerebral cortex. Neuron 46: 181–189
Zuo Y, Yang G, Kwon E, Gan W-B (2005b) Long-term sensory deprivation prevents
! dendritic spine loss in primary somatosensory cortex. Nature 436: 261–265.
70

APPENDIX
Autophosphorylation at the aCaMKII Threonine-286 site is essential for sensory-based
learning

Abstract

Alpha calcium/calmodulin-dependent protein kinase II (aCaMKII) is a major
synaptic protein that has been implicated in synaptic plasticity, learning, and memory. In
particular, its autonomous (calcium-independent) activity induced by
autophosphorylation at the Threonin-286 site plays a crucial role in plasticity and
cognition, as demonstrated by mutant mice. Mice lacking the ability to
autophosphorylate at the T-286 site display impaired hippocampal and neocortical
plasticity, as well as deficits in hippocampus-based learning tasks. To determine if
sensory learning is intact in T286A mutant mice, we trained them on the gap crossing
task, a whisker input- and barrel cortex-dependent learning paradigm. T286A mutants
display significant deficits on this task, even under rewarded conditions. Thus
autophosphorylation at the Thr-286 site is critical for tactile learning.

71

Introduction

aCaMKII is a highly abundant kinase that has long been shown to be critical for
synaptic plasticity, learning, and memory. It is large holoenzyme consisting of 12
subunits (Chao et al., 2011; Colbran, 2004) and is a major synaptic protein (Chao et al.,
2011). The opening of NMDARs during LTP induction results in an influx of calcium.
This rise in intracellular calcium levels activates calmodulin and CaMKII translocates
from the cytoplasm to the synapse where it binds to NMDARs (and PSD). This results in
long lasting potentiation of the AMPAR-mediated EPSC by two mechanisms. First, it
increases the average single channel conductance via phosphorylation of GluR1.
Second, it increases the number of AMPARs at the synapse; phosphorylation of
stargazin and subsequent binding to PSD95 traps and recruits extrasynaptic AMPARs
(for a thorough review, please see Lisman et al., 2012). Importantly,
autophosphorylation at the Threonine-286 site allows aCaMKII activity to persist even
after calcium levels fall back to baseline, hence inducing an autonomous state (Miller et
al., 1986). For this reason, aCaMKII has been considered to act as a molecular switch
in activating LTP (Lisman et al., 1985).
Advances in genetic techniques have allowed the study of one facet of aCaMKII’s
activity without knocking out the entire gene. This is particularly useful because CaMKII
is a multi-functional kinase with multiple roles. Utilizing knock-in technology, Giese et al.
(1998) was able to mutate the endogenous aCaMKII gene and substitute Threonine-286
for alanine. This point mutation results in a functional enzyme that lacks the ability to
72

autophosphorylate. The use of this mutant mouse line has shown that induction of
cortical LTP in adults is dependent on aCaMKII and its autophosphorylation at Thr286
(Glazewski et al., 1996, 2000; Hardingham et al., 2003). aCaMKII-Thr286A (T286A)
mutant mice also exhibit impaired experience-dependent plasticity, showing no
response potentiation of spared whiskers after whisker trimming (Glazewski et al., 2000;
Hardingham et al., 2003). On a structural level, these mice show no increase in new
spine stabilization at the border between spared and deprived barrel columns. The
density of new spines did not differ from wild-type mice, suggesting that CaMKII
autophosphorylation at Thr-286 plays a role in stabilization, but not the formation, of
new spines (Wilbrecht et al., 2010). Although neocortical plasticity in sensory cortices
has been shown to be impaired in T286A mutant mice, how this impacts sensory
learning remains unknown.

73

Materials and methods

Mice
Straws from aCaMKII-T286A mutant mice were kindly provided by Dr. Kevin Fox
and the line was re-derived at The Jackson Laboratory. Progeny were genotyped as
previously described (Giese et al., 1998). Mice homozygous for the T286A point
mutation were subsequently used for experiments and their wildtype littermates served
as controls. Mice were group housed with same-sex littermates and food and water
were available ad libitum. In the motivated studies, mice were food deprived and
maintained at 90% body weight throughout the course of gap cross training. Mice were
weighed everyday to ensure they did not drop below 90% body weight. Only male mice
were used in this study.
Mice were maintained and all experiments conducted according to protocols
approved by the Children's Hospital Los Angeles Institutional Animal Care and Use
Committee. Mice were anesthetized by isoflurane inhalation and euthanized by carbon
dioxide asphyxiation in accordance with approved protocols.

The gap cross assay
The gap cross assay system is a closed-loop robotic environment with motor
controlled units and sensing elements. The mouse behaves upon raised platforms
driven by independent linear actuators. The platforms are equipped with servo-motor
doors and positional sensors. Data acquisition and control algorithms are both executed
74

online for real-time dynamic control and offline for more advanced analysis.
Independent linear actuators move the Plexiglass platforms to generate a range
of gap distances. To monitor the rodent motion four IR motion sensors are at the back
and edge of each platform. Near the edge of each platform are servo-controlled doors
that prevent exploratory behavior during repositioning of the platforms. The linear
motors, servos, and motion sensors are USB controlled through microcontroller boards
(Arduino Mega 2560 and the Quadstepper Motor Driver) that communicate with a quad-
core CPU.
Motor positions are processed on a quad-core CPU using the Arduino and
Matlab programming environments. Platform position, door status (open/closed) and
feeders are real-time controlled using the Arduino C-based development environment
(ADE). Motion sensor data are continuously acquired and pre-processed within ADE
and are visualized and stored in real time within Matlab via serial communication.
Specifically, sensor activity are encoded as behavioral performance metrics:
success/failed crossing events, 1) successful: animal approaches the gap and crosses
to back of target platform; or 2) failed: animal approaches the gap and then retreats to
back of home platform. This information is computed in real-time.
To control the positional and door motors, the GCS employs a closed-loop finite
state machine algorithm (D.H. Herman, unpublished observations). Animal behaviors
are segmented into interactive events at the gap. Consequently, the system is
structured as a two state machine: Exploration and Adjustment. During Exploration, the
motors are disabled and the system continuously acquires behavioral data via motion
75

sensors. During Adjustment, the doors close to halt exploration, and the motors
reposition the platforms for the next exploration phase as determined by the
programmed protocol. Transitions between the two states are triggered by behavioral
events (i.e. successful/failed gap-crossing).
Mice were 10–12 weeks old at the start of the task. Animals were handled for 10
minutes a day for one week prior to beginning the task. The day before training began,
mice were habituated to the gap cross apparatus by placing each mouse in the chamber
with background white noise for 20 minutes in white light immediately followed by 20
minutes in the dark. A bridge was placed over the gap to prevent exploration of the gap
and gap crossing behavior during habituation. Mice in the motivated group received
food reward pellets (LabDiet) in their cages for a few days before beginning the task to
familiarize them with a new food.
Mice were subsequently trained once per day. Each session lasted for 20
minutes or 20 successful crosses, whichever came first. All GC training took place in the
dark with background white noise to mask any visual or auditory cues. Mice in the
motivated group received a food reward pellet upon each successful cross. All sessions
began with a trial at 3.0 cm, the shortest distance tested. Throughout the session, mice
were presented with nose distances (3-4 cm) and whisker distances (5-6 cm) in 0.5 cm
increments. Position of the mouse was tracked with motion sensors placed at the back
and near the edge of each platform. As a mouse traversed the platform, these sensors
record its progressive position. A successful trial was identified as any trial in which the
mouse successfully crossed the gap between the home and target platforms and
76

activated the motion sensor at the back of the target platform. A failed attempt was
defined as an attempt in which the mouse explored the edge of the home platform and
returned to the back of the platform. Regardless of whether the mouse succeeded or
failed, the gap distance would chance for the next trial in order to prevent the mouse
from having any previous knowledge of the gap distance. The next distance was
determined with a learning algorithm that randomly chose the distance from a Gaussian
distribution centered a gap distance 0.5 cm longer than the previous distance if the
preceding trial were successful, and a gap distance 0.5 cm shorter if the preceding trial
were a failure (D.H. Herman, manuscript in preparation). This approach decreases the
predictability of the subsequent gap distance relative to a ‘laddering’ learning paradigm
in which the next distance increased or decreased by a set distance depending on the
success or failure in the preceding trial.

77

Results

In order to investigate whether autophosphorylation at the T286 site was required
for tactile learning, T286A mutant mice were tested on the gap cross assay. Mice were
acclimated to the device and then tested at a range of distances spanning 3.0 to 6.0 cm
at 0.5 increments. Mice received one session a day for 9 days.
In spontaneous (unbaited) gap crossing, T286A mutant mice displayed impaired
performance compared to their WT littermates (P < 0.0001, Fig. A.1A). Multiple
comparisons reveal a significant difference in performance between genotypes at 5.0
and 5.5 cm (Fig. A.1A). T286A mutant mice perform similarly to WT mice at nose
distances (3-4 cm), indicating they are capable of performing this task when they can
use both their whiskers and touch receptors in their nose to detect the presence and
determine the distance of the target platform. Success at these short distances start and
remain high, thus sensory experience is not necessary for improvement. However, at
longer whisker distances (5-6 cm) when they must rely solely on tactile input coming in
from their whiskers, the performance of T286A mutant mice suffers. Performance at
these whisker distances normally improves with experience. This suggests that even
with sensory experience, T286A mutants are unable to improve their likelihood of
successfully crossing at whisker distances.
Next, we examined whether food motivation would improve the performance of
T286A mutant mice. WT and T286A mutant mice were food restricted and maintained at
90% body weight. Upon a successful crossing, mice were presented with a food reward
78

pellet. Even with food motivation, T286A mutant mice continue to display a deficit in
performance compared to WT mice (P < 0.0001, Fig. A.1B). This difference in
performance was again seen at whisker distances 5.0 and 5.5 cm (Fig. A.1B). However,
it was possible that the performance of T286A mutant mice was improving with food
motivation, just not to extent of WT mice. To test this, we compared the performance of
T286A mutant mice during spontaneous and motivated gap crossing at whisker
distances. We observe no difference in performance of T286A mutant mice with food
motivation (P = 0.766; Fig. A.1C).

79

5 5.5 6
0
20
40
60
80
100
Gap distance (cm)
Percent success
Motivated
Spontaneous
A C
B
Figure A.1. T286A mutant mice
display impaired performance on
the gap crossing task A, T286A
mutant mice display impaired
learning at whisker distances (WT
n = 10, T286A n = 12, two-way
ANOVA followed by Bonferroni’s
multiple comparisons test, *P <
0.05, **P < 0.01). B, T286A
mutant mice do not improve
performance with food motivation
(WT n = 11, T286A n = 10, two-
way ANOVA followed by
Bonferroni’s multiple comparisons
test, ****P < 0.0001). C, There is
no difference in performance of
T286A mutant mice at whisker
distances with or without food
motivation (Spontaneous n = 12,
motivated n = 10, two-way
ANOVA). Data presented as
mean + s.e.m.
3 3.5 4 4.5 5 5.5 6
0
20
40
60
80
100
Gap distance (cm)
Percent success
WT
T286A
**
*
3 3.5 4 4.5 5 5.5 6
0
20
40
60
80
100
Gap distance (cm)
Percent success
WT
T286A
****
****
80

Discussion

Mice lacking the ability to autophosphorylate at the aCaMKII-T286 site show
deficits in learning and memory. However, most studies examining learning in T286A
mutant mice have utilized hippocampus-dependent tasks (Giese et al., 1998; Elgersma
et al., 2002; Need and Giese 2003) although these mutant mice also display impaired
experience-dependent plasticity in sensory cortex (Glazewski et al., 2000; Hardingham
et al., 2003; Wilbrecht et al., 2010). To determine whether autophosphorylation at the
T286 site is also important for sensory-dependent learning, T286A mutant mice were
tested on the gap cross assay. We observe impaired performance of T286A mutant
mice at whisker distances that cannot be rescued using food motivation.
T286A mutant mice exhibit a deficit in spatial learning that cannot be rescued by
overtraining (Giese et al., 1998; Elgersma et al., 2002). Enriched environment, a
paradigm that has been repeatedly shown to improve learning and memory, cannot
rescue spatial learning deficits in these mutant mice (Need and Giese, 2003). T286A
mutants probably don’t have visual, motivational, or motor differences, as they
performed similarly to wildtype mice on the Morris water maze with a visible platform
(Need and Giese, 2003). Furthermore, T286A mutants exhibit normal performance at
shorter gap distances, indicating they are capable of gap crossing behavior. Wilbrecht
et al. (2010) demonstrated that T286A mutant mice exhibit impaired experience-
dependent new spine stabilization. As our previous work demonstrates an important role
for new spine stabilization during learning on the gap crossing task (Park and McGee, in
81

preparation), it is possible that the observed learning deficit may be in part due to
impaired structural plasticity. We conclude that autophosphorylation at the aCamKII-
T286 site is critical to tactile learning.

82

References

Chao LH, Stratton MM, Lee IH, Rosenberg OS, Levitz J, Mandell DJ, Kortemme T,
Groves JT, Schulman H, Kuriyan J (2011) A mechanism for tunable autoinhibition
in the structure of a human Ca2+/calmodulin-dependent kinase II holoenzyme.
Cell, 146(5): 732-45.

Colbran RJ (2004) Protein phosphatases and calcium/calmodulin-dependent protein
kinase II-dependent synaptic plasticity. J Neurosci 24: 8404-8409.

Elgersma Y, Fedorov NB, Ikonen S, Choi ES, Elgersma M, Carvalho OM, Giese KP,
Silva AJ (2002) Inhibitory autophosphorylation of CaMKII controls PSD association,
plasticity, and learning. Neuron 36: 493-505.

Giese KP, Fedorov NB, Filipkowski RK, Silva AJ (1998) Autophosphorylation at Thr286
of alpha calcium-calmodulin kinase II in LTP and learning. Science 279: 870-873.

Glazewski S, Chen CM, Silva A, Fox K (1996) Requirement for alpha-CaMKII in
experience-dependent plasticity of the barrel cortex. Science 272: 421-423.

Glazewski S, Giese KP, Silva A, Fox K (2000) The role of alpha-CaMKII
autophosphorylation in neocortical experience-dependent plasticity. Nat Neurosci
3: 911-918.

Hardingham N, Glazewski S, Pakhotin P, Mizuno K, Chapman PF, Giese KP, Fox K
(2003) Neocortical long-term potentiation and experience-dependent synaptic
plasticity require α-calcium/calmodulin-dependent protein kinase II
autophosphorylation. J Neurosci 23: 4428-4436.

Lisman JE (1985) A mechanism for memory storage insensitive to molecular turnover: a
bistable autophosphorylating kinase. Proc Natl Acad Sci USA 82: 3055-3057.

Lisman J, Yasuda R, Raghavachari S (2012) Mechanisms of CaMKII action in long-term
potentiation. Nat Rev Neurosci 13:169–182.

Miller SG, Kennedy MB (1985) Distinct forebrain and cerebellar isozymes of type II
Ca2+/calmodulin-dependent protein kinase associate differently with the
postsynaptic density fraction. J Biol Chem 260:9039–9046.

Need AC, Giese KP (2003) Handling and environmental enrichment do not rescue
learning and memory impairments in aCamKIIT286A mutant mice. Genes, Brain
and Behavior 2: 132-139.
83

Wilbrecht L, Holtmaat A, Wright N, Fox K, Svoboda K (2010) Structural Plasticity
Underlies Experience-Dependent Functional Plasticity of Cortical Circuits.
Journal of Neuroscience 30: 4927–4932.
84

Abstract (if available)

Abstract Structural plasticity through de novo growth and retraction of dendritic spines is a proposed mechanism for the remodeling of synaptic structures and functional rewiring of neural circuitry. We investigated the role of ngr1, a candidate gene proposed to restrict anatomical plasticity, in tactile learning and basal cortical spine dynamics. Furthermore, we examined the structural changes that occur during acquisition of a whisker-based learning paradigm. Here we have demonstrated that ngr1 does not limit sensory learning or baseline cortical spine dynamics and that tactile learning results in enhanced new spine stabilization in barrel cortex. These findings provide insight on the behavioral and anatomical role of ngr1 in barrel cortex, as well as advance our understanding of how tactile learning influences anatomical plasticity in sensory cortex.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Differential regulation of anatomical and functional visual plasticity by NgR1

PDF

Characterizing response and plasticity in sensory cortices of the Fmr1⁻⁄⁻ mouse

PDF

Imaging neuromodulator dynamics in somatosensory cortex

PDF

The effect of treadmill running on dendritic spine density in two neurodegenerative disorders: 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) mouse model of Parkinson’s disease and CAG₁₄₀ kn...

PDF

Experimental and computational explorations of different forms of plasticity in motor learning and stroke recovery

PDF

The roles of TRPM8 in cold sensation: the six sides of TRPM8

PDF

Functional properties of the superficial cortical interneurons

PDF

Plasticity in CMOS neuromorphic circuits

PDF

The role of the cofilin/Limk1 signaling pathway in axon growth during development and regeneration

PDF

Motor cortical representations of sensorimotor information during skill learning

PDF

Quantification and modeling of sensorimotor dynamics in active whisking

PDF

The selective role of GFRα3 in cold pain

PDF

Characterization of visual cortex function in late-blind individuals with retinitis pigmentosa and Argus II patients

PDF

Contextual modulation of sensory processing via the pulvinar nucleus

PDF

Functional circuits underlying sensory representation in mouse primary auditory cortex

PDF

Molecular diversity of neurons in developing and mature brain circuits

PDF

The expected and unexpected roles of TRPM8: cold pain and metabolism

PDF

Mapping multi-scale connectivity of the mouse posterior parietal cortex

PDF

Neural mechanisms of sensorimotor learning in cortico-basal ganglia pathways

PDF

Synaptic mechanism underlying development and function of neural circuits in rat primary auditory cortex

Asset Metadata

Creator Park, Jennifer I. (author)

Core Title Sensory learning and anatomical plasticity in barrel cortex

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Defense Date 12/01/2015

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

Tag barrel cortex,dendritic spines,gap cross,Learning,OAI-PMH Harvest,plasticity,Whiskers

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McKemy, David (committee chair), Bottjer, Sarah (committee member), McGee, Aaron (committee member), Pike, Christian (committee member)

Creator Email jen.park3@gmail.com,parkji@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-218256

Unique identifier UC11277114

Identifier etd-ParkJennif-4174.pdf (filename),usctheses-c40-218256 (legacy record id)

Legacy Identifier etd-ParkJennif-4174.pdf

Dmrecord 218256

Document Type Dissertation

Format application/pdf (imt)

Rights Park, Jennifer I.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA