Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Activity dependent synaptic plasticity and microfluidics Coquinco, Ainsley 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2012_spring_coquinco_ainsley.pdf [ 5.12MB ]
Metadata
JSON: 24-1.0072583.json
JSON-LD: 24-1.0072583-ld.json
RDF/XML (Pretty): 24-1.0072583-rdf.xml
RDF/JSON: 24-1.0072583-rdf.json
Turtle: 24-1.0072583-turtle.txt
N-Triples: 24-1.0072583-rdf-ntriples.txt
Original Record: 24-1.0072583-source.json
Full Text
24-1.0072583-fulltext.txt
Citation
24-1.0072583.ris

Full Text

Activity Dependent Synaptic Plasticity and Microfluidics     by   AINSLEY COQUINCO    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE STUDIES  (Neuroscience)      THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)            February 2012  ©Ainsley Coquinco 2012  ii Abstract  The visual cortex of the brain is one of the fundamental preparations to study critical periods and activity dependent changes in the brain.  During development, when sensory input from one eye is prevented, visual acuity and brain connectivity is lost in favour of inputs from the active eye. Because of the brain’s complexity, it is difficult to perform thorough analyses of synaptic mechanisms that exist during development.   Therefore, the development of simpler in vitro models would be advantageous.  In our studies, we used a 3-compartment microfluidic device and created a new model for dual input in vitro activity dependent synaptic plasticity. Microfluidics offered the advantage of being able to physically and chemically isolate neurons in distinct environments.  In chapter 2, we optimized previously developed microfluidic devices for use in our cell culture experiments and demonstrated that their application can create a dual input activity dependent system.  Using a 3-compartment microfluidic device, the activity of one neuronal group was reduced by application of tetrodotoxin or the GABA agonist, muscimol.  Treatment caused the formation of a greater number of synaptic contacts between the target ‘postsynaptic’ neurons and the ‘presynaptic’ inputs at normal working activity levels compared to the opposing ‘presynaptic’ inputs with reduced activity.  In chapter 3, we established that ‘critical periods’ exist in our in vitro model by varying the ages at which we reduced neuronal input activity.  Muscimol treatment had an earlier time window to induce activity dependent synaptic changes compared to inhibition by tetrodotoxin.  By the fourth week in culture, neither treatment induced any synaptic difference between inputs.  In chapter 4, we examined the mechanisms involved in our model.  We manipulated NMDAR activity, CamKII activity, or GluR2 internalization postynaptically under the same conditions that we previously established.  In our model, both treatments were NMDAR activity dependent while the requirement for CamKII activity and GluR2 internalization was dependent on the application of either muscimol or tetrodotoxin respectively.  Taken together, we showed the ability to create a new in vitro model for activity dependent synaptic plasticity and that even in a simple system multiple mechanisms can exist.  iii Preface  This research was performed in collaboration with the lab of Dr. Noo Li Jeon.  His lab was responsible for developing and creating the microfluidic devices used in this dissertation.  I further optimized the protocol to use the devices for the experiments performed and detailed here.  Plasmid constructs were generously donated from the lab of Dr. Anne Marie Craig.  The GluR3y and associated scramble peptide were generously provided by the lab of Dr. Yu Tian Wang.  Preparation of the dissociated primary neuron cultures were done by Dr. Wendy Wen.  All animal experiments were approved by the Animal Care Committee of the University of British Columbia (Protocol A10-0118; A06-0173; Certificates RBH-634-09, RA-367-09, RSHX- 280-09, 1097).  iv Table of Contents  Abstract .......................................................................................................................................... ii  Preface ........................................................................................................................................... iii  Table of Contents ......................................................................................................................... iv  List of Tables ............................................................................................................................... vii  List of Figures ............................................................................................................................. viii  Acknowledgements ....................................................................................................................... x  1. Introduction ............................................................................................................................... 1  1.1. Critical Periods and the Visual Cortex ................................................................................ 1  1.1.1. The Visual Cortex ................................................................................................................. 4  1.1.2. Ocular Dominance Columns and Plasticity .......................................................................... 5  1.1.3. Visual Cortex and NMDA Receptor Activity ....................................................................... 6  1.1.4. The Visual Cortex and Neurotrophins .................................................................................. 9  1.1.5. The Visual Cortex and Synaptic Plasticity ......................................................................... 10  1.2. The Synapse .......................................................................................................................... 13  1.2.1. Identifying a Synapse .......................................................................................................... 15  1.2.2. Presynaptic Assembly ......................................................................................................... 17  1.2.3. Postsynaptic Assembly ....................................................................................................... 18  1.2.4. Synapses and Activity ......................................................................................................... 19  1.2.5. Synapse Elimination ........................................................................................................... 21  1.3. Existing In Vitro Models of Activity Dependent Synaptic Plasticity ............................... 23  1.3.1. The Neuromuscular Junction (NMJ) .................................................................................. 24  1.3.2. Cerebellar Purkinje Cells .................................................................................................... 26  1.3.3. Genetic Modification of Individual Neurons ...................................................................... 29  1.3.4. Micro-Island Pairs ............................................................................................................... 31  1.3.5. Microfluidic Devices .......................................................................................................... 32  1.4. Objectives .............................................................................................................................. 34  2. An In Vitro Dual Input Activity Dependent Synaptic Plasticity Model Using Microfluidics ................................................................................................................................ 36  2.1. Introduction .......................................................................................................................... 36  2.2. Optimization of 3-Compartment Microfluidic Devices for Neuronal Cell Culture ...... 38  2.2.1. Materials ............................................................................................................................. 38  2.2.2. Procedure ............................................................................................................................ 40  2.2.3. Troubleshooting .................................................................................................................. 41   v 2.3. Procedure for In Vitro Dual Input Activity Dependent Synaptic Plasticity Model ....... 42  2.3.1. Treatment of Neurons ......................................................................................................... 42  2.3.2. Analysis of Treatment Effects ............................................................................................ 43  2.4. Anticipated Results .............................................................................................................. 44  3. A Critical Period Exists In Vitro in Our Model of Activity Dependent Synaptic Plasticity  ....................................................................................................................................................... 47  3.1. Introduction .......................................................................................................................... 47  3.2. Results ................................................................................................................................... 48  3.2.1. Nonspecific Silencing of Presynaptic Neuronal Activity ................................................... 49  3.2.2. Inhibition of Presynaptic Neuronal Activity with GABA Agonist, Muscimol .................. 56  3.2.3. The Physical Properties of the Microfluidic Device Do Not Affect the ‘Presynaptic’ Groups ........................................................................................................................................... 62  3.2.4. Cross Group Effects of Unilateral Inhibition ...................................................................... 67  3.2.5. Activity Inhibition Does Not Change Presynaptic or Synaptic Density on the Axon ........ 70  3.2.6. Age Dependent Plasticity .................................................................................................... 73  3.3. Discussion .............................................................................................................................. 80  4. Cellular and Molecular Mechanisms of Our In Vitro Activity Dependent Synaptic Plasticity Model Using Microfluidics ........................................................................................ 84  4.1. Introduction .......................................................................................................................... 84  4.1.1. The NMDA Receptor .......................................................................................................... 85  4.1.2. CamKII ............................................................................................................................... 86  4.1.3. The AMPA Receptor .......................................................................................................... 86  4.2. Results ................................................................................................................................... 87  4.2.1. Activity Dependent Changes Caused by TTX or Muscimol Are NMDA Receptor (NMDAR) Dependent ................................................................................................................... 88  4.2.2. Activity Dependent Changes Caused by Muscimol Requires CamKII Activity ................ 91  4.2.3. AMPA Receptor (AMPAR) Internalization is Required for Activity Dependent Changes Caused by TTX Inhibition ............................................................................................................ 94  4.2.4. Cellular Mechanisms .......................................................................................................... 98  4.3. Discussion ............................................................................................................................ 102  5. General Discussion ................................................................................................................ 108  5.1. A Novel Model for Activity Dependent Synaptic Plasticity In Vitro ............................. 108  5.2. Molecular Mechanisms of Our In Vitro Model ............................................................... 109  5.3. Activity Inhibition by TTX Reduces the Axon Growth Rate of Inhibited Neurons. Activity Inhibition by GABAR Activation (Muscimol) Increased the Axon Growth Rate of Uninhibited Neurons ................................................................................................................. 112  5.4. The Postsynaptic Age is a Greater Limiting Factor for Our In Vitro Critical Period 113  5.5. TTX versus Muscimol Inhibition ..................................................................................... 114  5.6. Future Directions and Applications ................................................................................. 116  5.7. Concluding Remarks ......................................................................................................... 118   vi References .................................................................................................................................. 120  Appendix .................................................................................................................................... 133  A. Calibration Curves ............................................................................................................... 133    vii List of Tables  Table 1: Summary of the observed axon changes in a field observed over 24 hours under various treatment conditions .............................................................................................................. 99    viii List of Figures  Figure 1: Illustration of the visual zones present in the mouse model ............................................ 5  Figure 2: Electron Microscopy image of a mature synapse .......................................................... 17  Figure 3:  A neuromuscular synapse ............................................................................................. 25  Figure 4: Illustration of a correctly innervated Purkinje cell. ....................................................... 27  Figure 5:  Schematic diagram on the process of hindbrain explant experiments ......................... 28  Figure 6: Schematic diagram of neuronal micro island pairs ....................................................... 31  Figure 7: Diagram of Campenot Chamber .................................................................................... 33  Figure 8: Schematic of two compartment microfluidic device ..................................................... 34  Figure 9:  3-compartment microfluidic device ............................................................................. 39  Figure 10:  Sample dual input activity dependent synaptic plasticity model results .................... 46  Figure 11:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment, DIV 8 - 10). ................................................. 51  Figure 12:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment, DIV 12 – 14) ................................................ 52  Figure 13:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment, DIV 19 – 21). ............................................... 54  Figure 14:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment, DIV 26-28). ................................................. 55  Figure 15:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment, DIV 8 – 10) .......................................... 57  Figure 16:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment, DIV 12 – 14). ....................................... 58  Figure 17:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment, DIV 19 – 21). ....................................... 60  Figure 18:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment, DIV 26 – 28) ........................................ 61  Figure 19: Comparison of the number of synapses between two sets of afferents from two different untreated presynaptic groups, DIV 8 - 10 .............................................................. 63  Figure 20:  Comparison of the number of synapses between two sets of afferents from two different untreated presynaptic groups, DIV 12 - 14. ........................................................... 64  Figure 21: Comparison of the number of synapses between two sets of afferents from two different untreated presynaptic groups, DIV 19 - 21 ............................................................ 65  Figure 22: Comparison of the number of synapses between two sets of afferents from two different untreated presynaptic groups, DIV 26 - 28 ............................................................ 66  Figure 23: Percentage of total PSD-95 colocalized with labeled synaptophysin from the ‘presynaptic’ groups, DIV 8 - 10 .......................................................................................... 67  Figure 24:  Percentage of total PSD-95 colocalized with labeled synaptophysin from the ‘presynaptic’ groups, DIV 12 - 14. ....................................................................................... 68  Figure 25: Percentage of total PSD-95 colocalized with labeled synaptophysin from the ‘presynaptic’ groups, DIV 19 - 21 ........................................................................................ 69  Figure 26: Comparison of presynaptic puncta density or synapse density, DIV 8 – 10 ............... 71  Figure 27: Comparison of presynaptic puncta density or synapse density, DIV 12 – 14 ............. 72  Figure 28:  Comparison of presynaptic puncta density or synapse density, DIV 19 – 21 ............ 73  Figure 29: Schematic of 3-compartment diagram set up for mixed age cultures. ........................ 74  Figure 30: Comparison of the number of synapses between two sets of afferents from two different presynaptic groups with an older postsynaptic target (TTX treatment) ................. 75   ix Figure 31:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups with an older postsynaptic target (muscimol treatment) ......... 76  Figure 32:  Comparison of the number of synapses between two sets of afferents from two different older presynaptic groups with a younger postsynaptic target (TTX treatment) ..... 78  Figure 33:  Comparison of the number of synapses between two sets of afferents from two different older presynaptic groups with a younger postsynaptic target (muscimol treatment)  ............................................................................................................................................... 79  Figure 34:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment) at DIV 12 -14 with postsynaptic group inhibition of NMDAR activity .............................................................................................. 89  Figure 35:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment) at DIV 12 – 14 with postsynaptic group inhibition of NMDAR activity .............................................................................................. 90  Figure 36:  Comparison of presynaptic puncta density or synpase density (postsynaptic group treated with APV) ................................................................................................................. 91  Figure 37:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment) at DIV 12 – 14 with postsynaptic group inhibition of CamKII activity ................................................................................................ 92  Figure 38:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment) at DIV 12 – 14 with postsynaptic group inhibition of CamKII activity ................................................................................................ 93  Figure 39:  Comparison of presynaptic puncta density or synpase density (postsynaptic group treated with KN-93) .............................................................................................................. 94  Figure 40:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (TTX treatment) at DIV 12 – 14 with the postsynaptic group treated with GluR23Y ............................................................................................................. 95  Figure 41:  Comparison of the number of synapses between two sets of afferents from two different presynaptic groups (muscimol treatment) at DIV 12 – 14 with the postsynaptic group treated with GluR23y ................................................................................................... 96  Figure 42:  Comparison of presynaptic puncta density of presynaptic afferent sets (with the postsynaptic group treated with GluR23Y) ............................................................................ 97  Figure 43:   Comparison of synapse density of presynaptic afferent sets (with the postsynaptic group treated with GluR23Y) ................................................................................................. 98  Figure 44:  Comparison of axon growth rate between 2 presynaptic groups. ............................ 101  Figure 45:  Comparison of axon growth rate against control conditions .................................... 102 Appendix A 1 : Concentration curves for fluorescein and crystal violet .................................... 133      x Acknowledgements  The work presented here would not have been accomplished without the generous help from so many.  Many thanks go to my supervisor, Dr. Max Cynader, who (quite possibly despite his better judgment) provided me the opportunity and support to achieve my academic goals.  Dr. Shiv Prasad, the person who introduced and helped me enters into the world of neuroscience.  Dr. Luba Kojic, for his continued help, support and advice towards my research and all things in general.  To the members of my supervisory committee, Dr. Ann Marie Craig, Dr., Yu Tian Wang, Dr. William Jia and honorary member Dr. Lynn Raymond.  For the valuable support they have given me through their advice, criticisms and generous donations of reagents and equipment time.  Swarni Sunner, who is always there to help me get through the bureaucracy of academia and science.  Dr. Wendy Wen, whose gracious help with providing me constant supply of neurons, made all my experiments possible.  To Dr. Noo Li Jeon and his lab whose extremely generous gift of microfluidic devices still remains in full use to this day.  To Dr. Dong Qiang, Dr. Li Pang and Dr. Wei Xiong who taught me many of the procedures I use today when I first started working in the lab.  To my friends, Dr. Austen Milnerwood, Dr. Sesath “Jiggawatts” Hewapatharine, Federal Agent Dr. Josh Levinson, The Earl of Dunfield, Philosopher Soroush Moghaddam, Dr. Dustin Hines, Alan Huang, Jedi Knight Ted Dobie, Fiona “Rat Whisperer” Choi, Dr. Soyon Ahn, Simon Chen, Edmund Lo, Dr. Herman Fernandes, Kevin She, Dr. Wei Xiong, Rick Kornelson, Teasup Cho, Carlos Smith, Kaiyun Yang, Shanshan Zhu, Guang Yang and the lunch time crew, whose company greatly helped me keep my sanity (arguable) throughout the years.  To my friends not in science, Chito Porkchop Rivera, Thu Hyunh, Carrie Au, Jason Ong, Terry Kam, Leo Cheng, Andrew Kwan, Orivia Cheng, Special Agent Kevin Leong, Hing Sham, Kevin Yuan, Sunzanne Tao, Jennsuwong, Dave’s Pop Culture, the BC Special Olympics, PBC, ULTIMATE BEAST INFECTION, the Vancouver Dodgeball league, and The Derek Zoolander Center For Children Who Can’t Read Good and Wanna Learn To Do Other Stuff Good Too Dodgeball Team, thank you for keeping me based in reality knowing that there is a world outside the confines of the lab.  Thank you Taco Louis for providing me sustenance throughout the years.  Last but most importantly, to my mom, dad and brother, thank you for the continued love, patience and support, despite the fact that I haven’t had a real job over the last 30 years.  I can never thank you all enough.  1 1. Introduction  1.1. Critical Periods and the Visual Cortex  Understanding the processes of neural circuit formation in the developing brain is one of the most challenging tasks in science today.  The formation of neural circuitry is not exclusively defined by genetic makeup but also includes a dynamic component undergoing changes induced by the environment [1, 2].  Early in development there are periods when the number of synapses is much greater than what will finally be maintained in the mature brain [1, 2].  Therefore, neural circuitry development involves at least two important steps (i) the formation of synaptic connections between neurons and (ii) the selection and removal of wanted and unwanted connections.  Both of these processes are highly dependent on the environmental cues the brain receives while undergoing development [1, 2].  Critical periods play an essential role in helping us understand brain development and neural circuitry establishment.  Critical periods of the nervous system are defined as times during development when the brain’s sensitivity to external stimuli is at its peak, thereby making its neuronal properties more susceptible to modification by experience [3].  The time windows during which these critical periods occur during development vary depending on the region of the brain.  Early in development, the existence of critical periods has been shown to exist across species (though our focus will be on mammals) and has been demonstrated in many systems and functions including vision, audition, somatosensensory and language [4].     2 Takeo Hensch’s review identified several distinguishable aspects that make up a critical period. Some of them included:  a specific role for inhibition, a role for electrical activity, unique molecular mechanisms across systems and possible stages of the same pathway, regulation of onset and duration by age and experience, unique timing and duration of critical periods across systems, and the potential for reactivation in adulthood.  Our studies will focus on several of these listed aspects.  Responding to changes in activity levels among neuronal inputs is a fundamental component of critical periods.  Because an essential role of the critical period is to shape the brain according to experience, changes in activity can affect the strength and connectivity of neuronal inputs. Therefore, the construction of neural circuits is a process that can involve both formation and elimination of synapses [5, 6].  How the balance of addition and removal of synapses is regulated in the nervous system still remains a key question in neural development.  A role for electrical activity is also essential for critical periods since electrical activity allows a neuron to transmit information to other neurons and locations.  Neuron communication is performed through the propagation of electrical action potentials leading to the release of neurotransmitter.  Previous studies have shown that synapses can still form in the absence of activity [7-9].  However, activity is known to play an important role in whether synapses are maintained and/or become functional.  Some examples of this phenomenon can be observed in the cerebellum and neuromuscular junction.  Early in development Purkinje cells and muscle fibres are multi-innervated by climbing fibres or motor neurons respectively [10, 11].  As development continues, leading eventually to single innervation of the target cells, the EPSPs of the selected connections continue to strengthen while the others do not.  Furthermore, it has been recently shown that neurons that are active have the ability to inhibit other neighboring neuron  3 domains but can facilitate neurons in the same domain (such as orientation domains in the visual cortex) [12].  Classical studies where strabismus, a condition where the eyes do not line up in the same direction,  is artificially induced in kittens have shown that neuronal interactions will become OD column specific [13].  From this, the authors concluded that circuit selection also required correlated activity [13].  Another essential property of critical periods is that they are finite.  As systems mature, developmental cues and experience will begin to reduce the amount of plasticity that is possible. One possible reason for this is that the high plasticity early in life allows for adaptability to variable conditions, while stability later in development and life allow for efficient operation in adulthood [14].  This can occur through structural changes whereby structures that are formed in development reduce the potential for further change in the adult brain or via functional changes between the balance of excitation and inhibition [14].  Understanding the process that restrict critical periods to a specific time window will not only aid in our understanding of neural development but may also allow us to delay, prolong or reintroduce these periods of enhanced sensitivity to the environment.  The importance of understanding critical periods does not apply simply to development.  The events and circuitry that are established during this time will remain relevant throughout the lifespan of the organism.  Studies of children with vision deficits in one eye demonstrated that early input was necessary even for functions that would appear later in development and that early deficits could have long lasting effects even into adulthood [15].  In more general terms, learning tasks as a child is often very different than as an adult; the earlier the exposure the more beneficial it appears to be later in life [16].  Therefore, understanding the mechanisms behind  4 critical periods will prove beneficial for increasing the effectiveness of rehabilitation and adult learning. 1.1.1. The Visual Cortex  The visual cortex is one of the most extensively studied models of activity dependent synaptic plasticity and critical periods.  One of its biggest advantages is having two distinct inputs providing information into the system: namely, inputs representing the two eyes.  These inputs have the added advantage of simple manipulations to differentially reduce the activity between them.  When one eyelid is closed, during development, the open eye gains greater visual acuity. Looking further into the brain, one can see a predominance of connections belonging to the lateral geniculate inputs of the open eye formed in the visual cortex [3, 17].  The convergence of signals from both eyes (from retinal ganglion cells) begins in the lateral geniculate nucleus (LGN) of the thalamus.  While inputs from the eyes are directed contralaterally through the optic chiasm, some retinal ganglion axons are directed ipsilaterally to the thalamus.  Neurons from the LGN then send their axons directly to the primary visual cortex. Regions of the visual cortex in which information is received from both eyes are known as the binocular zones. [18].         Figure 2003 [ eyes. R Copyr 1.1.2  One o [19]. for an termi patter Wies when three predo show the op   1: Illustration 18].   Sections eprinted from ight (2003), w   .Ocular D f the basic w  A cortical c y given rec nals that are n at the visu el were the f  one eye wa months, Hu minant shif ed that occlu en eye [23]  of the visual  labeled ‘3’ ar  Current Opin ith permission  ominanc ays the ma olumn refer eptive field  associated w al cortex to irst to show s deprived o bel and Wie t of cortical ding one ey . zones present e illustrative o ion in Neurobi from Elsevier e Column mmalian ne s to a group attribute [20 ith either t  form altern  that respon f visual stim sel recorded responses to e during de in the mouse m f the binocular ology, Aug;13 . s and Plas ocortex orga  of cells in a ].  In the vis he right or le ating ocular siveness of t uli from bir  the cortical ward the op velopment l odel and the i  zone in the m (4):413-20, M ticity nizes itself  vertical clu ual system, ft eye are a dominance he or cortic th [22].  Af  response to en eye.  Ad ead to expan nputs that they ouse as they re ark Hübener,  is by the us ster that sha clusters of t rranged in a columns [21 al cells in a ter closing o  visual stim ditional stain sion of the  receive from ceive inputs fr Mouse visual c e of cortical re the same halamocorti  tightly orga ].  Hubel an column chan ne eye at bi uli.  They no ing of the c columns for 5 Hubener om both ortex, columns  tuning cal axon nized d ged rth for ticed a ortex   6 However, when the sensory deprivation on one eye was performed one to two months later the shift in response was lessened.  No signs of an ocular shift were observed when the deprivation was performed three months later.  This landmark series of experiments gave us fundamental insights into the mechanisms of activity dependent changes in the brain and critical periods.  1.1.3. Visual Cortex and NMDA Receptor Activity  NMDA receptors have been identified as major players in mediating developmental plasticity and critical periods.  NMDARs are glutamate gated (requiring a coagonist to occupy the glycine binding site) ion channels that are widely expressed in the nervous system and have been shown to play an essential role in excitatory neurotransmission [24-26].  Previous studies by other labs have shown that when the NMDARs are blocked, ocular dominance shifts are reduced significantly [27].  By infusing the selective NMDAR antagonist APV into the cortex while monocular deprivation occurred, electrical recordings showed that cells in the binocular zone of the visual cortex were still responsive to the deprived eye [27].  However, these early experiments were unable to distinguish whether the reduction of ocular dominance shifts were due to NMDAR function or the loss of sensory response and orientation that also resulted from the NMDAR blockade [27].  Follow-up studies looked to further resolve this issue by using different NMDA antagonists or reducing the expression levels of NMDAR genetically [28]. Using the NMDAR open channel blocker MK801, observations further supported the idea that NMDAR function was key in the mediation of ocular dominance shifts [29].  Furthermore, this series of experiments observed MK801 treatment did not change in visual activity or AMPAR responses.  In addition, experiments that blocked the expression of NR1 (an essential subunit of NMDARs) showed a strong correlation between decreases in NMDAR mediated response and decreases in ocular dominance plasticity [28].  Furthermore, the authors noted that the reduction  7 of NR1 did not affect visual responsiveness or stimulus selectivity at the visual cortex.  Together these initial experiments demonstrated the requirement of NMDAR activity in visual cortex plasticity.  An important feature of NMDARs rests in the finding that its subunit composition changes as development continues.  The majority of NMDARs exist in heterotetrameric complexes consisting of two NR1 (glycine binding) and two NR2 (glutamate binding) subunits; currently, four NR2 subunits have been identified [25].  In particular, much attention has been given to the relationship between NR2A and NR2B subunits because of the differences between the timing of their appearance during development [30, 31], electrical properties [32, 33], calcium influx properties [33, 34], and intracellular binding partners [35].  Early investigation into the electrical properties of NMDARs during development discovered that as development progressed, NMDA mediated EPSCs in the visual cortex became shorter; this was in parallel to reductions in synaptic plasticity [36].  Furthermore, methods to prolong visual cortex plasticity such as dark rearing would also delay the shortening of NMDA mediated EPSCs [36].  In addition, investigations looking into the properties of NMDARs found that NR2B containing NMDARs had longer current durations than their NR2A counterparts [32, 37].  From these experiments, it has been established that NR2B is initially the predominant subunit in the NMDAR complex but, as development continues, a switch to an increased level of NR2A subunit occurs [38]; this switch parallels the end of the critical period [4].  However, currently there is no evidence that the subunit switch is responsible for the end of the critical period.  Interestingly, other investigations have observed conflicting results.  Investigations stemming from western blot analyses observed changes in NR2A levels ranging from a drastic increase [30] to a gradual increase over time during development [39].  Moreover, NR2B levels have been  8 observed to remain constant and at relatively high levels [30], have a slower gradual increase over time [39] or to decrease significantly by the time the brain reaches adulthood [40].  These discrepancies may partly be attributed to the species used for observation.  While many investigations used rodents, some investigators opted for an animal with greater binocular vision such as ferrets [30].  Another factor contributing to the different observations is sample preparation.  Since many of these initial observations were obtained through western blots or via immunohistochemistry, the inclusion of nonsynaptic receptor proteins and multiple cortical layers may have affected experimental quantifications [41].  In addition, the times at which critical periods occur vary between the cortical layers which can create further ambiguity in the results [42].  Later investigations used electron microscopy (EM) to look at specific layers of the cortex and locate proteins near the postsynaptic membrane [41].  With this methodology, investigators observed an increase of localization of NR2A subunits at synaptic sites and a concomitant decrease in NR2B levels as plasticity declines; however, this observation only occurred in layer 4 [41] which may explain the divergent results obtained from previous studies. Despite the varying results, changes in NMDA composition are clearly substantial throughout development.  As a further extension to the observed changes of NMDA subunit composition during development, the ratio between the 2A/2B subunits has been shown to affect OD shifts in the visual cortex [43].  Investigations using NR2A KO mice showed mice lacking the NR2A subunit had similar NMDAR mediated EPSC as dark reared animals [43].  Furthermore, ocular dominance shifts in NR2A KOs showed greater potentiation in the non-deprived eye [44].  These KOs studies showed a greater shift towards plasticity at later ages favouring enhancement of potentiation [43, 44].  While roles of NMDAR subunits in synaptic plasticity are still highly  9 debated [43, 45, 46], these experiments show the importance of NMDARs for developmental critical periods.  Furthermore, the importance of NMDARs has been emphasized because of its properties to act as a coincidence reporter for situations in which pre- and postsynaptic cells are excited.  For activation to occur, NMDARs require both glutamate binding and membrane depolarization making them ideal candidates for involvement in Hebbian synaptic plasticity [47, 48]. 1.1.4. The Visual Cortex and Neurotrophins  Neurotrophin production is rapidly enhanced by neuronal activity and many studies now show it is an important component of activity dependent synaptic plasticity [49].  Neurotrophic factors, such as BDNF, have been demonstrated to play important roles in LTP consolidation [50, 51]. The importance of neurotrophins is no exception in the visual cortex.  Application of the neurotrophic factor NGF has been shown to prevent OD changes caused by MD with the rationale being that the excess neuroptrophin was enough to compensate for the lack of visual stimulization to the closed eye [52].  Furthermore, the absence of any light (known as dark rearing) can delay the visual cortex development.  However, application of exogenous NGF also acted as a proxy for the lack of visual stimulation for dark reared animals [53] preventing the effects of dark rearing.  Another neurotrophin, BDNF, was demonstrated to have similar effects as NGF when exogenously applied [54].  In addition to preventing MD and dark rearing effects, BDNF also had effects on the inhibitory circuitry.  Evidence of this was first observed when investigators noticed application of BDNF could alter the OD shift in favour of the deprived in some areas of the visual cortex during MD [55].  A recent study also showed, that in BDNF heterozygous  10 knock out mice, the inhibitory response amplitude, frequency, presynaptic GABA release and overall strength of inhibition was reduced in visual cortex [56].  Taken together, these observations showed the importance of the involvement of neurotrophins in neural circuit development.  1.1.5.The Visual Cortex and Synaptic Plasticity  The idea of Hebbian Plasticity, simplified to the slogan that “neurons that fire together wire together,” has long been postulated to be one of the key mechanisms involved learning and memory [57-59].  As previously mentioned, inputs from the eye deprived during monocular deprivation (MD) will come to occupy less territory in the visual cortex.  Moreover, earlier studies have shown that when the cortical neurons were prevented from responding during MD by inhibiting them using muscimol, the deprived eye’s connections paradoxically became more dominant [60].  Therefore, much effort has been given to examining mechanisms in which synchronous pre- and postsynaptic activity correlates with increased synaptic strength and asynchronous activity leads to a decrease.  The two most highly studied mechanisms reflecting this phenomenon are long term potentiation (LTP) and long term depression (LTD).  LTP is defined as a persistent increase in synaptic strength [61], while LTD is defined as a persistent decrease in synaptic strength [62].  These mechanisms are thought to elicit change in very short time scales having more local effects with a positive feedback component (50).  Furthermore, the ability to induce these synaptic changes experimentally decreases as the brain develops, however, they do not always exactly match the time course of the critical period [42].  LTP is a process involving several phases with distinct molecular characteristics.  They can be divided into a fast acting early phase involving only the insertion or modification of AMPA  11 receptors (E-LTP) and a late phase requiring the synthesis of new proteins (L-LTP) [63].  L-LTP is thought to produce much longer lasting effects in synaptic strength [63] and has been further divided into 2 categories.  LTP3 is defined as L-LTP requiring the synthesis of new proteins from new mRNA.  A translation dependent form of LTP has also been defined; one that requires protein synthesis but not mRNA synthesis, LTP2 [63].  Because of its fast acting properties and the ability to result in long lasting changes via protein synthesis, LTP has long been thought to play a key role synapse selection.  LTP is considered a potential mechanism for the synaptic strengthening of the open eye in MD [64], however, investigations looking at the role of LTP with the open eye’s synaptic connections have been limited.  While there has been some indirect evidence for the role of LTP in MD such as the inhibition or removal of CamKII, an important molecule for LTP [65, 66], several genetic mutations shown to alter LTP have proven to be inconclusive [67].  LTD, much like LTP, can be broken down into different phases: a fast acting phase involving the internalization of AMPARs at the synapse and a late phase also requiring the synthesis of new proteins [68].  LTD and its resulting internalization of surface synaptic AMPA receptors have been postulated to be key mechanisms for MD [69, 70].  Previous experiments have shown that MD mimicked the effects of AMPA phosphorylation and internalization that are observed in experimentally induced LTD; MD also produced a saturating effect reducing the amplitude of induced LTD [69].  Therefore, LTD has been often postulated as a mechanism for loss of input from the deprived eye in MD.  Interestingly, it has been observed that the mechanisms of LTD can differ among layers of the visual cortex [71].  However, regardless of layer location, MD would occlude/reduce the effects of experimentally induced LTD [71, 72].  These observations indicated that activity dependent synaptic plasticity (OD in the case of the visual cortex) probably exists in the form of several distinct mechanisms [64].  12  To further complicate matters, Hebbian plasticity appears unlikely to be solely responsible for OD shifts because deprivation does not lead to a total loss of connections [73].  Homeostatic mechanisms are also thought to play a role in OD shifts and in other forms of synaptic plasticity. While LTP and LTD are known to rapidly alter a specific group of synapses, homeostatic mechanisms adjust synaptic strength slowly and globally [74].  The two potential homeostatic mechanisms thought to be involved are those postulated by Bienenstock, Cooper, and Munro (BCM theory) and synaptic scaling.  The BCM theory proposes that when activity is reduced the threshold to produce LTP decreases and LTD increases and when activity is higher the opposite occurs.  In the case of ocular dominance shifts in the visual cortex, when one eye is deprived the reduced activity in the area can lead to increased facilitation of LTP for the open eye inputs [64]; this increased LTP can further stabilize synapses with the open eye inputs and increase its territorial advantage.  Synaptic scaling is the other homeostatic mechanism proposed to play a role in OD shifts and MD.  Synaptic scaling is a form of synaptic plasticity that that increases or decreases the synaptic strength of all inputs for a neuron in relation to its total activity [75].  This was first originally observed in neuronal cultures.   When activity was blocked with TTX, the postsynaptic response (mEPSP) was observed to be significantly higher when compared to neurons with normal activity levels [75, 76].  Therefore, synaptic scaling as a mechanism of synaptic selection can be explained as followed: for overall strength to remain the same, the efficacy of weaker/less potentiated synapses may be reduced when other synapses at different neuronal regions receive more attention/potentiating signals [77].  Traditionally Hebbian plasticity was thought to be the major mechanism behind OD shifts in the visual cortex.  However, it has been shown visually that deprived animals had increased mEPSCs in layer 2/3 pyramidal neurons [78] suggesting  13 possible additional mechanisms such as synaptic scaling.  Furthermore, a recent study looking at Tumor necrosis factor-alpha (TNFalpha) showed that  mice lacking the gene showed a decreased loss in closed eye responses but the open eye response remained unaffected [74].  In addition, the investigators noted that TNFalpha knock out phenotype also prevented synaptic scaling in vitro. Therefore they offered the explanation that OD shifts were actually a two-step process separating gain and loss of response which involved a homeostatic mechanism [74].  Models such as OD shifts in the visual cortex allow us to observe in a macro scale the plastic developmental changes that occur in specific regions of the brain.  As previously mentioned, these changes are thought to involve mechanisms that can alter the synapse.  LTP, LTD and homeostatic mechanisms all act in ways that can strengthen or weaken existing synapses.  As we observe the territorial increases of the open eye in OD shifts, synaptogenesis, maturation, stabilization and elimination will all likely play important roles in these highly sensitive activity dependent environments. 1.2. The Synapse  During the critical period and throughout development, what determines the success of an afferent is connectivity; for neurons, connectivity comes from the synapse.  It is through the synapse that neurons relay information to each other from the inputs they receive from the environment (such as the eyes for the visual cortex).  In vertebrates, most neurons communicate via chemical synapses [79].  A chemical synapse is formed by the interaction of axon (presynaptic) and dendrite (postsynaptic) where electrical signals that propagate down an axon are converted into chemical signals at the dendrite and then back to an electrical signal once more [79].   14 For a synapse to form, contact is required between the axon and dendrite.  The mechanism by which contact occurs is still unclear.  Contact can be initiated from axon or dendritic growth cone filipodia [80-82], filipodia from axon or dendrites can form en passant synapses [83-85], or between axon and dendritic shafts [81, 83].  Regardless of how contacts occur, the contacts must be stabilized for further synapse development.  Signals postulated to stabilize contacts come from a variety of sources, including from cell adhesions molecules (CAMs).  Cell adhesion molecules hold a special interest in synaptogenesis because of their bidirectional ability to coordinate pre- and postsynaptic differentiation [86].  Scheiffele et al. first showed the ability of CAMs to induce presynaptic clustering.  By co-culturing neurons with non-neuronal cells transfected to express neuroligin-1 and -2, they observed the neuroligins’ to induce presynaptic cluster formation on contacting axons [87].  Further, they were able to block this effect with a soluble form of the neuroligin receptor, β-neurexin.  Graf et al. then observed that the reverse was also possible and using the same methodology showed that neurexins could induce postsynaptic cluster formation on contacting dendrites [88].  These series of experiments were not only the first to show that CAMs could induce pre- and postsynaptic formation but in conjunction with each other were some of the first experiments to show the bidirectional influence CAMs have on synaptogenesis [87, 88].  Because of their roles in both excitatory and inhibitory synapse formation [86, 88-90], influence on synapse numbers [89, 91, 92], and possible identifier of predetermined 'hotspots' where presynaptic machinery is recruited [83], presynaptic neurexins and postsynaptic neuroligins interactions may be the most studied of the CAMs.  However, the study of CAMs involvement in synaptogensis is not limited to neurexin/ neuroligin interactions, many of which have also been identified [79, 93, 94].   15 The timing of synaptogenesis is a highly debated topic.  Early experiments looking into the timing of synaptogenesis produced observations concluding that presynaptic assembly occurred prior to postsynaptic assembly [95, 96].  Freidman et al. utilized serial labeling of active presynaptic sites over a range of presynaptic ages and observed that younger active presynaptic sites were less likely to have an opposing postsynaptic component.  Okabe et al. observed labeled SYN and PSD and observed that PSD95 clusters would appear after SYN clusters in the same location further supporting the notion that the presynapse preceeded the postsynapse.  However, other observations appear to contradict these results.  Washbourne et al. observed that pre- and postsynaptic recruitment occurred almost simultaneously after axodendritic contact and that NMDAR clusters were present before presynaptic proteins [81].  Furthermore, Gerrow et al. demonstrated that preexisting postsynaptic protein complexes were not only recruited to form synapses, but that stable sites could also be locations at which presynaptic proteins would aggregate later [83].  Nevertheless, it is clear that synaptogenesis requires the clustering and assembly of proteins at two distinct and ultimately opposing locations. 1.2.1. Identifying a Synapse  Identifying what constitutes a synapse is not a trivial matter.  One of the most common ways to identify a synapse is through the use of fluorescent microscopy and immunohistochemical methods.  Numerous studies have shown that clusters of specific proteins preferentially localized to synapses are frequently at ideal locations where synapses will form [81-83, 94, 96-101]. However, only areas that have matching pre- and postsynaptic labeled clusters can be considered a synapse because pre- and postsynaptic clusters can be found without its appropriate partner [97,  16 102].  Even with matching pre- and postsynaptic markers, correctly labeled areas may not have synaptic function [101].  While immunocytochemistry is not always completely accurate, it still remains an effective method for identifying putative synapses.  Further confirmation of a synapse can be done through functional studies such as markers for presynaptic active zone recycling with FM dyes or electrophysiological techniques [101]. However, active zone labeling has also been shown to exist without postsynaptic partners [85] and electrophysiological techniques cannot identify synapses that exist in a silent state [101].  In addition, intracellular recording techniques have enough sensitivity to measure the function of a single synapse.  However, those techniques quickly affect cell viability making them inconvenient to use for long term observation of a synapse [101].  Therefore, functional studies will also require additional support.  One of the best ways to identify a synapse is through retrospective analysis such as electron microscopy (EM) [103].   EM allows the user to observe the complete structure of the synapse. Under EM (fig. 4), the pre- and postsynaptic active zones of a mature synapse separated by a uniform 20nm cleft can be observed. The presynaptic active zone is identified by a small number of spherical synaptic vesicles situated close to the plasma membrane and a larger vesicle pool nearby [101].  The postsynapse can be recognized by a thick specialized membrane region (the postsynaptic density) consisting of neurotransmitter receptors, ion channels, and associated proteins [101].  These characteristics are conserved from invertebrates to vertebrates, both in vivo and in vitro [101].  However, EM only gives a glimpse to the state of the synapse after death and how it developed to that point will always require further investigation.    Figure synapt (arrow E. Ahm from E      The comb 1.2.2  The f [93]. conta  2: Electron M ic vesicles and s). Scale bar: 2 ari and Steph lsevier. refore, to ac ination of m . Presyna ormation of  Presynaptic ined in one icroscopy ima  uniform syna 00 nm.  From en J. Smith, K hieve the go ethodologie ptic Assem  the presyna  proteins th of two vesic ge of a matur ptic cleft. Post  Ahmari 2002 nowing a Nasc al to unders s is required bly ptic bouton at are transp ular organel e synapse. Hip synaptic thick [101].  Reprin ent Synapse W tanding the . begins once orted along les, piccolo pocampal slice ening is eviden ted from Neur hen You See complete lif  contact is m the axon be transport ve  cultures show t, and electron on 2002 Apr 2  It, Copyright e cycle of a ade betwee gin to cluste sicles (PTV ing clusters o -dense materi 5;34(3):333-6 (2002), with p  synapse a n axon and r.  They are ) and/or syn 17 f round al is noted , Susanne ermission dendrite aptic  18 vesicle protein transport vesicles (STV) [85, 104].  Presynaptic proteins are recruited to form new synapses by accumulation at sites of axodendritic contact [85]; however, presynaptic active zone clusters without dendritic contact have been observed to exist in mature neurons and these clusters can also be recruited to form a synapse [105].  Whether or not axodendritic contact is required, clusters of presynaptic proteins are often observed to have a mobile existence [82, 106]. Furthermore, recent experiments show that locations on the axon at which STVs paused more frequently lead to a greater probability of axodendritic contact and stabilization [82].  As observed from studies of OD shifts in the visual cortex, afferents from the deprived eye layer of the LGN occupy/innervate less territory when compared to afferents representing the open eye [23].  Because MD changes the amount of cortical territory normally occupied by each eye’s thalamic afferents, how the presynapse (and axon) develops under different activity levels is an important area to be explored.  The importance of stable presynaptic boutons was demonstrated in axon branch stability identifying them as sites for branch extension [107].  Using the presynaptic marker synaptophysin (SYN), Ruthazer et al. observed that puncta with greater clustering for SYN were more stable and lead to more stable axon branches.  Furthermore, increased activity also stabilized puncta and the corresponding axonal branches. 1.2.3. Postsynaptic Assembly  The other side of the synapse contains a vast protein network of neurotransmitter receptors, scaffolding proteins, adhesion molecules, and signal transduction molecules collectively known as the postsynaptic density [108].  Recruitment of postsynaptic proteins into the synapse is thought to occur in a stepwise process and is fundamentally different than presynaptic assembly [109].  One of the earliest events of postsynaptic assembly is the recruitment of PSD-95 [79]. PSD-95 is one of the earliest proteins detected in the postsynapse and has been widely proposed  19 as a key player in synapse maturation [110, 111].  Initially, it was observed that the clustering of PSD-95 occurred by a gradual accumulation followed by recruitment of NMDARs and eventually AMPARs [95, 96, 112].  However, other investigations observed modular movement of PSD-95 and PSD-95 existed in a postsynaptic complex that when stationary could lead to presynaptic induction or recruitment to other locations of synaptogenesis [83, 113].  It has also been observed that NMDARs can be recruited to locations of synaptogenesis prior to or at the same time as PSD-95 accumulation [81, 114] which further complicates the exact timing and mechanisms of synaptogenesis.  Once the scaffolding proteins and NMDARs begin to cluster at sites at which synapses form, they are not functional until the presence of AMPAR appears.  NMDAR and AMPAR recruitment into synapses appear to be independently regulated [79].  Initially, nascent synapses will remain silent for a period of time [81, 83].  Silent synapses are formed synapses that contain postsynaptic NMDARs but still lack postsynaptic AMPARs [115-117].  Evidence that NMDARs and AMPARs have differing recruitment mechanisms comes from observations showing that NMDAR accumulation in synapses appears to be experience independent while AMPA accumulation is experience dependent [118].  Of course the postsynapse does not consist of just PSD-95, NMDARs and AMPARs.  There is evidence for over a thousand postsynaptic proteins, however, much of the early research involving the timing of postsynaptic assembly has been focused on these key elements [79, 93]. 1.2.4. Synapses and Activity  It has long been observed that neuronal activity is not required for synaptic connections to form [119].  However, activity is still required for synapse development and integration into the  20 neuronal circuitry.  At the excitatory synapse, NMDARs and AMPARs are the main ionotropic receptor channels that are responsible for activity in the neuron.  Because NMDARs require both electrical activity and neurotransmitter binding to be activated, they can act as coincidence reporters for the synapse/neuron [47, 48].  While some recent observations have indicated that synaptic connectivity mediated by NMDARs is experience independent [118], its activation has been shown to regulate synapse formation/numbers and the recruitment of AMPARs into synapses [120, 121].  Recent investigations have demonstrated that removal of NMDAR activity by genetic ablation in only a few neurons during synaptogensis dramatically increases the number of functional synapses compared to silent synapses [120].  However, earlier investigations showed that chronic blockade of NMDARs during neuronal development lead to increased NMDAR distribution but little change in AMPAR [7].  Furthermore, NMDAR antagonism was observed to prevent synapse unsilencing in 3 week old in vitro neuronal cultures [122].  These differences can be attributed to the method and timing of inhibition but they also exemplify the complexity and numerous functions that NMDARs regulate during synaptogenesis and maturation.  Another important and widely examined function of NMDARs is their role in LTP and LTD. LTP represents an enduring increase in synaptic strength and LTD is a decrease in response. One of the major ways that this is regulated is by increasing or decreasing the number of AMPARs at a given synapse.  Early observations indicated that LTP was a potential mechanism for synapse maturation as its activation could transform synapses from a silent to active state [115, 116] .  Furthermore, LTP has been shown to lead to enlarged dendritic spines [123, 124] and has recently been proposed as a likely mechanism for spine stabilization [125].  Recent investigations show that LTP induction can result in an increase in dendritic spine size within  21 5hrs and while transient, a strong correlation between the temporary spine size increase and prolonged synapse stability was recorded [125].  As previously mentioned, early studies into the functional aspects of LTP showed LTP could result in dramatic structural changes to the activated synapse with size being the most visible change [123].  Induction of LTP not only increased spine size but subsequently lead to the incorporation of GluR1 subunit containing AMPARs [124].  Furthermore, a recent study showed that incorporation of the GluR1 C-tail was sufficient to induce postsynaptic spine enlargement [126] indicating the role of GluR1 as an important molecule for increased synaptic response and synapse stability [126].  Conversely, LTD is associated with a decrease in synaptic response and a reduction in surface expressed AMPARs at the synapse.  LTD induction has also been observed to reduce dendritic spine size [127, 128].  In addition, synaptic removal of AMPARs has been demonstrated to be necessary and sufficient to result in loss of dendritic spines and synaptic NMDAR responses [129].  However, several studies have also indicated that spine size and AMPAR density may be regulated by independent pathways increasing the complexity of the situation [129].  Nevertheless, the removal and insertion of synaptic AMPARs have been shown to have a direct correlation with the stability of presynaptic inputs. This, provides a potential mechanism by which transsynaptic signaling of AMPARs may further stabilize the synapse [130]. 1.2.5. Synapse Elimination  Early in development there are periods during which the number of synapses is much greater than can be maintained in the mature brain [2].  Therefore, to form proper and correct connections in the neural circuitry, synapse elimination is an integral part of development.  This  22 process is thought to depend on neuronal activity [131].  This is observed in the visual cortex in which synapse elimination plays an important part of development [132] and global activity inhibition has been shown to prevent OD shifts caused by MD [133, 134].  Studies looking into this phenomenon have hypothesized a role for LTD because of its role during critical period plasticity and receptor internalization [70].  LTD induction by low frequency stimulation has been shown to induce synapse loss [131].   However, LTP may also play a role in elimination since induction of LTP has been observed to promote spine loss, specifically among unstimulated spines [125].  Recent advances in imaging techniques have shown that synapse formation and elimination are dynamic processes and occur concurrently [135].  In addition, elimination may not be an all or none process but one that is more of a dynamic balancing act because some inputs that have lost postsynaptic territory have been observed to regain it later on [136].  Many of the details of synapse elimination still remain vague because it is difficult to identify when a synapse is about to removed and until recently much of our knowledge about synaptogenesis has come from observations using fixed tissues at fixed time points [137]. Despite these difficulties, several investigations have provided us with some insight into the possible mechanisms of removing a synapse.  Much of our knowledge on the cellular mechanisms of synapse elimination and activity has come from NMJ and cerebellum [6, 57]. The NMJ begins with a single muscle fibre being innervated by multiple motorneurons but over time is reduced to one [57].  Elimination of the other motorneurons does not occur if activity is blocked [138, 139] and selectively reducing the activity of an axon has been shown to lead to its removal, further emphasizing the activity dependence of the connections [140].  Furthermore, the elimination process is observed to be dependent on postsynaptic activity as local inhibition at the muscle itself can lead to elimination [141].  Moreover, postsynaptic activity also appears to  23 mediate synapse loss in the cerebellum [6] as postsynaptic expression of mGluR1 was shown to be required for the development of proper connectivity in the cerebellum [142].  Recent studies have begun to recognize the role of immune system molecules in the process of synapse elimination.  The complement protein, C1q has been linked to synapse elimination as mice lacking that gene exhibit signs of synaptic elimination malfunction such as the lack of anatomical refinement of retinogeniculate connections and excess innervation of LGN neurons [132].  Furthermore, Major histocompatibility complex class I (MHCI) molecules have been shown to negatively regulate synapse density and surface MHCI levels themselves are regulated by activity [143].  However, whether MHCI and the complement system work in concert remains to be explored. 1.3. Existing In Vitro Models of Activity Dependent Synaptic Plasticity  It is difficult to do detailed analyses in the central nervous system (CNS) because of the heterogeneity of inputs to individual neurons and the complexity of synaptic organization.  As previously mentioned, even among layers of the same region of the brain different mechanisms can exist for the same physiological processes.  Compounded by the fact that numerous molecular mechanisms are proposed to exist for individual neuronal plasticity models, there is a definite need to simplify the approach to discovery so as to not overwhelm with the abundance of existing potential mechanisms for development of the synapse and neural circuitry.  Fortunately, several simplified models have been developed.  These reduced models, will have differences but they have greatly contributed to our understanding of how connections between neurons (and their appropriate targets) are formed and changed with activity.  24 1.3.1. The Neuromuscular Junction (NMJ)  The neuromuscular junction provides greater simplicity and accessibility to synapses when compared to the connections contained in the brain [57].  The NMJ consists of parts of three cells: motor neuron, muscle fibre and Schwann cell [144].  Early in development, vertebrate muscle fibres are innervated by several motor axons.  However, as development continues innervation of a single muscle fibre undergoes a transition from multiple innervations by several motor axons to innervation by a single motor axon [145, 146].  This reduction of axons appears to be an activity dependent effect because altering the activity of the connecting motor afferents can change how the muscle fibre is innervated [147].  Therefore, this mass establishment of connections followed by a refinement may be a suitable comparison to some events that occur in the central nervous system.  Many classical experiments using the NMJ as a model have helped the research field gain significant progress in understanding neuronal activity dependent plasticity.  Isolated muscles are used either in fixed preparations or in vivo and by using lipophilic dyes (fixed) or direct transfection (in vivo) the different innervating nerves can be individually labeled [148, 149].  By varying the developmental times  when the isolated muscles are fixed, researchers were able to observe a gradual loss of synapses until the axon with no synapses would withdraw from the muscle fibre [148].   Furthermore, they observed the segregation of axonal branches into non overlapping compartments leading to the eventual withdrawal of all but one axon [149].  In addition, Gan and Lichtman (1998) observed that the axons closest to the selected axon would withdraw first followed by the more distal ones indicating a systematic removal occurring [149]. Further studies have shown that the refinement of NMJ synapses is an activity dependent effect. Recent studies have shown that asynchronous firing of inputs could lead to elimination and that  impo preve chann  Figure the mo Spring synaps    Becau under betwe us to sing synchro ntion of pos el proteins  3:  A neurom tor axon and t er Science+Bu es, 32(5-8), 20 se of its red standing of en axons ca use addition nous firing tsynaptic ac could also p uscular synap he muscle fibe siness Media: 03, 883-903, uced compl synapse dev n alter neur al models to on additiona tion potenti revent axon se comprising r that is to be i  Journal of Ne Bruce L. Patto exity, the N elopment an al circuitry d  help under l inputs cou als by overe  selection fro of a Schwann nnervated.  Fr urocytology, B n, figure 1. MJ has been d more imp evelopment stand the sy ld block the xpression of m occurrin cell that caps t om Patton 200 asal lamina an  a valuable ortantly, ho .  However, napse.  The ir eliminatio  inward rect g [151]. he motor nerv 3 [150].  With d the organiza tool in cont w differenc  several key distinct diff n [147].    A ifying potas e, the nerve ter  kind permissi tion of neurom  ributing to o es in activity  differences erences of ta 25 lso, sium minal of on from uscular ur  levels  require rget and  26 effector cells (pre- and postsynaptic components), the type of neurotransmitter released (acetylcholine versus glutamate) and cell composition indicate that direct correlations to NMJ and CNS synapse development may not always be possible.  Therefore models using CNS neurons would help us further broaden our understanding of mechanisms underlying synaptic development and neuronal interaction with differing activity levels. 1.3.2. Cerebellar Purkinje Cells  In the olivo-cerebellar system of rodents, climbing fibres (CF) converge and form synapses with Purkinje cell (PC) targets.  Initially during development, PCs are innervated by multiple CFs [10] and as the development process continues the  connections of only one CF will be strengthened while the other inputs are eventually eliminated [152].  This results in the majority of PCs being innervated by a single CF [11].  This process of extensive synaptic reorganization occurs during the first 2-3 weeks of postnatal development.  This developmental process for CF – PC connections makes it a good model for the study of activity dependent changes in the CNS.  Using cerebellar slices, much information has been elucidated on how CF – PC connections are formed, selected and eventually eliminated.  With PCs under whole cell patch clamp, CF mediated EPSC can be recorded by stimulating the CF.  Gradually increasing stimulation strength allows one to approximate the number of CFs innervating a single PC by observing the number of EPSC steps that are evoked [153].  Single innervated PCs would show an all or none response while multi-innervated PCs would show discrete steps.  In addition, researchers have observed that during development, the selected fibers would elicit stronger EPSP responses compared to soon to be removed CFs [155].  Investigators also hypothesized that specific forms of LTD and LTP may be important in the weakening and strengthening of those signals respectively [152].  Figure cerebe by two cerebe smoot the dis 2009 [ depend Elsevi     4: Illustration llar cortex. Th  types of excit llar granule ce h, proximal pa tal part of the 154]. Reprinte ent plasticity er  of a correctly ey are innerva atory afferents lls, which form rt of the Purkin Purkinje cell d d from Neuros of developing  innervated Pu ted by two typ , the climbing  the ascendin je cell dendrit endritic tree. G cience, Sep 1; climbing fiber  rkinje cell. Pu es of inhibitory  fibers originat g branches and ic tree while th ranule cells a 162(3):612-23 –Purkinje cell rkinje cells ar  interneurons ing from the i  the parallel fi e parallel fibe re in turn activ , L.W.J. Bosm synapses, Cop e the sole outp , the stellate an nferior olive a bers. Climbing rs make synap ated by mossy an and A. Kon yright (2009), ut neurons of t d the basket c nd by the axon  fibers innerv ses with the sp  fibers. From B nerth, Activit  with permissi  27 he ells and s of the ate the ines on ossman y- on from  Figure 2009;1 explan plates   In ad exper Letel roles cereb of CF PCs o depen on th   5:  Schematic 06:14102-141 ts (red) attach or brainstems dition, Chéd imental man lier et al., us of CF inner ellum and a  – PC could f differing a dent on the e age of the  diagram on th 07 (Permissio ed to mature (A are made to ex otal et al. de ipulations b ed this in vi vation and s ttached brain  also occur ges.  These  maturation postsynaptic e process of h n granted by N  and B) or im press GFP to e veloped and y culturing tro system t ynapse elim stem [156] in vitro.  Fut  experiment of both pre a  target (PCs indbrain expla ational Acade mature (C and nable the visu  in vitro sys cerebellar e o create a m ination usin .  With this m hermore, th s lead them nd postsyna ). nt experiment my of Science  D) cerebellar alization of (re tem that all xplants on a odel to stud g mouse hin odel, they ey could cul to conclude ptic partner s by Letellier s Copyright © plates or brain )innervating f owed for ea  millicell m y the pre- an dbrain expla showed that ture the sys that CF mul s but elimin M et al. PNAS 2009)  .  Here  stems.  Cereb ibres [156]. sier access t embrane [15 d post- syna nts contain  synaptic se tem with CF ti-innervatio ation was d 28  we see ellar  o 7]. ptic ing the lection s and n was ependent  29 The CF – PC network is an excellent model that provides researchers with another way understand the underlying mechanisms of activity dependent plasticity in the CNS.  However, this model is still restricted to specific cell types and it remains to be seen whether other neuronal types in the brain operate in the same manner.  Therefore, it continues to be an important endeavor to have models with the flexibility to incorporate the different types of neurons that exist throughout the brain. 1.3.3. Genetic Modification of Individual Neurons  Until recently, one of the most common methods to examine synaptic changes in a population of neurons was activity manipulation of the entire population of neurons [7, 59, 75, 119].  However, these whole scale changes were unable to address issues about how neurons with different activity levels interacted with each other.  To resolve this issue, several labs have begun to examine the effects of differing activity levels by genetically modifying individual neurons in the population.  Burrone et al. (2002) utilized this strategy to examine the effects of neurons with suppressed excitability within a network of active neurons by overexpressing the inward rectifier potassium channel Kir2.1 [119].  The overexpression of Kir2.1 hyperpolarizes the neuron and decreases its resting membrane resistance making it more difficult to excite [119] compared to the other surrounding neurons in the network.  The authors observed that fewer synapses were formed with neurons that had reduced excitability.  Furthermore, by varying the times at which the neurons were transfected with Kir2.1, Burrone et al. were able to observe some manner of a critical period.  When the neurons were transfected later in development, the observed difference in synapse number between inhibited and wild type neurons were no longer apparent.  While Kir2.1 is a postsynaptic modification that prevents the transfected neuron from being excited, Harms et al (2005) took an alternate approach and genetically prevented a subset of  30 neurons from being able to excite the surrounding network [158].  The presynaptic vesicle release of neurotransmitters was prevented by expressing the light chain of tetanus toxin in select neurons.  Without the release of neurotransmitter, the transfected neurons were unable to excite neurons postsynaptically and their axons should have some similar characteristics to inactive neuronal axons.  Interestingly, when this model was used to observe the effects of differing activity levels, no differences in synapse number were observed when comparing axons with vesicle release prevented and wild type axons.  A possible explanation for the lack of change is that preventing all synaptic vesicle release also prevents spontaneous vesicle release from occurring.  Therefore, it is possible that some cross talk between axon and dendrite may still be required even at minimal amounts.  Genetic modification to reduce the activity of individual neurons can provide a model that is not specific to any one neuronal cell type or group.  This allows further investigation into activity dependent mechanisms that occur in neurons in other regions of the brain such as the cerebral cortex and the hippocampus.  However, the rate of neuronal transfection is often low and the behaviour of a neuron with reduced activity in a region in which the vast majority of cells are normal may not be indicative of the mechanisms that occur in a more even setting.  Furthermore, genetic modifications may also cause additional unknown potentially unwanted changes in the modified neuron.  1.3.4  The b basic comp the ne speci reduc Figure (Copy togeth the ba    Tarsa of neu . Micro-I iggest adva  molecular a lexity of the twork activ fic axons an ed culture m  6: Schematic right © 2002, p er and allowed lance between   and Goda ( rons consis sland Pair ntage of red nd cellular c  intact brain ity still exis d dendrite fr odels.  diagram of ne ermission gra  to form synap the 2 choices o 2002) create ting of a wi s uced models hanges that .  However, ts in the cult om a particu uronal micro i nted by The N ses with each f synapse form d a further s ld type and m  is the simp  a neuron un  even an in ure dish.  D lar set of ne sland pairs fro ational Acade other or thems ation. implified th utant phen licity they a dergoes wit vitro system etermining t urons can b m Tarsa L , G my of Science elves.  Experi e in vitro m otype [159] fford the res hout the add , the additio he specific e a very ard oda Y PNAS 2 s) [159].   Two ments can them odel by cult .  By isolatin earcher to e itional leve nal complex interactions uous task ev 002;99:1012-  neurons are i  be designed  uring 2 cell g pairs of n 31 xamine ls of ity of between en in 1016 solated to change islands eurons  32 on glial islands, the investigators were able to compare mutant and wild type synapse formation under symmetrical conditions.  Because the system consisted of only two neurons the number of autapses and heterosynapses each cell formed were used for comparison.  The authors were able to compare the number of synapses formed and with additional classifications of autapses and hetrosynapses.  The elegance of this model allowed the investigator to observe effects of specific phenotypes under equal conditions and distribution in vitro to a wild type neuron since only two options for synaptic partners are available.  While the micro island model provides an effective way to observe synaptic preference between two individual neurons, its advantage of simplicity also acts as its greatest limitation.  In developmental situations, such as visual cortex plasticity, the resulting phenomena are not exclusive to synapse number or two specific neurons.  Additionally, there is often expansive territorial takeover by the more active neuronal group which includes growth and synaptogenesis. 1.3.5. Microfluidic Devices  The first culture system to compartmentalize parts of a neuron to separate regions with fluid control was devised by Campenot [160].  The Campenot chamber consists of compartments divided by Teflon attached to a thin layer of silicon grease in which axons are able to grow from one side to the other.  To promote the axon growth through the grease, the use of Nerve Growth Factor (NGF) was required.  In addition, Campenot chambers were not well adapted for more advanced microscopic techniques [161].  Microfluidics is a field of study that uses devices with small channels to control and manipulate minute amounts of fluids.  Therefore, microfluidic devices have the ability not only to separate  neuro envir from did n allow Figure then th from m growth Creati     ns into disti onment.  Pa axon but als ot require th ing for easie  7: Diagram o e surface is sc igrating from  will be able t ve Commons A nct groups b rk et al. intro o maintain d e use of gro r use of adv f Campenot C ratched.  A Te  one compartm o separate neu ttribution Lic ut allows fo duced the u istinct chem wth factors anced micro hamber from F flon divider w ent to another rites when they ense. r each group se of microf ical enviro and are cons scopic tech iller et al., 20 ith grease at th .  Neurons can  enter the oth  to have its luidic devic nments [162 tructed with niques 10 [163].  A cu e bottom acts  then be place er compartmen  own distinc es to not on ].  Furtherm  poly(dimet lture dish is fi as a stopper fo d in any comp ts.  Permissio t chemical ly separate c ore, their de hylsiloxane lled with colla r migration ce artment where n granted by th 33 ell body vices ) gen and ll bodies  eventual e   Figure Here y neuron cross o microg projec constr channe on the the oth microf (2005)    1.4.  Using group target devic group  8: Schematic ou can see the  using this dev ver to the othe rooves 10um tions to enter t uction but with ls allows us to  wells (seen in er.  Reprinted luidic culture .  Objective  a multi-com s of neuron ).  This was e.  The neur s; however,  of two compa  top and side v ice.  Neurons r side of the d wide and 3um he other side.  3 compartme  further limit  the second im  by permission platform for C s partment m s (the affere  made possi onal cell bo  the interact rtment microf iew of the dev are placed on evice.  Howev  high.  This ch The device we nts and microg entry into the o age) one can c  from Macmil NS axonal inju icrofluidics nts) formed ble by utiliz dies and den ion between luidic device il ice.  The illust one side of the er, the device annel prevents  use (also dev roove channel ther compartm reate hydrosta lan Publishers ry, regenerati  chamber, w synaptic con ing the prop drites were  the three gr lustrated and d ration shows h  device and al compartments  the soma from eloped by the s of 500um in ents to axons tic forces that Ltd: Nature M on and transpo e created a nections wi erties of a th completely oups is mai eveloped by T ow one can se lowed to grow  are separated  crossing, onl authors of this  length.  The 5 . Furthermore, limit chemical ethods 2005 A rt,  Anne M T model in wh th a third gr ree compar isolated from ntained thro aylor et al., 2 parate soma a .  As axons gr by a channel o y allowing neu  figure) is simi 00um long, 10  by adjusting t  flow from one ug;2(8):599-6 aylor et al., co  ich two sep oup of neur tment micro  the other n ugh their ax 34 005 [164]. nd axon a ow, they f ronal lar in um wide he volume  side to 05, A pyright arate ons (the fluidic euronal ons.  35 Moreover, the microfluidic properties of the device created distinct chemical environments for each separate neuronal group.  We chemically altered the activity of one neuronal afferent group and determined the synaptic changes that occurred compared to the other unaltered afferent group.  This enabled us to develop a model of cortical neurons with similar characteristics to the visual cortex, specifically dual afferent inputs, thereby creating a dual input activity dependent in vitro model.  With this model, we observed the influences opposing sets of afferents have with each another when sharing a common target in vitro.  Furthermore, by varying the time points at which activity is manipulated, we determined whether a critical period exists in our new model.  Upon establishing this model, we elucidated the molecular mechanisms involved.  The device properties provided an environment to block several postsynaptic effects of the target neurons with limited affect to the incoming afferents.  We determined the postsynaptic molecular mechanisms in our model by chemically inhibiting postsynaptic signaling pathways such as NMDARs, CamKII and Glur2 trafficking.  In summary, we created a novel activity dependent synaptic plasticity model with dual inputs that we can selectively control using microfluidics and in vitro culture techniques.   Using this model, we also investigated some underlying mechanisms of neuronal synaptic plasticity.   36 2.  An In Vitro Dual Input Activity Dependent Synaptic Plasticity Model Using Microfluidics 2.1. Introduction  In the developing brain, neurons undergo a substantial number of morphological and synaptic changes [1, 2].  A constant balance of synapse formation, stabilization and elimination is required for the proper development of neural circuit formation [2].  A major contribution to these mechanisms is driven by the experience/activity input that the brain receives during development [19].  The most classic and extensively studied example of this phenomenon occurs in the development of the visual cortex.  When one eye is closed during development, electrophysiological observations show increases in activity and territory in favour of the open eye in the visual cortex [3, 17].  This is also clearly observable structurally with anatomical techniques in the formation and maintenance of ocular dominance columns [22, 165].  The mechanism by which neural activity shapes neural circuitry during development remains a key question to be resolved in neuroscience research.  While visual cortex plasticity is one of the more robust and well studied models of neural development, it is not without its own difficulties. Because of the complexity of the mammalian brain, it becomes technically difficult or expensive to observe changes on the cellular and molecular scale.  Furthermore, there is the additional difficulty of genetic manipulations either yielding extremely low gene expression efficiency or requiring the use of transgenic mice.  To work around these issues in this and many other systems, many labs have chosen to use in vitro models.  The use of primary neuronal cultures allows for easier observation of neurons and delivery of specific genes via transfection [96, 166, 167].  These reduced models make it easier for the researcher to genetically manipulate and/or visualize individual neurons.   37 An important advantage of in vitro models is the ease of visualizing and genetically manipulating neurons.  However, changing the activity of neurons in culture is currently limited to non selective chemical treatment of the entire culture or genetic modification of a small population of neurons within the culture [119, 120, 158, 168-170].  The use of chemical treatments to simulate a situation that occurs between two eyes in development cannot be used because treatment would affect influences the entire culture.  Therefore, most models that look to change the activity of a subset of neurons in culture require the use of use genetic modifications.  This enables the user to examine the effects of manipulating a particular neuron’s activity relative to its unaltered surrounding environment.  The disadvantage is that it is difficult to specify a target group and control the proximity of altered and unaltered neurons.  We are now able to address these problems using microfluidic technology.  Microfluidics is a field of study that uses devices with small channels to control and manipulate minute amounts of fluids that enables the creation of chemically isolated environments [171].  Neurons can send long processes spanning several of these microenvironments while conventional in vitro culture methods cannot adequately create or control multiple environments effectively [119, 162].  By manipulating the volumes of the individual compartments, distinct chemical environments can be created due to the hydrostatic forces present between compartments.  Park et al, (2006), extensively detailed the fabrication and use of two compartment microfluidic devices.  Here, we used a three compartment microfluidic device to create a new model to study in vitro activity dependent synaptic plasticity (centre compartment) with dual inputs (lateral compartments). Using this three compartment device, we designated one of the compartments (centre) as target/postsynaptic neurons while the other two (lateral compartments) acted as presynaptic neurons.   Utilizing the microfluidic properties of our system, we selectively altered the activity of one or two of the neuronal compartments pharmacologically.   Furthermore, all postsynaptic  38 proteins viewed in each compartment originated from neurons grown in that region since only axons could traverse between compartments.  This added benefit prevented any confusion about which group of neurons the presynaptic axons formed synaptic connections with. Here we outline in detail the methodologies we used in developing our model using a three compartment microfluidics device. 2.2. Optimization of 3-Compartment Microfluidic Devices for Neuronal Cell Culture 2.2.1. Materials  We used a three compartment microfluidic device made from Sylgard 184 polymer similar in design to that described in Park et al, 2006. The only difference in the device we used was the addition of a third central chamber 500 µm in width.  The three channels of the microfluidic device were separated by 500 µm long 2.5 µm thick microgrooves.  All images reported here were taken from the central compartment, centered at the midline to eliminate any bias that might occur due to relative proximity of the field to one the lateral compartments.  Images were taken with a Zeiss Axiobserver microscope using numerical aperture 1.4 63x (immunostains) or a numerical aperture 0.95 40x (time lapse) objective.  The materials required for the assembly of the devices included treated No 1.5 (0.17mm) coverglass (Deckglaser), Poly-D-Lysine in borate buffer, rat primary cortical neurons, nucleofection reageants as detailed in Zeitelhofer et al., 2007 [166], DMEM with 10% FBS, Neurobasal (with B27 supplement, Glutamax and PenStrep), and YFP or CFP tagged synaptophysin plasmid constructs (pLL 3.7 backbone with a synapsin promoter).  Coverglass was incubated in a 50% HCl solution at room temrperature overnight followed by a distilled water rinse followed by sonication for 30min.  The glass was then rinsed with 100%  Ethan 75%  The P acid, All n regul Figure outer c target. marke croppe  ol followed ethanol unti oly-D-Lysi 1.9 g sodium euronal cultu ations specif  9:  3-compar ompartments    B  show the rs of the 2 inpu d image of co  by three sub l required fo ne (PDL) in  tetraborate re experim ied by relev tment microflu are designated microgrooves ts (synaptoph localized synap sequent wa r use.  borate buffe , 400 ml wa ents were pe ant authorit idic device.  A  as ‘presynapti that allow for ysin), the PSD tophysin and shes with 75 r consisted ter, pH 8.5) rformed in a ies. . Illustration c’ and the cen axon interactio -95 of the targ PSD-95 of a c % ethanol. of 0.1 M bo . ccordance w 3-compartmen tral compartm n between com et groups and ontrol experim  The glass w rate buffer s ith the gui t device dual i ent is designat partments.  C a merged imag ent. as then stor olution (1.2 delines and nput experime ed as the posts  shows the pre e of the 3.  D 39 ed in 4 g boric nt.  The 2 ynaptic’ synaptic is a  40 2.2.2. Procedure  2.2.2.1. Assembly of Microfluidics Chambers  Fabrication of the microfluidic devices has been detailed extensively in Park et al., 2006 [162]. Their protocol described the procedures required when the use of a plasma chamber was available.  Our experience showed that several additional preparation steps were required to provide consistent cell health and growth when a plasma chamber was not used.  Microfluidic devices were first sterilized with a quick wash in 70% ethanol.  They were then incubated in PBS (pH 7.4) overnight at 37 C and 5.5% CO2.  After incubation, devices were rinsed with distilled water and allowed to dry overnight at room temperature.  The cover glass was then dried thoroughly.  Finally, the devices and glass were UV sterilized for 30 minutes prior to assembly.  Once completely dry, chambers were pressed firmly onto treated No 1.5 cover glass.  After attachment of the device to the coverglass, we ensured no air bubbles were formed in between the glass and chamber.  Next, poly-D-lysine was put into half of the wells allowing the coating solution to pass through the channel into the other connected well.  The assembled devices were then incubated at 37 C for 2 days.  After incubation, PDL was removed from the wells without drying the channels.   The devices were washed with H2O by placing sterile ddH2O in the top half of the wells and allowing the water to flow through rinsing the inner part of the device.  This H2O wash step was repeated three times.  After washing, all wells were filled with sterile H2O and incubated overnight at 37C and 5.5% CO2.   Finally, H2O was removed from the device and replaced with Neurobasal (NB) with B27 supplement and glutamax.  The media was allowed to flow through as indicated previously.  Assembly was  41 completed upon replacement of H2O with NB.  Ideally, the devices were used the day of completion.  In our hands, devices were still usable for up to 3 days after complete assembly. 2.2.2.2.  Cell Culture, Nucleofection and Plating of Rat Primary Cortical Neurons  Rat cortical neurons were prepared from embryonic day (E) 18 rats and were dissociated into a cell suspension to be used for nucleofection [166].  Our nucleofection protocol required a cell suspension consisting of 4 million cells.  The cell suspension was aliquoted to contain 4 million cells and then spun down in a centrifuge at 600rpm for 30 seconds.  110 uL of electroporation buffer was mixed with 2ug of FP-tagged (CFP or YFP) synaptophysin plasmid and chilled for about 10min on ice.  The media of the pelleted cell suspension was removed and the cells were resuspended in the electroporation buffer.  Cells were then electroporated using setting O-003 on the Amaxa Nucleofector.  After electroporation, 1 mL of DMEM with FBS was added to the cells.  10 µL of the nucleofected cell suspension was placed on the top half of a channel and allowed to flow through.  Afterward, another 10 µL of the cell suspension was placed at the bottom of the channel.  Cells were allowed to settle for 10 min, before a quick wash of NB was applied to remove the DMEM/FBS media.  Finally, each well was filled 200 µL of NB (Invitrogen) with B27 supplement (Invitrogen) and glutamax (Invitrogen). 2.2.3.Troubleshooting  If the cell density inside the channel was too low, the flow of the cells through the channel was reexamined.  If the cell flow through was too slow the device was tilted to increase the flow force.  If more force was required, quick vacuum suction at one end of the channel was used. After a waiting period for cells to settle once again, the density in the channels was reexamined to ensure the desired density was obtained.  42 Initially, we qualitatively observed cell growth to be slower compared to cultures in classical culture systems.  We resolved this problem by ensuring all surfaces that neurons were exposed to were coated with poly-d-lysine.  Occasionally, axons did not stay within the micro-channels between the compartments.  We determined this occurred because both chamber and cover glass were not sufficiently dried during assembly.  Any moisture prior to assembly would affect the hydrophobic bond formed between the glass and sylgard.  If leaking of chemical into the other chambers was inappropirately high, we again determined this problem to be a result of improper assembly.  Our solutions included ensuring, prior to assembly, that all components were sufficiently dried and no air pockets were present between the sylgard and glass. 2.3. Procedure for In Vitro Dual Input Activity Dependent Synaptic Plasticity Model  2.3.1. Treatment of Neurons  At specified days, neurons were pharmacologically treated for 24 to 48 hours to manipulate their activity.  With the use of the multicompartment chambers, we were able to differentially alter the activity of each group of neurons separately.  Tests using fluorescent dyes (fluorescein or crystal violet) placed in the compartment wells showed that after 48hrs less than 1%  dye concentration is found in the other 2 compartments (with total volumes of 600µL each) if one outer compartments was filled with a total volume of 120 µL of dye.  When the middle compartment is filled with dye at a total volume of 250 µL, there is a 4.1 ± 2.2% concentration of dye found in  43 an adjacent well with that has volume of 120 µL after 48hrs.  If the adjacent well has a total volume of 600µL there is a less than 1% leak after 48hours. %concentration based on starting dye concentration of the filled well (100µM fluorescein or crystal violet).  Tests were performed by placing one of the aforementioned dyes in one channel while the other channels were filled with water.  After 48hrs, the water from the other two wells was collected and absorbance readings for the dyes were taken and compared to calibration curves (appendix A).  Experimental treatment was performed by removing the media from one compartment only.  The new media with the desired activity enhancer/inhibitor was added to one well only.  A wait time for about 10-30s was required to allow for the new media (with chemical) to flow through the channel, removing the excess untreated media from the channel.  (The other well was observed to fill as fluid flowed through the channel).  The media was removed again and then 50 µL of treated media was added to each well of the treated compartment.  A total volume difference between the treated and untreated channels was kept around 150 – 400ul depending on the treatment being performed.  Additionally, because the lateral compartments were labeled with different fluorescently tagged synaptophysin markers we alternated the treatment between markers to remove any potential bias caused by the differential labeling. 2.3.2. Analysis of Treatment Effects  After the desired treatment period ended, cells were fixed with 1% paraformaldehyde in PBS (Phosphate Buffered Saline pH 7.4, GibCo) for 4min followed by cold methanol (-20°C) fixation for 6min.  Cells were then washed 3 times with PBS.  Afterward, cells in the central compartment (target/postsynaptic cells) were immunostained for PSD-95 using PSD-95 antibody (monoclonal mouse, 1:1000 dilution from Affinity Bioreagents) in 3% BSA (bovine serum albumin, Sigma) in PBS overnight at 4°C.  Cells were again washed with PBS and then  44 incubated for 1hr at room temperature with Alexa 647 anti mouse secondary antibody (1:1000 dilution, Invitrogen).  The immunostaining for PSD-95 acted as a postsynaptic marker for all neurons in the compartment.  After fixation and immunostaining, neurons were viewed under a fluorescent microscope (Zeiss Axio Observer) at various points along the centre line of the middle compartment.  Comparisons were made to determine if the number of morphological (putative) synapses formed with the ‘postsynaptic’ group differed between the treated and untreated lateral compartments.  For the purposes of this thesis, a synapse was defined as a colocalization between a labeled synaptophysin puncta and a PSD-95 labeled postsynaptic puncta.  Our analysis was performed blinded while using the ImageJ colocalization plug-in available on the program’s website. Puncta colocalization was determined by thresholding puncta (blinded) to remove background immunofluorescence.  For all experiments, n equals the number of fields taken from at least 3 independent experiments (except Fig. 24 in which the number of independent experiments was 2). 2.4. Anticipated Results  A three compartment microfluidic device provides a way to create a situation in which a central group of neurons can act as a target/source for two other adjacent neuronal groups.  We are able to utilize the properties of the microfluidic device and inhibit the activity of a single group of neurons with chemical reagents such as the sodium channel blocker tetrodotoxin (TTX).  In addition to the chemical isolation of neuronal groups, a secondary advantage to this device is that all postsynaptic proteins/dendrites belong to the neurons seeded in the compartment.  This prevents any discrepancies regarding whether the postsynaptic targets belong to neurons other than the designated target cells.  45 This design allows groups of neurons with differing levels of activity levels to form connections with a common target group of cells.  This enables us to create an in vitro model with the similarity of two activity controllable inputs like the visual cortex in vivo.  Under normal conditions, axons from either presynaptic group would form similar numbers of synapses with the postsynaptic neurons.  However, when one of the presynaptic groups has its activity reduced by TTX, that group forms significantly fewer synapses when compared to the untreated presynaptic group.  Throughout this thesis, our assessments will use morphological synapses, i.e. puncta of synaptophsyin overlapping with puncta of PSD-95, which we will refer to as synapses for simplicity.  Figure a singl group   10:  Sample d e presynaptic (or blank for c ual input activ group (tetrodo ontrol).  Green * ity dependent toxin, 1µM) an  indicated the  synaptic plast d control (onl  untreated grou icity model re y volume diffe p. sults from a D rence). Blue i IV 12 – 14 tre ndicated the tr 46 atment of eated  47 3. A Critical Period Exists In Vitro in Our Model of Activity Dependent Synaptic Plasticity  Differences in activity levels between neurons can affect its affinity for target cells.  This phenomenon is classically seen in the visual cortex where.  During development, when one eye is closed, visual acuity favours the open eye.  Here, we constructed an in vitro model using microfluidics to simulate the basic dual input situation in the visual cortex.  Primary cortical neurons were separated into three chemically independent groups with one group designated as the common postsynaptic target (centre compartment) and the other two (lateral compartment) groups presynaptic afferents.  Using TTX or muscimol, we reduced the activity of a single ‘presynaptic’ group of neurons for 48h and compared its relative success of target innervation with the unaltered ‘presynaptic’ group by the number of synapses formed.   We showed that inhibition treatment of one group lead to a decrease or increase synapses being formed by the treated  or  untreated group respectively.   By varying the developmental ages in which we inhibited a particular group of neurons, we observed an in vitro “critical period” during which changes in synapse number were strongly dependent on input conditions.  In addition, we observed that differing treatments of neuronal activity inhibition exhibited different time windows of synaptic change.  Furthermore, as an additional step, we were able to integrate neurons of different ages into the system.  We observed in our model that the in vitro critical period was dependent specifically on the age of the post-synaptic targets when presynaptic cells were inhibited by GABA receptor activation. 3.1. Introduction  Elucidating the mechanisms underlying the formation of neural connections in the developing brain is one of the most challenging tasks in neuroscience.  The formation of neural circuitry is  48 not exclusively defined by genetic makeup but also includes a dynamic component regulated by the environment [1, 2].  Early in brain development, there are periods when environmental cues can greatly change the structure and function compared when it is mature [1, 2].  One way brain structure and function is determined is through establishment of  synaptic connections; activity levels could be one way of assessing where these connections should be made. Therefore, the construction of neural circuits is an activity dependent process that adapts to external signals from the environment [6, 57].  Where synapses form are often determined by the inputs that the brain receives during development.  The time window during which the brain is most susceptible to change from the influences it receives from its environment is often referred to as a ‘critical’ or ‘sensitive’ period.  Understanding the basis of how the brain develops during these periods is crucial as the effects of what happens early in development have a substantial impact in how adults behave and learn later in life [3]. 3.2. Results  In order to create our model, we used a three compartment microfluidic device similar in design to the two compartment design described in Park et al (Fig. 8, [162]).  This device allowed us to create three chemically isolated environments yet all three groups were still able to interact physiologically.  This enabled us to create an environment in which two ‘presynaptic’ neuronal groups could interact and form synapses with a common ‘postsynaptic’ group, therefore, allowing us to observe activity dependent synaptic changes between two sets of afferents in our in vitro system.  By utilizing the hydrostatic properties of the device and reducing the working volume of one of the ‘presynaptic’ groups, we created a chemical environment distinct from the other two neuron groups.  49 3.2.1. Nonspecific Silencing of Presynaptic Neuronal Activity  Tetrodotoxin (TTX) is a potent sodium (Na+) channel blocker capable of silencing neural activity [75, 172].  By binding to fast voltage gated sodium channels, TTX disables their function and in doing so prevents the propagation of action potentials.  This makes TTX a very potent nonselective (i.e. non-receptor based) inhibitor of neural activity.  To simulate the reduced activity levels of ocular occlusion in vivo, we maintained one of our ‘presynaptic’ groups under the conditions of 1µM of TTX in normal neural media for 48 hrs (where leak into the other compartments was determined to be less than 1%, Ch 2).  It takes approximately four days for axons from one compartment to enter the adjacent compartment [162].  Our preliminary observations showed that it took anywhere from 4-7 days for the axons from a ‘presynaptic’ compartment to enter the ‘postynaptic’ group compartment.  To determine the presence of an activity dependent synaptic changes between the sets of afferents occurred in our model, we treated one ‘presynaptic’neuronal group with TTX beginning at DIV 8 (12, 19 and 26 also) for two days.  To distinguish the ‘presynaptic’ afferents, each presynaptic group was transfected at DIV 0 with a different fluorescent protein tagged synaptophysin (FP-syn) construct.  Once treatment ended, neurons were immediately fixed and immunostained for postsynaptic protein, PSD-95.  Afterward, we examined the cultures to determine the number of synapses each presynaptic group formed with the postsynaptic target group in the centre compartment.  For our purposes, we defined a synapse as FP-syn being colocalized with with PSD-95 [85, 100, 173].  Counting the total number of labeled synapses in randomly selected fields of view, there was no significant difference between the treated (6.96 ±  50 1.64 synapses per field) and untreated (6.42 ± 1.19 synapses per field, p > 0.05) ‘presynaptic’ groups (Fig. 11).  It was possible that the previous time point was too early in the neuron’s development to show much of a difference.  Earlier investigations looking at synaptogenesis concluded that synapse maturation and formation, in cultured neurons, occurred approximately two weeks into development [122, 174].  Therefore, we repeated the experiments later in development beginning treatment at DIV 12.    Figure groups applic 24, un     11:  Compari  (TTX treatme ation while the treated = 6.42 0 2 4 6 8 10 treatm nu m be r of  s yn ap se s pe r fie ld son of the num nt, DIV 8 - 10  other group r ± 1.16, n = 24 DIV TT X ent of 'pre ber of synaps ).  From DIV emains at norm ; Mean ± SEM 8-10 un tre ate synaptic'  es between tw 8 – 10, one pre al levels.    t t  d ( TT X) groups o sets of affere synaptic grou est: p = 0.79; s nts from two d p has its activi ynapses/field: ifferent presy ty reduced by  TTX = 6.96 ± 51 naptic TTX  1.61, n =   Figure groups applic = 46, u     12:  Compari  (TTX treatme ation while the ntreated = 16 0 5 10 15 20 treatm nu m be r of  s yn ap se s pe r fie ld son of the num nt, DIV 12 –  other group r .78 ± 2.08, n = DIV 1 TT X ent of pre  ** ber of synaps 14).  From DIV emains at norm  46; Mean ± S 2 - 14 un tre ate synaptic g es between tw  12 – 14, one al levels.    t t EM d ( TT X) roups o sets of affere presynaptic gr est: p = 0.005; nts from two d oup has its act  synapses/field ifferent presy ivity reduced b : TTX = 9.80 52 naptic y TTX ± 1.26, n   53 After treatment and fixation, there was a significant difference in the number of labeled synapses between the treated and untreated presynaptic groups.  The TTX treated presynaptic group had an average of 9.80 ± 1.28 synapses per field while the untreated group had an average of 16.78 ± 2.11 synapses per field, p < 0.05 (Fig. 12).  Therefore, at this stage of in vitro development, differences in activity levels affected the balance of synapse number between the two input pathways.  To determine the existence and extent of a critical period in vitro, we performed the same experiment at two additional time points, DIV 19-21 (Fig. 13) and DIV 26-28 (Fig. 16).  Again, there was a significant difference in the number of synapses between TTX treated and untreated ‘presynaptic’ groups, 9.55 ± 1.26 and 16.28 ± 1.62 synapses per field respectively at DIV 19 – 21, p < 0.05.  However, at DIV 26 – 28, differences in activity between ‘presynaptic’ groups were no longer present using (47.10 ± 7.23 synapses per field for the TTX treated group and 39.80 ± 9.30 synapses per field for the untreated group, p > 0.05).  These results indicated that when the neural activity of a presynaptic group was reduced (by TTX) in comparison to another ‘presynaptic’ group, fewer synapses are formed by the group with reduced activity.  However, these differences only occurred between the second and third week of development in our model.  This indicated a critical period occurs around the second and third week of development in cultured cortical neurons in vitro in our model.   Figure groups applic = 33, u     13:  Compari  (TTX treatme ation while the ntreated = 16 0 5 10 15 20 treatm nu m be r of  s yn ap se s pe r fie ld son of the num nt, DIV 19 – 2  other group r .28 ± 1.62, n = DIV TT X ent of 'pre  ** ber of synaps 1).  From DIV emains at norm  33; Mean ± S 19-21 un tre ate synaptic' es between tw  19 – 21, one al levels.    t t EM d ( TT X) groups o sets of affere presynaptic gr est: p = 0.003;  nts from two d oup has its act  synapses/field ifferent presy ivity reduced b : TTX = 9.55 54 naptic y TTX ± 1.26, n   Figure groups applic = 20, u   14:  Compari  (TTX treatme ation while the ntreated = 39 0 20 40 60 treatm nu m be r of  s yn ap se s pe r fie ld son of the num nt, DIV 26-28  other group r .80 ± 9.30, n = DIV 2 TT X ent of 'pre ber of synaps ).  From DIV emains at norm  20; Mean ± S 6 - 28 un tre ate synaptic' es between tw 26 – 28, one p al levels.    t t EM d ( TT X) groups o sets of affere resynaptic gro est: p = 0.54; s nts from two d up has its activ ynapses/field: ifferent presy ity reduced by  TTX = 47.10 55 naptic  TTX ± 7.23, n  56 3.2.2.Inhibition of Presynaptic Neuronal Activity with GABA Agonist, Muscimol  GABA is the main inhibitory neurotransmitter in the adult brain.  GABA specific receptors on the synapse are open ligand gated chloride channels (GABAA receptors) or activate metabotropic receptors that cause the activation of inward rectifying potassium channels (GABAB receptors) [175].  This allows GABAR to regulate excitability by hyperpolarizing the membrane potential or through shunting of excitatory inputs [176].  Muscimol is a potent agonist of the GABAA receptor.  Our previous experiments used a method of inhibition (TTX treatment) that was nonspecific to pre- and postsynaptic neurotransmission.  In this section, we examined whether the effects of a more specific inhibition, GABAA receptor (GABAAR) inhibition, lead to a change in the synaptic balance between  ‘presynaptic’ groups with differential activity.  Once again, we began our experiment at DIV 8 and allowed the treatment to persist for 48hrs until DIV 10.  However, this time we used 10 µM muscimol to inhibit the activity of a ‘presynaptic’ neuronal group.  After treatment, fixation and immunostaining, there was a significant difference in the number of synapses between the muscimol treated and untreated ‘presynaptic’ groups (3.14 ± 0.71 and 6.60 ± 0.67 synpases per field respectively, p < 0.05, Fig 15).  This result indicated that GABA inhibition affected activity dependent synaptogenesis at an earlier time in development in vitro when compared to our TTX results.  We repeated the experiment at with muscimol treatment at DIV 12-14 (Fig. 16) and began to elucidate if an in vitro critical period also existed with GABA inhibition.  At this age range, there was still a significant difference in synapse number when comparing the muscimol treated and untreated ‘presynaptic’ groups (15.41 ± 2.40 and 28.69 ± 4.19 synapses per field respectively, p < 0.05).  Figure groups muscim = 3.14    15:  Compari  (muscimol tre ol application  ± 0.71, n = 35 0 2 4 6 8 treatm nu m be r of  s yn ap se s pe r fie ld son of the num atment, DIV 8  while the oth , untreated = 6 DIV mu sc im ol ent of 'pre *** ber of synaps  – 10).  From er group rema .60 ± 0.67, n = 8 - 10 un tre ate synaptic'  es between tw DIV 8 – 10, on ins at normal l  35; Mean ± S d ( mu s) groups o sets of affere e presynaptic evels.    t test: EM nts from two d  group has its p = 0.0007; sy ifferent presy activity reduce napses/field: m 57 naptic d by uscimol   Figure groups muscim 15.41     16:  Compari  (muscimol tre ol application ± 2.40, n = 32 0 10 20 30 40 treatm nu m be r of  s yn ap se s pe r fie ld son of the num atment, DIV  while the oth , untreated = 2 DIV 1 mu sc im ol ent of pre ** ber of synaps 12 – 14).  From er group rema 8.69 ± 4.19, n 2 - 14 un tre ate synaptic es between tw  DIV 12 – 14 ins at normal l = 32; Mean ± d ( mu s) groups o sets of affere , one presynap evels.    t test: SEM nts from two d tic group has i p = 0.008; syn ifferent presy ts activity redu apses/field: m 58 naptic ced by uscimol =   59 However, at DIV 19 - 21 (Fig 17), the differences causes by differences in neuronal activity were no longer present (22.84 ± 4.41 and 27.76 ± 4.01 synapses per field for the muscimol treated and untreated groups respectively, p > 0.05).  Furthermore, no synaptic changes occurred when a single ‘presynaptic’ group was reduced in activity from DIV 26-28 (Fig. 18).  From this set of experiments, we determined that neuronal activity inhibition by GABAAR could also affect the synaptic balance between two groups of ‘presynaptic’ neurons.  Interestingly, the critical period caused by this form of inhibition began and ended earlier when compared to our earlier observations using TTX.  With this, we conclude d that different types of inhibition can lead to distinct critical periods and potentially have different mechanisms involved.  Figure groups muscim 22.84    17:  Compari  (muscimol tre ol application ± 4.41, n = 19  0 10 20 30 40 treatm nu m be r of  s yn ap se s pe r fie ld son of the num atment, DIV  while the oth , untreated = 2 DIV 1 mu sc im ol ent of 'pre ber of synaps 19 – 21).  From er group rema 7.76 ± 4.01, n 9 - 21 un tre ate synaptic' es between tw  DIV 19 – 21 ins at normal l = 21; Mean ± d ( mu s) groups o sets of affere , one presynap evels.    t test: SEM nts from two d tic group has i p = 0.41; syna ifferent presy ts activity redu pses/field: mu 60 naptic ced by scimol =  Figure groups muscim 23.55     18:  Compari  (muscimol tre ol application ± 6.95, n = 11 0 10 20 30 40 treatm nu m be r of  s yn ap se s pe r fie ld son of the num atment, DIV 2  while the oth , untreated = 2 DIV 2 mu sc im ol ent of 'pre ber of synaps 6 – 28).  From er group rema 0.82 ± 4.00, n 6 - 28 un tre ate synaptic' es between tw  DIV 26 – 28 ins at normal l = 11; Mean ± d ( mu s) groups o sets of affere , one presynap evels.    t test: SEM nts from two d tic group has i p = 0.74; syna ifferent presy ts activity redu pses/field: mu 61 naptic ced by scimol =   62 3.2.3. The Physical Properties of the Microfluidic Device Do Not Affect the ‘Presynaptic’ Groups  In order to maintain the distinct microenvironments, the microfluidic devices have special physical properties that could potentially affect the neuronal groups.  To determine that the physical properties of these devices were not the cause of the effects, we repeated the same experiments but the ‘treated’ side only received the same mechanical treatments and pressure gradients.  The ‘presynaptic’ group treatment consisted of maintaining the reduced volume required for chemical isolation (without the activity inhibitors) and the wash steps involved to replicate the liquid flow through the channels when replacing the culture media.  Observation of the 4 experimental time points (Fig.  21 - 24) showed no difference between ‘treated’ and untreated presynaptic groups when comparing against the number of synapses in the field.  Therefore, we concluded that the effects we observed in the previous effects were a result of the chemical inhibition and not due to the properties of the microfluidic device.             Figure presyn n = 27 nu m be r of  s yn ap se s pe r fie ld  19: Compari aptic groups, D ; Mean ± SEM co n 0 5 10 15 son of the num IV 8 - 10.  t t  DIV 8 tro l ber of synapse est: p = 0.70; s  - 10 co ntr s between two ynapses/field: ol  sets of affere  control = 8.44  nts from two d  ± 1.31, n = 27 ifferent untrea , control = 9.4 Treatment Group: Blue  Untreated Group: Green 63 ted 4 ± 2.25,   Figure presyn 2.14, n  nu m be r of  s yn ap se s pe r fie ld  20:  Compari aptic groups, D  = 45; Mean ± co n 0 5 10 15 20 son of the num IV 12 - 14.  t  SEM DIV 12 tro l ber of synaps  test: p = 0.58;  - 14 co ntr es between tw  synapses/field ol o sets of affere : control = 13  nts from two d .51 ± 1.28, n = ifferent untre  45, control = Treatment Group: Blue  Untreated Group: Green 64 ated 14.89 ±   Figure presyn ± 2.90   nu m be r of  s yn ap se s pe r fie ld      21: Compari aptic groups, D , fields = 20; M co n 0 5 10 15 20 son of the num IV 19 - 21.  t ean ± SEM DIV 1 tro l ber of synapse  test: p = 0.39; 9 - 21 co ntr s between two  synapses/field ol  sets of affere : control = 12  nts from two d .05 ± 1.92, fiel T G B  U G G ifferent untrea ds = 20, contr reatment roup: lue ntreated roup: reen 65 ted ol = 15.05   Figure presyn 4.32, n nu m be r of  s yn ap se s pe r fie ld  22: Compari aptic groups, D  = 17; Mean ± co n 0 10 20 30 40 50 son of the num IV 26 - 28.  t  SEM DIV 2 tro l ber of synapse  test: p = 0.49; 6 - 28 co ntr s between two  synapses/field ol  sets of affere : control = 42  nts from two d .29 ± 5.26, n = ifferent untrea  17, control = Treatment Group: Blue  Untreated Group: Green 66 ted 37.53 ±   67 3.2.4.Cross Group Effects of Unilateral Inhibition  Figure 23: Percentage of total PSD-95 colocalized with labeled synaptophysin from the ‘presynaptic’ groups, DIV 8 - 10.  ANOVA: p = 0.0004, post hoc: Turkey test; control = 1.89 ± 0.26%, 54; TTX = 1.19 ± 0.24%, 24; untreated (TTX) = 1.16 ± 0.20%, 24; muscimol = 0.62 ± 0.12%, 34; untreated (mus) = 1.52 ± 0.14%, 34; Mean ± SEM, n (*** indicates significant difference compared to control)  DIV 8 - 10 co ntr ol TT X un tre ate d ( TT X) mu s un tre ate d ( mu s) 0 1 2 3 4  *** treatment of 'presynaptic' groups % PS D -9 5 w ith  la be le d sy na pt op hy si n  We then compared our treatment groups against our control experiments for the three time points in which induced synaptic differences were caused by at least one of the inhibition treatments. One technical aspect to using these particular microfluidic devices is that it is difficult to perfectly control the cell density in the area of observation.  Therefore to compensate for that, we normalized the number of synapses to the total number of PSD-95 puncta in the field of view. This gave us the percentage of PSD-95 used to form synapses with our labeled synaptophysin puncta.  Looking at the %PSD-95 at DIV 8-10 (Fig. 23), 0.62 ± 0.12% of the PSD-95 in the field of view colocalized with the labeled synaptophysin of the muscimol treated group.  This was a significant decrease compared to the 1.52 ± 0.14 % of PSD-95 associated with the untreated muscimol group (t test: p < 0.0001).  Furthermore, PSD95% for the muscimol treated afferents was significantly less compared to a control group in which both ‘presynaptic’ groups remained  68 at ‘normal’ neural activity level (1.89 ± 0.25%).  However, there was no significant difference between the control and untreated muscimol group.  From this, we conclude that at DIV 8 – 10, inhibition by GABAR primarily affects the ‘presynaptic’ group with reduced activity causing it to form fewer synapes. Figure 24:  Percentage of total PSD-95 colocalized with labeled synaptophysin from the ‘presynaptic’ groups, DIV 12 - 14.  ANOVA: p < 0.0001, post hoc = Turkey test; control = 2.25 ± 0.17%, 91; TTX = 1.66 ± 0.20%, 39; untreated (TTX) = 2.60 ± 0.31%, 39;  muscimol = 2.39 ± 0.34%, 32; untreated (mus) = 4.53 ± 0.61%, 32; Mean ± SEM, n   (*** indicates significant difference compared to control) DIV 12 - 14 co ntr ol TT X un tre ate d ( TT X) mu sc im ol un tre ate d ( mu s) 0 2 4 6 *** ANOVA p < 0.001 * treatment of presynaptic groups % ps d9 5 co lo ca liz ed  w ith  la be le d sy na pt op hy si n   At DIV 12-14 (Fig. 24), there was a large increase in the PSD-95 percentage associated with the synapses formed with the opposing ‘presynaptic’ group that remained untreated by muscimol. Moreover, there were no other significant differences among the other ‘presynaptic’ neuronal groups compared to the control (ANOVA: p < 0.0001).  However, comparisons made between the TTX treated and TTX untreated groups showed a significant difference (t test: p = 0.013) supporting our previous results.  At this time point, when a ‘presynaptic’ group was inhibited by  69 muscimol, the opposing untreated afferents formed more synapses than control.  This would signify that inhibition by GABAR activation on one ‘presynaptic’ group also exerted some influence to the opposing afferents to form additional synapses with a target group.  However, TTX treatments appear to result in a modest decrease in preference for the treated group and a modest increase to the untreated group, neither which is individually different from control. Interestingly, the muscimol treated afferents showed no reduction in the synapses compared to control.  We concluded that inhibition by muscimol induced an affect that resulted in an increase number of synapses for the opposite untreated ‘presynaptic’ neurons.  Therefore, we showed that that changing the activity of one ‘presynaptic) group could result in influences that affected the opposing untreated ‘presynaptic’ group in our model. Figure 25: Percentage of total PSD-95 colocalized with labeled synaptophysin from the ‘presynaptic’ groups, DIV 19 - 21.  ANOVA: p = 0.016, post hoc = Turkey test; control = 1.95 ± 0.26%, 40; TTX = 1.304 ± 0.23%, 32; untreated (TTX) = 2.10 ± 0.22%, 32;  muscimol = 1.92 ± 0.33%, 21; untreated (mus) = 2.74 ± 0.37%, 21; Mean ± SEM, n  DIV 19 - 21 co ntr ol TT X un tre ate d ( TT X) mu sc im ol un tre ate d ( mu s) 0 1 2 3 4 treatment of presynaptic groups % PS D -9 5 w ith  la be le d sy na pt op hy si n   *  70 At DIV 19-21 (Fig. 25), there were no significant differences when comparing the percentage of PSD-95 colocalized with labeled synaptophysin to the control group.  However, once again there was a significant difference between TTX treated and TTX untreated ‘presynaptic’ group (t test; p = 0.014). From these analyses, we concluded that in our dual input model, changes in activity from one set of afferents can influence changes to an opposing set of afferents. 3.2.5. Activity Inhibition Does Not Change Presynaptic or Synaptic Density on the Axon  Additionally, we looked at the synaptic and presynaptic puncta density along the axon for each presynaptic group for the three developmental time points where we observed changes in synaptic numbers (Fig. 26 - 28).  There was no significant difference in synaptic or presynaptic density along the axon between the groups at each time point.  This observation leads us to believe that the major changes associated with synapses number in our model were in fact growth related.  Figure ANOV counte untrea  A 0 0 1 1 sy na pt op hy si n/ 10  µ m  o f a xo n le ng th   26: Compari A: p = 0.29) o rparts shows n ted (mus) resp  co ntr ol t .0 .5 .0 .5 treatmen son of presyna f presynaptic o significant d ectively.  (Syn  DIV 8 - 10 tx un tre ate d ( ttx ) mu s t of 'presyna ptic puncta den afferent sets un ifferences.  n aptophysin = g  cim ol un tre ate d ( mu s) ptic' groups  sity or synaps der TTX or m = 29, 34, 25, 2 reen, PSD-95           B  0.00 0.05 0.10 0.15 0.20 0.25 sy na ps es /1 0 µm  o f a xo n le ng th e density, DIV uscimol treatm 1, 21 for contr  = red) co ntr ol ttx un treatmen  8 – 10, (A; A ent and their ol, TTX, untre DIV 8 - 10 tre ate d ( ttx ) mu sc i t of presynap NOVA: p = 0 accompanying ated (TTX), m mo l un tre ate d ( mu s) tic groups 71 .86, B;  untreated uscimol,   Figure ANOV counte untrea  A 0 0 1 1 sy na pt op hy si n/ 10  µ m  o f a xo n le ng th  27: Compari A: p = 0.20) o rparts showed ted (mus) resp  co ntr ol t .0 .5 .0 .5 treatmen son of presyna f presynaptic  no significant ectively.  (Syn  DIV 12 - 14 tx un tre ate d ( ttx ) mu s t of 'presyna ptic puncta den afferent sets un  differences.  n aptophsyin = g  cim ol un tre ate d ( mu s) ptic' groups sity or synaps der TTX or m  = 33, 26, 29, reen; PSD-95          B  co ntr ol 0.0 0.1 0.2 0.3 0.4 sy na ps es /1 0 µm  o f a xo n le ng th e density, DIV uscimol treatm 25, 25 for con  = red) DI  [n on e] TT X un tre treatment of  12 – 14, (A; ent and their trol, TTX, untr V 12 - 14 ate d ( TT X) mu sc im u 'presynaptic' ANOVA: p = accompanying eated (TTX), ol ntr ea ted  (m us )  groups 72 0.06, B;  untreated muscimol,  Figure ANOV counte untrea     A 0 0 0 0 0 1 sy na pt op hy si n/ 10  µ m  o f a xo n le ng th   3.2.6  An ad differ origin  28:  Compari A: p = 0.81) o rparts shows n ted (mus) resp  co ntr ol t .0 .2 .4 .6 .8 .0 treatmen . Age Dep ditional adv ent compart  (and theref son of presyna f presynaptic o significant d ectively. (Syna  DIV 19 - 21 tx un tre ate d ( ttx ) mu s t of 'presyna endent Pl antage to u ments with ore, age) of ptic puncta de afferent sets un ifferences.  n ptophsyin = g  cim ol un tre ate d ( mu s) ptic' groups asticity sing the mic varying age an axon or c nsity or synap der TTX or m = 24, 24, 20, 1 reen, PSD-95          B  c 0.0 0.1 0.2 0.3 0.4 sy na ps es /1 0 µm  o f a xo n le ng th rofludic dev s and still be ell body in se density, DIV uscimol treatm 7, 21 for contr = red) D on tro l ttx un t treatment ices was the  able to dist the central c  19 – 21, (A; ent and their ol, TTX, untre IV 19 - 21 rea ted  (tt x) mu sc im u of 'presynapt  ability to g inctly identi ompartmen  ANOVA: p = accompanying ated (TTX), m ol ntr ea ted  (m us ) ic' groups row cultures fy the point t.  We asked 73  0.85, B;  untreated uscimol,   in  of   wheth regar  Our p in wh reduc 26 – 2  We s older week either  Figure When occur  er the age o ds to the pla revious obs ich effects w ed activity w 8 as the age et up two sy  than the ‘po s younger th  TTX or mu  29: Schemati  the ‘postsy red between f the ‘presy sticity we ob ervations sh ith either T ere no long s in our het stems (Fig. stsynaptic’ an the ‘post scimol for 4 c of 3-compar naptic’ grou  treated and naptic’ grou served in o owed that tr TX or musc er observed erochronic e 29): (A) one group and (B synaptic’ gr 8hrs (A at D tment diagram p was older  untreated g  p or ‘postsy ur model. eatments du imol overla .  Therefore xperiments  in which th ) one in wh oup.  Then a IV 12 – 14  set up for mix  at DIV 26 - roups (Fig. naptic’ grou ring DIV 12 pped, while , we decided . e ‘presynap ich the ‘pre  single ‘pre , B at DIV 2 ed age culture  28 (A), no 30, 31). p had a grea  – 14 was a by DIV 26 –  to use DIV tic’ groups w synaptic’ gr synaptic’ gr 6 – 28). s. differences ter influenc  common tim  28 the effe  12 – 14 and ere two we oups were t oup is treate in synapse n 74 e in e point cts of  DIV eks wo d with  umber  Figure groups reduce weeks ± 1.60   30: Compari  with an older d by TTX app  older than the , n = 33; Mean Presyn Postsy 0 5 10 15 treatme nu m be r of  s yn ap se s pe r fie ld son of the num  postsynaptic t lication while  presynaptic gr  ± SEM aptic neuron naptic neuron TT X nt of 'presyn ber of synapse arget (TTX tre the other grou oups, t test: p s DIV 12 - 14 s DIV 26 - 28 un tre ate d ( TT X) aptic' group s between two atment).  From p remains at no = 0.18; synaps s   sets of affere  DIV 12 – 14 rmal levels. es/field: TTX nts from two d , one presynap In addition, po = 6.73 ± 1.04, ifferent presyn tic group has i stsynaptic neu  n = 33, untrea Treatment Group: Blue  Untreated Group: Green 75 aptic ts activity rons are 2 ted = 9.30  Figure groups activit neuron 29, un  31:  Compari  with an older y reduced by m s are 2 weeks treated = 8.86 Presyn Postsy mu s 0 5 10 15 treatmen nu m be r of  s yn ap se s pe r fie ld son of the num  postsynaptic t uscimol appli older than the ± 0.96, n = 29 aptic neuron naptic neuron cim ol t of 'presyn ber of synaps arget (muscim cation while th presynaptic gr ; Mean ± SEM s DIV 12 - 14 s DIV 26 - 28 un tre ate d ( mu s) aptic' group es between tw ol treatment). e other group oups, t test: p  s o sets of affere  From DIV 12  remains at no = 0.77; synaps nts from two d  – 14, one pres rmal levels.   I es/field: musc ifferent presy ynaptic group n addition, pos imol = 8.45 ± Treatment Group: Blue  Untreated Group: Green 76 naptic  has its tsynaptic 1.05, n =   77  Conversely, if treatment was administered when the ‘presynaptic’ group was older (B), the muscimol treated group had fewer synapses per field compared to the corresponding untreated ‘presynaptic’ group (Fig. 35).  However, there was no difference  if ‘older’ presynaptic groups are treated with TTX (Fig. 34).  From our observations, we concluded that for non receptor specific neural activity inhibition (TTX), the pre- and postsynaptic plasticity contributions were equally important as the cell ages and contributions from both the presynapse and postsynapse were required for the activity dependent effects we monitored to take place.  However, for inhibition by muscimol, we observed a greater importance postsynaptically since ‘presynaptic’ neurons at an advanced age (DIV 26 – 28) continued to display lower numbers of synapses when their activity was lowered as long as the ‘postsynaptic’ targets were still at DIV 12 – 14.  Figure presyn has its postsy 1.81, n  32:  Compari aptic groups w  activity reduc naptic neurons  = 19, untreate Presyn Postsyn 0 5 10 15 20 treatmen nu m be r of  s yn ap se s pe r fie ld son of the num ith a younger ed by TTX app  are 2 weeks y d = 12.78 ± 1 aptic neurons aptic neuron TT X t of 'presyna ber of synaps postsynaptic t lication while ounger than th .67, n = 19; M  DIV 26 - 28 s DIV 12 - 14 un tre ate d ( TT X) ptic' groups es between tw arget (TTX tre  the other grou e presynaptic ean ± SEM o sets of affere atment).  From p remains at n groups, t test: nts from two d  DIV 26 – 28 ormal levels. p = 0.21; syna ifferent older , one presynap  In addition, pses/field: TTX Treatment Group: Blue  Untreated Group: Green 78 tic group  = 9.61 ±  Figure presyn group postsy = 5.38  33:  Compari aptic groups w has its activity naptic neurons  ± 0.68, n = 24 Presyn Postsy mu s 0 5 10 15 treatmen nu m be r of  s yn ap se s pe r fie ld son of the num ith a younger  reduced by m  are 2 weeks y , untreated = 1 aptic neuron naptic neuron cim ol t of 'presyn  *** ber of synaps postsynaptic t uscimol applic ounger than th 0.63 ± 1.07, n s DIV 26 - 28 s DIV 12 - 14 un tre ate d ( mu s) aptic' group es between tw arget (muscim ation while th e presynaptic  = 24; Mean ± s o sets of affere ol treatment). e other group r groups, t test:  SEM nts from two d From DIV 26 emains at nor p = 0.0001; sy ifferent older – 28, one pres mal levels.   In napses/field: m Treatment Group: Blue  Untreated Group: Green 79 ynaptic  addition, uscimol   80 3.3.Discussion  We used a three compartment microfluidic device to create a model of dual input activity dependent synaptic plasticity.  The microfluidic devices allowed us to designate two compartments as ‘presynaptic’ neurons and the central compartment as the ‘postsynaptic’ neuronal group.  The ‘postsynaptic’ group acted as a common target for both ‘presynaptic’ neuronal groups.  The additional advantage to this device was that we were able to create three distinct chemical environments allowing each neuronal group to be treated separately but still able to interact with each other [162].  Using these properties, we created an in vitro situation that could parallel the dual input conditions similar to monocular ocular occlusion in the visual cortex.  We then sought to ascertain whether our microfluidic model would have comparable characteristics to that observed the visual cortex.  By inhibiting one of the two ‘presynaptic’ groups, we did observe a ‘critical period’ in vitro.  Interestingly, depending on the type of inhibition the time course of this ‘critical’ period differed.  The effects due to inhibition by GABAR activation occurred sometime in the first two weeks while the effects of TTX inhibition peaked during the second and third week in vitro.  Whether these results are caused by changes in synapse formation, elimination or some combination of the two processes remains to be determined in future studies.  The earlier start time for the muscimol induced activity dependent changes in our model could be due in part by the maturation of the GABA receptor.  Early in the neuronal development, GABAR activation results in a depolarizing effect [177, 178] and the inhibitory effects of GABA begin to appear around DIV 8 in culture and complete inhibitory effects begins around DIV 14 [177, 179, 180].  While GABAR activation can accelerate the switch to inhibition [177], without additional controls to test for activity, we cannot rule out possible excitatory effects early in neuronal culture development (in particular our experiments from DIV 8 – 10).  81  During the first three time points that we examined,  there was significant difference in the number of synapses present between the ‘postsynaptic’ target group and the ‘presynaptic’ group with reduced neural activity (either by TTX, muscimol or both).  However, when we examined the ‘presynaptic puncta’ density (by FP tagged synaptophysin) and the synaptic density per length of axon, there was no difference between the groups with reduced and basal level activity. This suggested that at least some of the effects observed were related to differential axon growth as a function of activity.  This observation is supported by previous studies that identified molecules important to closing the critical period in the visual cortex  belonged regulators of axonal growth such as chondroitin sulfate proteoglycans and the myelin-signaling proteins NgR [181, 182].  Furthermore, activity has been shown to lead to increased branching in more dense clusters of FP-synaptophysin and to retraction in faintly labeled locations [107].  From this, we can propose a possible mechanism for our model in which (1) increased activity promotes the stability or formation of a synapse allowing the axon to continue to extend and increase the axons propensity to form additional synapses and (2) reduced activity could lead to the opposite effect.  Indeed, we saw evidence of the first scenario (1).   When we reduced DIV 12-14 ‘presynaptic’ neural activity by GABAAR activation, there was a greater number of synapses formed with the opposing labeled ‘presynaptic’ afferents.  For TTX inhibition, it was possible that growth and retraction occurs because while there were no significant differences in the number of synapses between control, treated and untreated groups, there was a difference when comparing only the TTX treated and opposing untreated group.  It should also be noted that chronic application of GABA agonist can lead to internalization of GABAR [183-186].  While the concentrations used in those experiments were much higher [185,  82 186], additional controls will need to be performed before we can rule out any potential effects caused by GABAR internalization.  As neurons develop, the composition of the extracellular matrix and receptors at the synapse changes over time [187-189].  As we gain more information on how new neurons generated during adulthood progress, how they integrate into a mature system becomes a more relevant question that needs to be further explored [190-192].   Adult neurogenesis highlights an interesting situation where one can study the factors that make mature neuronal circuitry (which historically was thought of as rigid) a permissive environment for functional integration [190]. Therefore, an increased understanding about the integration of newly generated neurons into a mature system may have significant implications in neuronal cell replacement therapy for degenerative neurological diseases or regeneration after injury [193].  Under conditions of inhibition by GABAR, it was possible to reintroduce the synaptic changes observed in our model when the age of the ‘postsynaptic’ group was 2 weeks younger than the ‘presynaptic’ neurons.  Meanwhile in the reverse situation, there was no effect.   Therefore, we concluded that under inhibitive conditions of GABAAR activation the age of the ‘postsynaptic’ neurons was a limiting factor for our model.  Recent studies have shown that newly born neurons in adult systems do experience a critical period [194], however, the main assessments were postsynaptic responses such as long term potentiation (LTP).  Our experiments are supported by these previous findings since renewed plasticity occurred when the postsynaptic group was replaced with younger neurons at ages within our previously determined plastic age range.  Our experiments show that different methods of chemical inhibition (in our case, TTX and muscimol) led to distinct differences in the model’s properties.  It is possible that this occurs  83 because of the drugs targets and the associated channel properties.  TTX acts by blocking voltage gated sodium (Na+) channels thereby preventing the propagation of action potentials; this effectively and completely silences the neuron and limits the communication between the ‘presynaptic’ and ‘postsynaptic’ neurons to only spontaneous glutamate release from presynaptic vesicles.  This differs from muscimol which works as an agonist to the GABAA receptor (GABAAR).  GABAAR is a ligand gated ion channel and when activated allows the flow of chloride ions (Cl-) into the neuron [175].  Inhibition by GABAAR activation  is therefore caused by hyperpolarization or shunting which makes the start of an action potential more difficult. Furthermore, a recent study has shown that even under saturating conditions of GABAAR agonist, the majority of GABAAR only had an open probability of approximately 50% [195]. Under these conditions, activity is severely impaired; however if a depolarization was large enough, action potentials could still be generated.  This would lead us to hypothesize that spontaneous activity and reduced levels of neurotransmission (via GABAAR activation) results in different mechanisms of activity dependent synaptic plasticity.    84 4. Cellular and Molecular Mechanisms of Our In Vitro Activity Dependent Synaptic Plasticity Model Using Microfluidics   Critical periods are one of the most commonly studied models for activity dependent morphological and synaptic changes.  One of the more studied models is the visual cortex where cortical connections associated with inputs from 2 different sources (the eyes) can be drastically altered during specific periods of heightened sensitivity to visual stimuli.  Altering the activity between these inputs at specific moments will change how the neural circuitry is formed during development.  The drastic changes that can occur at this time will also last throughout development and adulthood.  Understanding the mechanisms that influence the synaptic balance during development is critical to our understanding about how early influences to the brain affect learning and behaviour in adulthood.  Using our model dual input activity dependent synaptic plasticity in vitro model, we showed that different types of inhibition in primary cortical neurons have different molecular mechanisms.  Treatments to reduce activity in one group of neurons using either the sodium channel blocker, tetrodotoxin (TTX) or the GABA agonist muscimol, both led to less synapses for the inhibited afferents compared to the untreated group.  However, TTX treatment showed dependence on GluR2 internalization while the muscimol treatment showed no such relationship. 4.1. Introduction  Brain development represents an amazing and complex process. Not only does it require the appropriate genetic cues, but it is also dependent on the stimulation it receives from the environment [1, 2].  Because the formation of neural circuits is an extremely dynamic process, activity is a very important part of neural refinement as it can modulate the stability, formation and elimination of synapses [2, 158].  Therefore, any experience that the brain receives during  85 development can significantly alter its function later in life.  Going beyond the synapse, Vaughn and colleagues formulated the ‘synaptotrophic hypothesis’ stating that a stable synapse would in turn stabilize its associated dendritic branch [196, 197] leading to further development.  This hypothesis appears to also apply for axon development [107] suggesting that increased activity would lead to stabilization of new synapses and axon branches while decreased activity would lead to a loss of synapses. 4.1.1. The NMDA Receptor  Activity is a very important part of neural refinement as it can modulate the stability, formation and elimination of synapses [1, 2].  Special interest is given to the NMDA receptor because of its unique property which requires both depolarization and receptor activation to permit ion flow; therefore, the pre- and postsynaptic cells must be excited in concert for activation to occur [198]. Activity by patterned neurotransmission, especially through the NMDA receptor, can promote synapse maturation and stability.  Activation of NMDA receptors by spontaneous activity has been shown to recruit AMPA receptors into previously silent synapses [199].  Interestingly, another study from the same group, later observed that NMDA receptor activation prevented synapse unsilencing [120].  While the two findings appear contradictory, these observations indicate that NMDAR activity is important for the selection process in which both presynaptic and postsynaptic parts will interact to form a synapse.  Furthermore, NMDA receptor activation has been shown to slow turnover of actin filaments within the spine pool, increase actin filament content, recruit AMPA receptors to the postsynaptic membrane and regulate expression of structural proteins to promote synapse maturation [200-203].  Another important factor is that NMDA receptors are highly permeable to Ca2+ ions when activated.  This Ca2+ ion influx can contribute to many downstream intracellular events [198].  86 4.1.2. CamKII  Activation of  NMDA receptors allows calcium to enter the cell leading to the subsequent activation of Ca2+/calmodulin-dependent protein kinase II (CaMKII) [204].  CaMKII is activated when Ca2+ entering the cell binds calmodulin, which subsequently binds CaMKII [204].  Under physiological conditions, neuronal activity can drive CaMKII to cluster postsynaptically [205]. Once activated, CaMKII interacts with the NMDA-R which would increase its translocation to the synapse [206-208].  Furthermore, CaMKII when activated plays an important role to the targeting of AMPA receptors to the synapse [207, 208]. 4.1.3. The AMPA Receptor  AMPA receptors are the primary mediator of fast synaptic transmission in the glutamatergic synapse and are usually found at the postsynaptic region of excitatory synapses.  The majority of AMPA receptors appear to exist in heteromers of GluR1/Glur2 subunits[174, 209].  Their expression in the maturing synapse is thought to appear in stepwise fashion, after the appearance of NMDARs, through specific patterns of network activity [210].  Therefore, its expression is observed to be a response to the activity in the brain.  Moreover, one of the first studies on silent synapses showed that AMPAR blockade increased the number, size and fluorescent intensity of AMPA receptor clusters and rapidly induced the appearance of AMPA receptors at 'silent' synapses [122].  Additional studies also showed that overexpression of the AMPA subunit glutamate receptor 2 (GluR2), lead to an increase in dendritic spine size and density[211, 212]. Recently, it has been shown that postsynaptic AMPA receptors, specifically the GluR2 subunit, played an important role in synapse stability and its expression levels were directly correlated to synapse number [213].   87 The critical period of the visual cortex serves as an excellent model to observe changes in the connective balance that occurs during development when two different afferents interact in the same region of the brain [1, 3].  To examine the molecular mechanisms involved we used our in vitro dual input model.  Using a three compartment microfluidic device, we created an environment in which two separate groups of neurons interacted to form synapses with a third group of neurons, the target.  Because we were able to maintain three distinct chemical environments, we were able to alter the activity the one presynaptic group and simultaneously inhibit some of the downstream mechanisms of the postsynaptic target group chemically. 4.2. Results  In our previous study, the critical periods for activity dependent synaptic plasticity in our model for TTX or muscimol treatments overlapped during DIV 12-14.  We also showed postsynaptic effects at that age were important to maintaining this type of plasticity.  Therefore, we chose to use this time point for further investigation of the molecular mechanisms involved.  To gain further insight into the possible molecular mechanisms in our model, we used our three compartment device to not only decrease the activity of a single ‘presynaptic’ group but also interrupt postsynaptic signaling of the target group simultaneously.  To examine the properties of our model, we once again looked at the number of synapses and the changes that occurred under our experimental conditions.  For our study, we defined a ‘synapse’ as a PSD-95 puncta colocalized with a fluorescent protein tagged synaptophysin puncta (FP-syn).    88 4.2.1. Activity Dependent Changes Caused by TTX or Muscimol Are NMDA Receptor (NMDAR) Dependent  NMDARs role in critical periods and activity dependent plasticity have been extensively studied because of their ability to act as a sensor for synchronous activity [198] and their change in composition as neurons mature [187, 188, 214].  We decided to determine if NMDA-R activation played a role in our model by applying the NMDAR antagonist APV (50 µM) to the ‘postsynapatic’ target neurons in our three compartment device.  The NMDAR block occurred while one of the ‘presynaptic’ groups had their neural activity inhibited by either TTX or muscimol.  All treatments occurred for the same 48hr duration from DIV 12-14.  In our previous study, at DIV 12 – 14, when one of the ‘presynaptic’ groups was reduced in activity, it had fewer synapses compared to its opposing untreated group (Fig. 14, 18).  However, when APV was added to the media of the ‘postsynaptic’ group the differences generated TTX or muscimol were no longer present (Fig 34, 35).  Furthermore, the presynaptic puncta and synaptic density remained similar throughout the groups (Fig 36).  This showed that the result was not simply that NMDAR inhibition prevents the formation of synapses.  Synaptic densities remained similar when compared to groups in which APV was not applied ‘postsynaptically.’  This shows that NMDAR activation is an important component in the processes involved in our dual input model.        Figure groups group treatm postsy 1.59, n nu m be r of sy na ps es pe r fie ld  34:  Compari  (TTX treatme has its activity ent is indicated naptic NMDA  = 39; Mean ± DIV 12 0 5 10 15 treatmen nu m be r of  s yn ap se s pe r fie ld son of the num nt) at DIV 12  reduced by T  by [ ].  The ta R activity.  t te  SEM -14 (postsyna TT X t of presyna ber of synaps -14 with posts TX application rget postsynap st: p = 0.051; ptic NMDA blo un tre ate d ( TT X) ptic groups es between tw ynaptic group  while the oth tic group is si synapses/field ck, APV) o sets of affere inhibition of N er group remai multaneously : TTX = 8.44 ± nts from two d MDAR activi ns at normal l treated with A  0.94, n = 39, ifferent presy ty.  One presy evels.  Postsyn PV (50 um) to  untreated = 1 Treatment Group: Blue  Untreated Group: Green 89 naptic naptic aptic  block 1.59 ±  Figure groups presyn Postsy um) to untrea  35:  Compari  (muscimol tre aptic group ha naptic treatme  block postsyn ted = 12.29 ± DIV 12 mu s 0 5 10 15 20 treatme nu m be r of  s yn ap se s pe r fie ld son of the num atment) at DI s its activity re nt is indicated aptic NMDAR 157, n = 38; M - 14 (postsyna cim ol nt of presyn ber of synaps V 12 – 14 with duced by mus  by [ ].  The tar  activity.  t te ean ± SEM ptic NMDA bl un tre ate d ( mu s) aptic groups es between tw  postsynaptic cimol applicat get postsynap st: p = 0.95; sy ock, APV) o sets of affere group inhibitio ion while the o tic group is sim napses/field: m nts from two d n of NMDAR ther group rem ultaneously t uscimol = 12 T G B  U G G ifferent presy  activity.  One ains at norma reated with AP .13 ± 2.20, n = reatment roup: lue ntreated roup: reen 90 naptic  l levels. V (50  38,  91 Figure 36:  Comparison of presynaptic puncta density or synpase density (postsynaptic group treated with APV) (A; ANOVA: p = 0.30, B; ANOVA: p = 0.084) of presynaptic afferent sets  of various treatments and their accompanying untreated counterparts.  Postsynaptic treatment is indicated by [ ].  Postsynaptic target group is treated with APV.  Presynaptic groups are treated with TTX or muscimol.  n = 33, 34, 27, 19, 27, 27 for control [none], control [APV], TTX [APV], untreated (TTX) [APV], muscimol [APV], untreated (mus) [APV] respectively. A                                                                      B DIV 12 - 14  Postsynaptic group treated with APV co ntr ol [no ne ] co ntr ol [A PV ] TT X [ AP V] un tre ate d ( TT X)  [A PV ] mu sc im ol [A PV ] un tre ate d ( mu s) [A PV ] 0.0 0.5 1.0 1.5 Treatment of 'presynaptic' groups APV treatment only on postsynaptic group sy na pt op hy si n/ 10  µ m  o f a xo n le ng th  DIV 12 - 14 Postsynaptic group treated with APV co ntr ol [no ne ] TT X [ AP V] un tre ate d ( TT X)  [A PV ] mu sc im ol [A PV ] un tre ate d ( mu s) [A PV ] 0.0 0.1 0.2 0.3 0.4 treatment of 'presynaptic' groups sy na ps es /1 0 µm  o f a xo n le ng th    4.2.2. Activity Dependent Changes Caused by Muscimol Requires CamKII Activity  CamKII activation occurs downstream of NMDAR activation when calcium binds calmodulin leading to CamKII’s activation [215]. Once activated, one important role for CamKII is the phosphorylation of AMPARs leading to their insertion into the synaptic membrance [202, 216]. To determine if CamKII activity was essential for the changes induced in our model, we used the CamKII inhibitor KN-93 (1 µM) to block CamKII activity in the ‘postsynaptic’ neurons in our model.  Figure groups group treatm postsy 1.07, n nu m be r of sy na ps es pe r fie ld  37:  Compari  (TTX treatme has its activity ent is indicated naptic CamKI  = 35; Mean ± DIV 12 - 1 0 5 10 15 treatmen nu m be r of  s yn ap se s pe r fie ld son of the num nt) at DIV 12  reduced by T  by [ ]. The ta I activity.   t te  SEM 4 (postsynapt TT X t of presyna  * ber of synaps – 14 with post TX application rget postsynap st: p = 0.025; s ic CamKII bloc un tre ate d ( TT X) ptic groups es between tw synaptic group  while the oth tic group is sim ynapses/field: k, KN93) o sets of affere  inhibition of er group remai ultaneously t  TTX  = 6.86 ± nts from two d CamKII activi ns at normal l reated with KN  1.03, n = 35, T G B  U G G ifferent presy ty.  One presy evels.  Postsyn 93 (1 um) to  untreated = 1 reatment roup: lue ntreated roup: reen 92 naptic naptic aptic block 0.26 ±   Figure groups presyn Postsy to bloc 12.32  nu m be r of  s yn ap se s pe r fie ld  38:  Compari  (muscimol tre aptic group ha naptic treatme k postsynaptic ± 2.23, n = 37 DIV 12 - 1 mu s 0 5 10 15 treatme son of the num atment) at DI s its activity re nt is indicated  CamKII activ ; Mean ± SEM 4 (postsynap cim ol nt of presyn ber of synaps V 12 – 14 with duced by mus  by [ ]. The tar ity.  t test: p =  tic CamKII blo un tre ate d ( mu s) aptic groups es between tw  postsynaptic cimol applicat get postsynapt  0.60; synapse ck, KN93) o sets of affere group inhibitio ion while the o ic group is sim s/field: muscim nts from two d n of CamKII ther group rem ultaneously tr ol = 11.14 ± T G B  U G G ifferent presy activity.  One ains at norma eated with KN 1.79, n = 37, u reatment roup: lue ntreated roup: reen 93 naptic l levels. 93 (1 um) ntreated =  94 Figure 39:  Comparison of presynaptic puncta density or synpase density (postsynaptic group treated with KN-93) (A; ANOVA: p = 0.18, B; ANOVA: p = 0.15) of presynaptic afferent sets of various treatments and their accompanying untreated counterparts.  Postsynaptic target group is treated with KN93.  Postsynaptic treatment is indicated by [ ]. Presynaptic groups are treated with TTX or muscimol.  n = 33, 32, 24, 23, 21, 26 for control [none], control [KN93], TTX [KN93], untreated (TTX) [KN93], muscimol [KN93], untreated (mus) [KN93] respectively.  A                                                                      B DIV 12 - 14 Postsynaptic group treated with KN93 co ntr ol [no ne ] co ntr ol [K N9 3] TT X [ KN 93 ] un tre ate d ( TT X)  [K N9 3] mu sc im ol [K N9 3] un tre ate d ( mu s) [K N9 3] 0.0 0.5 1.0 1.5 Treatment of 'presynaptic' groups KN93 treatment only on postsynaptic group sy na pt op hy si n/ 10  µ m  o f a xo n le ng th  DIV 12 - 14 Postsynaptic group treated with KN93 co ntr ol [no ne ] TT X [ KN 93 ] un tre ate d ( TT X)  [K N9 3] mu sc im ol [K N9 3] un tre ate d ( mu s) [K N9 3] 0.0 0.1 0.2 0.3 0.4 treatment of 'presynaptic' groups sy na ps es /1 0 µm  o f a xo n le ng th    Interestingly, synaptic changes caused by TTX inhibition were unaffected by inhibiting CamKII activity (Fig. 37). However, activity induced changes caused by muscimol treatement were no longer present when KN-93 was applied ‘postsynaptically.’  This indicated a CamKII requirement for synaptic change under these in vitro conditions (Fig. 38). Further examination into the ‘presynaptic puncta’ and synaptic density also showed no differences between treated, untreated and control conditions (Fig. 39).  4.2.3. AMPA Receptor (AMPAR) Internalization is Required for Activity Dependent Changes Caused by TTX Inhibition  GlurR2 expression levels have been shown to correlate with synapse number in vitro [213]. Furthermore, GluR2 expression on the synaptic membrane correlates with  Figure groups has its group synaps Postsy the act 17.00 green)  A P 0 5 10 15 20 trea nu m be r of  s yn ap se s pe r fie ld   incre in sur been  40:  Compari  (TTX treatme  activity reduc is simultaneou es/field: TTX naptic treatme ivity induced ± 1.74, n = 49 .  ostsynaptic gr TT X [ sc ram ] tment of 'pre ased synapse face Glur2 c shown to be  * son of the num nt) at DIV 12 ed by TTX app sly treated wit + 3Y = 16.69 nt is indicated changes (A; t t ; Mean ± SEM  oup treated w un tre ate d ( TT X) [ synaptic' gro  size, densi ould play a  regulated b ber of synaps – 14 with the p lication while h Glur23y (1 u ± 1.49, n = 48  by [ ]. Applica est: p = 0.039; ).  Associated  DIV 12 - 1 ith 3Y scrambl  s cra m] ups ty and stabil  potential ro y tyrosine p es between tw ostsynaptic g  the other grou m) to block po , untreated + 3 tion of a scram  synapses/field images below                  B 4 e   ity [211, 21 le in our in v hosphorylat o sets of affere roup treated w p remains at n stsynaptic Glu y = 20.17 ± 1. ble peptide ( : TTX + 3Y = graphs (Treate  Posts TT X 0 5 10 15 20 25 treatment o nu m be r of  s yn ap se s pe r fie ld 2].  Therefo itro model. ion at its car nts from two d ith GluR23Y.  O ormal levels. R2 internaliza 78, n = 48; Me 1 um) postsyna  12.08 ± 1.58, d group: blue, ynaptic group  [3 Y] un tre a f 'presynapti re, we specu   The endoc boxyl tail [2 ifferent presy ne presynapt The target pos tion (B; t test: an ± SEM). ptically did no  n = 49, untrea  Untreated gro DIV  treated with G ted  (T TX ) [3 y] c' groups lated that a ytosis of Glu 17].  We us 95 naptic ic group tsynaptic p = 0.14; t prevent ted + 3y = up, 12 - 14 lur23y   decrease R2 has ed a  TAT- to det Figure groups group postsy test: p ± SEM not pre 63; un Untrea  A nu m be r of  s yn ap se s pe r fie ld   Durin with GluR23y pep ermine if en  41:  Compari  (muscimol tre has its activity naptic group i = 0.0083; syn ).  Postsynapt vent the activ treated + 3y = ted group, gre  Postsynaptic mu sc im ol [sc ram ] 0 10 20 30 treatmen g neural act TAT-GluR2 * tide, which docytosis o son of the num atment) at DI  reduced by m s simultaneous apses/field: mu ic treatment is ity induced cha 22.49 ± 2.58, en).   group treated un tre ate d ( mu t of 'presyna ivity inhibit 3y for the sam **  has been sh f GluR2 is e ber of synaps V 12 – 14 with uscimol applic ly treated with scimol + 3Y =  indicated by [ nges (A; t test n = 63; Mean ±  DIV 12  with 3Y scram s) [sc ram ] ptic' groups ion of one ‘p e duration own to bloc ssential in o es between tw  the postsynap ation while th  Glur23y (1 µm  16.90 ± 20.4  ]. Application : p = 0.0001; s  SEM). Assoc              B - 14 ble  1 2 3 4 nu m be r of  s yn ap se s pe r fie ld  resynaptic’ .  Under TTX k regulated ur model. o sets of affere tic group treat e other group r ) to block pos , n = 42, untrea  of a scramble ynapses/field: iated images b Postsyna mu sc im ol [3y ] 0 0 0 0 0 treatmen  group, the  inhibition  ** AMPA end nts from two d ed with GluR2 emains at nor tsynaptic GluR ted + 3y = 26  peptide (1 µm  muscimol + 3 elow graphs ( ptic group tre un tre ate d ( m t of 'presyna ‘postsynapti  with TAT-G  ocytosis [21 ifferent presy 3y.  One presy mal levels.  Th 2 internalizat .62 ± 2.95, n = ) postsynaptic Y = 10.97 ± 1 Treated group DIV 12 - ated with Glur2 us ) [3 y] ptic' groups c’ group wa luR23y pre 96 7-219], naptic naptic e target ion (B t  42; Mean ally did .27, n = : blue, 14 3y   s treated sent in  97 the postsynaptic group, there was no difference in number of labeled synapses observed for the inhibited presynaptic group compared to untreated (Fig 40). Figure 42:  Comparison of presynaptic puncta density of presynaptic afferent sets (with the postsynaptic group treated with GluR23Y) of various treatments and their accompanying untreated counterparts.  Postsynaptic target group is treated with either GluR23y or scramble peptide.  Postsynaptic treatment is indicated by [ ].  Presynaptic groups are treated with TTX or muscimol.  ANOVA: p = 0.058.  n = 33, 38, 27, 32, 22, 26, 29, 21, 28, 32, from left to right on graph.  DIV 12 - 14  Postsynaptic group treated with 3Y (3Y) or scramble (scram) peptide co ntr ol [no ne ] co ntr ol [3y ] co ntr ol [sc ram ] TT X [ 3Y ] TT X [ sc ram ] un tre ate d ( TT X)  [3 Y] un tre ate d ( TT X)  [s cra m] mu sc im ol [3Y ] mu sc im ol [sc ram ] un tre ate d ( mu s) [3Y ] un tre ate d ( mu s) [sc ram ] 0.0 0.5 1.0 1.5 Treatment of 'presynaptic' groups 3Y or scram treatment only on postsynaptic group sy na pt op hy si n/ 10  µ m  o f a xo n le ng th   However, when presynaptic inhibition was caused by the GABA agonist, muscimol, TAT- GluR23y did not block the synaptic changes.  The GABA inhibited ‘presynaptic’ neurons still formed fewer labeled synapses with the ‘postsynaptic’ target group compared to the opposing untreated group (Fig 41).  Consistent with the rest of our results, there were no significant differences in the presynaptic puncta or synaptic densities between the ‘presynaptic’ groups’ axons (Fig 42, 43).    98 Figure 43:   Comparison of synapse density of presynaptic afferent sets (with the postsynaptic group treated with GluR23Y) of various treatments and their accompanying untreated counterparts.  Postsynaptic target group is treated with either GluR3y or scramble peptide.  Postsynaptic treatment is indicated by [ ].  Presynaptic groups are treated with TTX or muscimol.  ANOVA: p = 0.33.  n = 33, 38, 27, 32, 22, 26, 29, 21, 28, 32, from left to right on graph.   DIV 12 - 14  Postsynaptic group treated with 3Y (3Y) or scramble (scram) peptide co ntr ol [no ne ] co ntr ol [3y ] co ntr ol  [s cra m] TT X [ 3y ] TT X [ sc ram ] un tre ate d ( TT X)  [3 y] un tre ate d ( TT X)   [s cra m] mu sc im ol [3y ] mu sc im ol  [s cra m] un tre ate d ( mu s) [3y ] un tre ate d ( mu s) [sc ram ] 0.0 0.1 0.2 0.3 0.4 0.5 Treatment of 'presynaptic' groups 3Y or scram treatment only on postsynaptic group sy na ps es /1 0 µm  o f a xo n le ng th   Our results showed that the changes induced in our dual input activity dependent synaptic plasticity model were NMDA dependent.   However, depending on the method of inhibition, the pathways diverged to one that required the internalization of AMPAR and one that was independent of it. 4.2.4.Cellular Mechanisms  Neural circuit formation is a dynamic process [2] involving the simultaneous regulation of addition, stabilization and elimination of dendritic and axonal processes by neural activity [220- 222].  Early investigations examining the cellular effects of systems in which the activity  99 between neurons differed observed mechanisms resulting in branch elimination [5, 156].  These experiments used models such as NMJ or purkinje cells where differences in activity resulted in single innervation of a postsynaptic target.  However, many parts of the CNS allow for multi- innervation of postsynaptic targets meaning the cellular dynamics of less active afferents may not behave the same compared to uni-innervated neuronal circuitry.  Therefore, it is important to understand how our model reflects the cellular mechanisms that occur in an activity dependent setting.  We decided to take a qualitative look at the types of dynamic changes that exist in our model at two weeks in vitro.  Using long term time lapse imaging, we examined the axon growth of the treated and untreated ‘presynaptic’ neurons in the ‘postsynaptic’ neuron compartment.  Table 1: Summary of the observed axon changes in a field observed over 24 hours under various treatment conditions.  ‘Growth’ is defined as an increase in axon length, ‘retraction’ a decrease in axon length and ‘unchanged’ as no significant/minimal change of axon length from initial to final observation.  (% ± SEM)  ‘Presynaptic’ Neuron Group % Growth  % Retraction % Unchanged Control (n = 20) 68 ± 9 11 ± 5 21 ± 8 Muscimol treated (n = 16) 70 ± 9 18 ± 8 11 ± 5 Muscimol untreated (n = 15) 89 ± 4 4 ± 3 7 ± 4 TTX treated (n = 17) 71 ± 9 7 ± 4 22 ± 8 TTX untreated (n = 14) 90 ± 6 6 ± 4 4 ± 4  We decided to qualitatively classify the change in axon size into three groups: growth, retraction and unchanged.  We defined ‘growth’ as an overall net gain in axon length from initial to final observation,  ‘retraction’ as an overall net loss of axon length from initial to final observation and ‘unchanged’ as no significant or minimal change of axon length.   100 In our experiments, regardless of treatment, the majority of axons experienced an overall net gain in axon length over time (Table 1).   ANOVA showed no significant differences between the groups for %growth, %retraction, %unchanged (p = 0.14, p = 0.33, p = 0.2 respectively).  Because the majority of axon dynamics in our model was composed of increased growth (and there were no differences in the type of growth cone advancement among the treatment groups), we decided to further examine how changes in activity to one ‘presynaptic’ neuronal group affected positive axon growth.  Therefore, we used time lapse imaging to estimate the rate of growth for positively growing axons under our treatment conditions.  When a ‘presynaptic’ neuronal group’s activity was inhibited with TTX, the axon length increased at a slower pace compared to the opposing neuron group not treated with TTX (Fig 44a).  The same effect occurred when one ‘presynaptic’ group was treated with muscimol (Fig 44b).  When muscimol was used, the untreated ‘presynaptic’ group’s axon once again increased at a faster pace compared to the treated side.  Furthermore, when TTX was used to reduce ‘presynaptic’ neuronal activity, the rate of axon growth decreased for the treated group when compared to a control (where both presynaptic groups remain untreated) (Fig. 45).   Moreover, rate of axon growth for the opposing untreated group remained the same compared to our control group (Fig. 45).  When muscimol was applied instead, the rate of axon growth for the treated group remained similar to control but the opposing untreated ‘presynaptic’ group showed an increase in axon growth rate compared to control (Fig. 45).      Figure reduce 0.78, a 0.58, a norma  A  44:  Compari d by TTX (A; xons = 14; Me xons = 11; un l levels. Image  0 2 4 6 8 treatment D is ta nc e tr av el ed  e ve ry  3 0 m in  (u m /3 0m in ) son of axon gr  t test: p = 0.00 an ± SEM) or treated = 9.25 s were collect  ttx u  of 'presynap  ** owth rate betw 4; axon growt  muscimol (B; ± 0.88, axons= ed every 30mi  un tre ate d ( ttx ) tic' groups een 2 presyna h rates: TTX t  t test: p = 0.00 11; Mean ± SE n after treatme            B ptic groups.  O reated = 2.74 ± 06; axon grow M) applicatio nt up to 24hrs 0 5 10 15 treatmen D is ta nc e tr av el ed  e ve ry  3 0 m in  (u m /3 0m in ) ne presynapt  0.42, axons = th rates:  mus n while the ot . mu s t of 'presyna  *** ic group has it  11; untreated cimol treated = her group rem un tre ate d ( mu s) ptic' groups 101 s activity  = 5.78 ±  4.99 ± ains at     102  Figure 45:  Comparison of axon growth rate against control conditions.  (ANOVA: p < 0.0001, post hoc = Turkey test; axon growth rate of control = 6.34 ± 0.73, axons = 17; Mean ± SEM).  (** and * indicate significance against control) co ntr ol TT X un tre ate d ( TT X) mu sc im ol un tre ate d ( mu s) 0 5 10 15 treatment of 'presynaptic' groups (opp mus and opp ttx are the untreated  competing 'presynaptic' groups) D is ta nc e tr av el ed  e ve ry  3 0 m in  (u m /3 0m in )   We once again saw the divergent pathways that existed depending on the type of presynaptic chemical inhibition used.  Consistent with our previous observations, TTX inhibition exerted a greater influence on the inhibited group while inhibition by GABAAR activation (muscimol) showed a greater influence towards the untreated ‘presynaptic’ group at normal basal activity levels. 4.3. Discussion  In this series of experiments, we showed that NMDA activation was a crucial part of synaptic change under our dual input with differing activity level conditions.  We demonstrated this by blocking NMDAR function with APV while reducing the neural activity of one presynaptic group.  The NMDAR block prevented our previously observed synaptic differences (in the ** *  103 number of synapses formed) when one of the ‘presynaptic’ groups had reduced neural activity. The changes in our model being NMDAR activation dependent was consistent with observations in the adult mouse visual cortex  where depriving the dominant contralateral eye of vision leads to a persistent, NMDA receptor-dependent enhancement of the weak ipsilateral-eye inputs [223]. Moreover, our result was consistent with those obtained in systems other than the visual cortex as  NMDAR dependence has also been observed in the critical period of the mouse barrel cortex [224].  NMDA dependence and critical plasticity plasticity is tightly linked with long term potentiation (LTP) and long term depression (LTD) [225, 226].  NMDAR activation results in an influx of Ca2+ which eventually leads to further downstream activation of CamKII.  Our results showed than CamKII activity was necessary for the synaptic changes induced by muscimol treatment in our model.  CamKII has been widely considered an essential molecule for LTP [227].  CamKII increases synaptic strength by either enhancing the channel conductance of AMPARs [228] or by prompting insertion of additional AMPAR into the synapse [206-208, 229].  The latter has also been associated with an increase in dendritic spine size.  Previous studies have demonstrated that LTP driven AMPAR insertion into the synapse is necessary and sufficient to increase and stabilize spine density [230] and that this process is also CamKII dependent [208].  Furthermore, because of the multiple scaffolding protein binding partners that AMPARs have, their translocation can further change the scaffolding structure and stability of the synapse [129, 231].  Additional studies have shown that a stable synapse can stabilize its asscociated dendritic branch [196, 197].  This phenomenon was also observed to occur in axons with studies showing that stable presynaptic terminals would result in further branching and growth of the axon [107].  In addition, stable presynaptic sites on axons were also predictive of where synapses would form [82].  Therefore, the importance of CamKII activation in our model  104 may be due to its role in stabilizing synapses with higher activity and promoting the growth of those associated axons.  AMPAR internalization has also been observed to play an important role in activity dependent plasticity and the elimination of synapses [213, 232].  Long term depression (LTD) has long been implicated as an important factor for the ocular dominance column shifts seen in the visual cortex during the critical period [69, 70, 233].  One major form of LTD involves the internalization of AMPAR from the synapse, more specifically, the GluR2 subunit.  It has been observed that the extracellular N-terminal domain of the GluR2 subunit is important to maintain the stability of opposing presynaptic terminals [213].  In addition, Rap2, which has been shown to control synaptic removal of AMPAR during depotentiation [234] has also been implicated in synaptic depression, the loss of surface GluR2 and the retraction of axons and dendrites [235]. Using the TAT-GluR23y peptide, which has been shown to block the regulated endocytosis of sGluR2 [217-219], would determine if AMPA endocytosis was a necessary component of the activity dependent changes in our model.  This was indeed the case when one of the two presynaptic groups was treated with TTX.  However, the prevention of GluR2 internalization had no effect when inhibition by GABAAR activation (muscimol) was used to reduce presynaptic neural activity.   These experiments further supported our previous observations (previous chapter) showing that TTX and GABA inhibition involve different mechanisms in regards to activity dependent plasticity in our model.  We found that at DIV 12-14, TTX treatment of a presynaptic group resulted in a fewer labeled synapses compared to the opposing untreated group.   Additional analysis of the TTX treatment showed a decrease in axon growth relative to the untreated ‘presynaptic’ group.  Furthermore, it appeared the changes resulted from a contribution of both ‘presynaptic’ groups.  However, with  105 muscimol induced inhibition at DIV 12 – 14, the observed differences in the number of synapses between treated and untreated groups appeared to be more closely related to changes in the untreated presynaptic group.  One possible explanation of these results is the possible roles that LTP and LTD play in our in vitro model. There also exists the possibility for differentiating factors involving Hebbian and homeostatic plasticity.  Previous studies looking into the mechanism of OD shifts in the visual cortex determined that NMDAR activity was a critical component to the process [27].  Our model also supports those previous findings as blocking NMDARs prevented the synaptic changes that occurred with both of our chemical inhibition treatments.  Moreover, Ca2+ influx through NMDARs can trigger either LTP or LTD depending on the type of stimulation [236]. Our models showed that further downstream of NMDAR activation that the synaptic changes caused by muscimol treatment required CamKII activity while TTX treatment did not.  NMDA dependence coupled with CamKII activity dependence shows evidence for an LTP dependent mechanism [206, 207, 216, 237].  Synaptic shifts requiring NMDAR activity and GluR2 internalization hints towards a possible LTD mechanism [45, 61, 217]) for our TTX treatment model.  Furthermore, a recent investigation observed that NMDAR dependent LTD did not require the involvement of CamKII [238] providing us with further evidence towards a difference in Hebbian mechanisms between our two treatments.  This suggests that changes that occurred in our model may depend on LTP and LTD for muscimol and TTX treatment respectively.  However, we also cannot rule out the possible involvement of homeostatic mechanisms potentially playing a role in our model.  Several investigations also have shown the involvement of homeostatic mechanisms on critical period OD shifts in the visual cortex [74, 78].  One form  106 of homeostatic plasticity, synaptic scaling, has been shown to require the Glur2 subunit of AMPARs [239].  The investigators also observed that synaptic scaling could be blocked by interfering with C-tail interactions of the GluR2 subunit.  The Glur23y peptide we use to prevent the internalization of AMPAR also interferes with a specific segment of the C-tail on the subunit [217].  While there is no documented information regarding whether the specific interaction the 3Y peptide interrupts is involved with synaptic scaling, we also cannot currently rule out that possibility.  Therefore, additional experiments would be required to further determine the possible roles that Hebbian and homeostatic plasticity have in our model.  Our preliminary live imaging experiments suggested that changing the activity of a presynaptic group resulted in changing the rate at which axons grow.  Our experiments supported previous findings from other investigators that have demonstrated the importance of molecules related to axonal growth in ending the critical period [240].  Interestingly, TTX inhibition affected the growth rate of the ‘presynaptic’ group with reduced activity while inhibition by GABA agonist affected the growth rate of the ‘presynaptic’ group with normal activity levels.  TTX inhibition reduced the axon growth rate for the treated neurons while inhibition by GABA agonist increased the axon growth rate for the untreated ‘presynaptic’ neurons.  This series of experiments supported our previous observations that one of the major influences of the induced synaptic shifts in our in vitro system was due to changes in axon growth.  However, follow up experiments should be performed to determine whether changes in axon growth are directly related to increased synapse number and stability in our in vitro model as seen in other systems [107].  Previous time lapse imaging experiments have shown that neurons grow by a very dynamic process of addition and retraction of filopodia [241].  These branches have a relatively short life  107 spans and it is widely hypothesized that forming proper synaptic connections allows for the branch to be stabilized and promote growth [107, 241].  To properly observe the branch dynamics in our model, future experiments with a high frequency of image collection would be required.  This would provide further insight into how changes in activity can affect the absolute motility of interacting axonal branches.  In conclusion, we propose that our model in which presynaptic neuronal activity was reduced by sodium channel block resulted in fewer synapses being formed by the group with reduced activity by using mechanisms LTD and synaptic scaling.  However, when activity is reduced by GABA agonists, the induced synaptic differnces in our model no longer required the internalization of the AMPAR, Glur2.  This would potentially signify a greater role for LTP mechanisms involving the ‘presynaptic’ group at normal activity levels.  In addition, activity inhibition by TTX or muscimol changes the rate of axon growth between the two designated ‘presynaptic’ neuronal groups in our model.  However, while TTX treatment reduced the axon growth rate of the inhibited group, inhibition by GABAR activation (muscimol) increased the growth rate of the untreated ‘presynaptic’ group.  These results further demonstrated that changes in neural activity between afferents could shift the synaptic balance between them and that the type of inhibition also could \ lead to very distinct mechanisms.  108 5. General Discussion  5.1. A Novel Model for Activity Dependent Synaptic Plasticity In Vitro  Using a three compartment microfluidic device, we created a new model of dependent synaptic plasticity with dual inputs of varying activity.  By designating the two side compartments as ‘presynaptic’ and the central compartment as ‘postsynaptic,’ we created a situation in which two different sources of afferents can form synapses with a common postsynaptic target.  In addition, this device can create three separate chemical environments allowing each neuronal group to be in different chemical environments while maintaining interactions with each other [162]. Therefore, the physical properties of these microfluidic devices allowed us to create an in vitro situation which has similarities to the ocular occlusion situation in the visual cortex (2 inputs with different activity levels) but allowed for greater flexibility and control than is possible in vivo.  In the visual cortex, when one eye is closed during development, afferents belonging to that eye maintain fewer connections in the visual cortex compared to the open eye.  Our model allowed us to chemically reduce the activity one of the ‘presynaptic’ groups without affecting the other. In our model, we chose to reduce neuronal activity either by application of the sodium channel blocker, TTX, or the GABAR agonist, muscimol.  When one ‘presynaptic’ group’s activity levels was reduced two weeks into development (DIV 12 – 14), more synapses found ‘postsynaptically’ were associated with the uninhibited ‘presynaptic’ group relative to the inhibited group.  Furthermore, at this time point, both chemical applications produced similar results.   109 In vivo, these drastic changes on connectivity occur at specific times during development known as critical periods [19].  Critical periods are time points during which neurons are most susceptible to change from environmental experience and influence.  To determine if our model also contained a critical period, we varied the time points when neural activity inhibition of a single ‘presynaptic’ group would occur.  Spanning four weeks of in vitro development, we saw a defined time period during which induced synaptic differences occurred for both of our chosen chemical treatments.  Those experiments allowed us to conclude that a critical period does exist in our model.  Interestingly, the time window for these induced synaptic changes in our model depended on the specific treatment we administered.  The visual cortex is one of the most popular models to study activity dependent plasticity and critical periods but its complexity does pose some problems.  The heterogeneous nature of all the multiple inputs and the multi mechanistic possibilities that can exist even in the same region of the brain greatly increases the difficulty in understanding this process.  By establishing a new in vitro model for activity dependent synaptic plasticity, we provided another simpler approach to further study the mechanisms behind synaptic plasticity involving dual inputs. 5.2. Molecular Mechanisms of Our In Vitro Model  NMDAR activation has been shown to be an important step in causing OD shifts in the visual cortex and critical periods in general [27, 223, 242].  By infusing NMDAR antagonists while MD is being administered, investigators have prevented OD shifts by MD [27].  Consistent with previous investigations, we observed that NMDA activation is a crucial part of synapse selection in our model.  We blocked NMDAR function with APV in the postsynaptic group, while simultaneously reducing the neural activity of one ‘presynaptic’ group (DIV 12 – 14).  Under these conditions, the synaptic changes that occurred in our model no longer happened.  110 Therefore, we concluded that postsynaptic NMDAR activation was a requirement for our model regardless of the chemical inhibition methods used.  Furthermore, NMDAR dependence and critical period plasticity are linked with long term potentiation (LTP) and long term depression (LTD) [27, 71, 242].  NMDAR activation results in an influx of Ca2+ which can eventually lead to downstream activation of CamKII [215].  CamKII activity has also been identified as an important requirement for OD shifts [65, 66, 243]. Using CamKII deficient mice, investigators observed that visual cortex plasticity was greatly diminished [65].  We applied the CamKII inhibitor, KN-93, postsynaptically during our activity deprivation experiments and examined the synapse distribution between the two presynaptic groups induced by muscimol application.  Interestingly, the changes caused by TTX treatment remained under reduced CamKII activity.  Several previous studies have associated CamKII activation with LTP [244-247].  CamKII is widely considered an essential molecule for LTP [227] because of its contributions to increasing synaptic strength by enhancement of AMPAR channel conductance [228] and its role in facilitating the insertion of additional AMPARs into the synapse [206-208, 229].    Furthermore, previous studies have demonstrated that LTP driven AMPAR insertion into the synapse is necessary and sufficient to increase and stabilize spine density [230] which would indicate a possible role for LTP in our model when muscimol is applied.  Moreover, LTD has been observed to occur independently of CamKII activity [238]. Therefore, the difference in mechanism we observed in our model under CamKII activity inhibition may be due to different requirements for LTP and LTD.  The role of LTD is more frequently investigated for visual cortex plasticity and critical periods. It is also considered an important factor for ocular dominance column shifts seen in the visual cortex during the critical period [69, 233].  One major change resulting from LTD is the  111 internalization of AMPAR from the synapse, more specifically GluR2.  AMPAR internalization also plays an important role in the elimination of synapses [130].  AMPAR internalization is regulated by phosphorylation of the C-tail and blocking interaction is an effective method to prevent GluR2 internalization [217, 219].  The TAT-GluR23y peptide (3Y) is short peptide sequence derived from the C-tail of GluR2 and shown to effectively prevent the regulated endocytosis of GluR2 [217-219].  When 3Y was applied to the postsynaptic group in our model, TTX induced the synaptic shifts in our model were blocked.  However, 3Y application had no preventative effect when muscimol was used as a ‘presynaptic’ neural activity inhibitor.  While our model appears to be dependent on similar essential receptors and molecules that are associated with LTP and LTD, we cannot definitely conclude that those specific synaptic plasticity mechanisms are involved.  LTP and LTD result in long lasting increase or decrease of synaptic response respectively which is often measured electrophysiologically.  However, a limitation to our model is the difficulty to perform electrophysiological recordings.  Therefore, alternative approaches must be used to provide further information if LTP and LTD mechanisms are important mechanisms in our model.  Experiments utilizing time lapse imaging techniques can potentially be used to examine the turnover of AMPAR at the synapse [248].  Moreover, previous investigations have shown that changes in synaptic plasticity can change the local excitability at dendrites, which is possible to observe using calcium imaging [249].  Future experiments observing AMPAR trafficking calcium imaging at synapses could help us further our understanding of the potential involvement of LTP and LTD mechanisms in this model.  112 5.3. Activity Inhibition by TTX Reduces the Axon Growth Rate of Inhibited Neurons.  Activity Inhibition by GABAR Activation (Muscimol) Increased the Axon Growth Rate of Uninhibited Neurons  Our experimental results lead us to hypothesize that axon growth is an important part of our in vitro model.  This is also observed in vivo as many genes currently associated with ending the critical period are related to cell growth [30].   To complement our previous experiments, we utilized time lapse imaging to examine involvement of axon growth.  In our in vitro model, the axon growth rate was slower for the ‘presynaptic’ group with reduced activity relative to its opposing ‘presynaptic’ counterpart.  Interestingly, TTX treatment reduced the axon growth rate for the inhibited neurons while muscimol treatment resulted in increased axon growth rate of the untreated ‘presynaptic’ neurons.  Our findings coincide with a recent study, using ON/OFF bipolar cells in the retina, that showed changes from reducing activity between cells was a result of reduced synapse formation rather than synapse elimination [250].  These observations provided additional support for our previous conclusion in which we determined that having fewer synapses associated with the inhibited ‘presynaptic’ neuronal group in the ‘postsynaptic’ compartment was a result of differences in axon growth between treated and untreated ‘presynaptic’ neuronal groups.  Synaptic connections are thought to be able to stabilize axonal and dendritic branches allowing for further growth [196, 241, 251].  Our results showed ‘presynaptic’ groups that formed more synapses with the common ‘postsynaptic’ group also had a greater rate of axon growth in that region further supported those initial studies.  Our results also provided support for the synaptotropic hypothesis which postulates that synaptic connections can stabilize exploratory axonal and dendritic branches allowing for further growth at those points [196].   113 Furthermore, according to the synaptotropic hypothesis, there should be more frequent turnover of axonal branches if successful synaptic connections were not established.  Therefore, in our in vitro model, we would expect a greater turnover of axonal branches and filopodia associated with the activity inhibited ‘presynaptic’ neuronal group.  However, our experiments were unable to address the branch dynamics that occurred in our dual input setting.  Branch life times are very fast and to accurately estimate their life span high frequency imaging is required (since a branch must appear in at least 2 consecutive images) [241] which our chosen 30 minute intervals was not.  Therefore, additional experiments with a higher rate of image capture will further our understanding of branch motility during under influences of afferents with different levels of activity. 5.4. The Postsynaptic Age is a Greater Limiting Factor for Our In Vitro Critical Period  As neurons develop, the composition of the extracellular matrix and receptors at the synapse changes over time [187-189].  Some of these developmental changes are thought to be responsible for ending the critical period [41].  Recent studies are beginning to show the importance of adult neurogenesis and the potential therapies can be derived from it [252-254]. Therefore, how new neurons integrate themselves and the effect they have in a mature system have become an important area of study [190-192].  In our activity dependent model, regardless of the chemical treatment we chose, the effects of our model were no longer occurred by DIV 28. Furthermore, both chemical treatments we used produced changes in synaptic distribution around DIV 14.  Using our model, we determined if replacing either the pre- or postsynaptic neurons (at DIV 28) with younger neurons would reintroduce the synaptic effects in our model caused by changes in input activity level.  By culturing the different ages of the pre- and post- synaptic neurons (a two week difference) together, we show that our model was dependent on the age of  114 the postsynaptic neurons when GABA agonist was used for activity inhibition.  Interestingly, the reverse situation with younger ‘presynaptic’ neurons showed no reintroduction of the synaptic shifts.   However, when TTX inhibition was used in the mixed age experiments no reintroduction occurred for either age situation. The postsynaptic importance of age in critical periods has also been reported in previous studies. Critical periods in PC-CF systems are observed to be postsynaptic age dependent [156].  Recent studies have shown that newly born neurons in adult systems experience a critical period [194], as measured by the ability to induce long term potentiation (LTP).  One future direction using heterochronic cultures in three compartment devices might be to determine whether younger neurons can shift the synaptic balance in a developing neural circuit. By varying the in vitro ages between ‘presynaptic’ neuronal groups, we would be able to determine if younger neurons would have an advantage or disadvantage is gaining synaptic territory in a dual input environment.  In addition, using our current established methodology for in vitro dual input changes, we would be able to determine whether additional changes in neuronal activity levels would further amplify or remove any observable effects.  Furthermore, results once again differed depending on the how activity inhibition was induced. Further investigations into this phenomenon would further increase our understanding of activity and synaptogenesis. 5.5. TTX versus Muscimol Inhibition  In vivo, it has been shown that several distinct mechanisms can be involved in ocular dominance shifts in the visual cortex during the critical period.  Previous investigations the mechanisms of ocular dominance shifts have found evidence for multiple mechanisms.  Not only has it been  115 observed that different layers of the visual cortex undergo LTD with differing mechanisms [72] but there is also evidence for the involvement of homeostatic mechanisms [73, 74].  Therefore, it is possible that multiple mechanisms exist in vitro also.  In our studies, we used two forms of chemical treatment, TTX and muscimol, to reduce neural activity of select ‘presynaptic’ groups.  TTX blocks sodium ion channels to prevent the propagation of action potentials.  This results in limiting the communication between pre-and postsynaptic terminals (after initial contact) to spontaneous presynaptic vesicle release [255]. Muscimol acts as a GABAR agonist to further hyperpolarize the neuron making it much more difficult for an axon potential to be generated [195].  Interestingly, these two types of chemical inhibition lead to different molecular and cellular pathways in our model.  Replacing TTX inhibition with a combination of AMPAR and NMDAR antagonism [172] would help determine whether limiting postsynaptic neurotransmission to spontaneous vesicle release is different from hyperpolarization of presynaptic neurons.  ‘Presynaptic’ group AMPAR/NMDAR antagonism might result in similar neurotransmission properties to TTX inhibition.  Furthermore, investigations previous to ours showed neurons transfected with an inward rectifier potassium channel, Kir2.1 formed fewer synapses compared to existing wild type neurons [119].  Kir2.1 reduces neuronal activity by further hyperpolarizing the neuron making the excitation threshold for an action potential to occur more difficult to reach.  However, the investigators of that study only looked at the postsynaptic effect of Kir2.1 on neuronal synaptogenesis.  Later investigations did use Kir2.1 neurons to monitor the effects on axon growth and motility [222], however, they observed a reduction in axonal arbor length. This could be due to model differences as their observations were obtained using retinal ganglion cells of the zebrafish.   116 Future experiments looking at Kir2.1 transfected neurons as presynaptic afferents might provide further information on synaptogenesis involving ‘presynaptic’ neurons with reduced activity resulting from hyperpolarization.  These additional experiments will help determine if the differences in mechanism between TTX and muscimol inhibition arise from differences in neurotransmission between synapses or possible secondary effects of the applied drugs.  To further evaluate the differences between TTX and muscimol inhibition in our model, we can use fluorescent indicators for calcium or sodium.  Another limitation of our model is that we are looking at how a small subset of incoming axons with reduced activity interacts in an environment abundant with additional axons. Therefore, the overall activity in the target ‘postsynaptic’ neurons may not vary significantly.  Previous studies have shown that it is possible to observe local calcium transients at synaptic sites [27].  Future experiments using calcium indicator dyes could look at the postsynaptic transients that occur at synapses and determine whether our chemical treatments lead to differences in synapse activation.  Therefore, we can observe activity at specific synaptic sites in the postsynaptic compartment which we can distinguish between inputs originating from the inhibited or normal ‘presynaptic’ group. Presynaptically, high speed fluorescent imaging using the Na+ indicator, sodium-binding benzofuran isophthalate (SBFI) [28, 29] can quantify sodium dynamics along the axon [30] enabling us to image axon potentials by observing the Na+ dynamics [30].   This would allow further characterization of differences between our two methods of neural activity inhibition. 5.6. Future Directions and Applications  Similar to in vivo studies of visual cortex plasticity, our model and experimental methods currently focus on reducing the activity of an incoming ‘presynaptic’ input.  In vivo (in the visual cortex), it is much easier to close one eye (reducing patterned input) rather than continuously  117 stimulate it without affecting the other eye.  This makes it difficult to create an experimental situation in which one afferent pathway’s activity is changed by sustained increased stimulation. However, it would be relatively easy in our model to change one compartment to an environment with increased neuronal activity relative to normal without affecting the other two groups. Therefore, follow up experiments with our model could consist of examining the synaptic consequences when one ‘presynaptic’ neuronal group has increased neural activity (above basal levels).  This could be done using treatments with high extracellular potassium K+ concentrations or GABA antagonists, for example.  Furthermore, we do not have to focus solely on spontaneous activity.  While induction of LTP and LTD through electrophysiological methods in our model is difficult, several methods exist for induction of these processes chemically [168-170, 256].  Using these induction protocols, we can determine if ‘presynaptic’ cells undergoing LTP or LTD can achieve greater success in regards to synapse formation.  Previous investigations have looked at LTP and LTD as a postsynaptic consequence of stimulation [131, 257, 258].   However, there is little information available about the plastic consequences when a group of ‘presynaptic’ neurons undergoes LTP or LTD while an opposing group of afferents does not.  This type of study will also provide us with additional information on the role synaptic plasticity in multi-input situations.  Much effort has gone into examining mechanisms that can extend or reintroduce the critical period time window later in life.  Recent investigations have hypothesized the potential for ‘molecular brakes’; molecules whose expression would signal the end of the critical period [240].  To identify these molecules, investigators have turned to the use of transcriptome analysis [240, 259].   Information from this methodology has the potential to produce several candidate  118 genes that may play essential roles in developmental critical periods.  Furthermore, the resulting proteins from these genes are also potential targets.  These identified proteins have the potential to be developed in novel pharmacological therapies. Often the biological activities of proteins can be replicated by shorter peptides taken from the primary sequence [260].  Recently, peptide arrays have become valuable tools in finding active peptide sequences that can potentially be used for future pharmacological applications. However, the success of an array can generate a multitude of candidates and biological verification of all identified molecules can quickly become quite arduous.  A disadvantage for in vivo methods is the time takes to screen potential genes or peptide candidates.  Our model would allow for fast initial screens allowing for only the most effective candidates to be used in vivo.  For example, recent investigations have identified Lynx1 as an important molecule to halt the critical period in the visual cortex [240].  Lynx1 has direct interactions with nicotinic acetylcholine receptors (nAChR) [261].  Using a peptide array, amino acid sequences can be identified that are responsible for the Lynx1 – nAChR interaction which can then be used to interfere with this interaction.  If several candidates are discovered the time and resources to screen each one individually (and in combination) in vivo may be too great.  Our model would allow initial screens of the candidate peptides to be quickly screened allowing only the ones with the highest potential to be used in the more arduous in vivo experiments. 5.7. Concluding Remarks  In conclusion, we established a novel in vitro model for activity dependent synaptic plasticity using a three compartment microfluidic device.  The three compartment device allowed us to create a dual input situation where the activity of the inputs can be altered chemically.  Upon  119 further investigation using our model, we observed that it had some mechanistic similarities to the in vivo systems such as the visual cortex.  Our model also had NMDA dependent mechanisms and an observed critical period for synaptic plasticity.  Interestingly, variations in the time frame and mechanisms of the critical period that existed in our model differed between the chemical inhibition methods we used.  Furthermore, neural activity inhibition affected the relative axon growth rate between opposing afferents but once again whether the rate was increased or decreased depended on the type of inhibition.  Moreover, the age of the postsynaptic neurons was an important limiting factor for our model under inhibition by GABAR activation. We concluded from our model that activity changes from one group of afferents can influence the synaptic behavior of an opposing set of afferents when they share a common target.  Future experiments performed using this model not only will provide insight into the synaptic influence opposing afferents have towards each other but may also provide potential pharmacological applications in vivo.  In addition, we will be able to easily visualize the cellular and molecular dynamics that occur in such a setting.  Furthermore, our model can be used as an efficient screen for genes and novel drugs aimed at prolonging or reintroducing critical periods allowing for better selection for future application in vivo.  While not a replacement for current in vivo and in vitro models of activity dependent synaptic plasticity, our newly established model can act as an additional important complement to current and future studies of neural circuit formation and synaptogenesis.  120 References  1. Katz, L.C. and C.J. Shatz, Synaptic activity and the construction of cortical circuits. Science, 1996. 274(5290): p. 1133-8. 2. Hua, J.Y. and S.J. Smith, Neural activity and the dynamics of central nervous system development. Nat Neurosci, 2004. 7(4): p. 327-32. 3. Hensch, T.K., Critical Period Regulation. Annual Review of Neuroscience, 2004. 27(1): p. 549-579. 4. Berardi, N., T. Pizzorusso, and L. Maffei, Critical periods during sensory development. Curr Opin Neurobiol, 2000. 10(1): p. 138-45. 5. Lichtman, J. and H. Colman, Synapse Elimination and Indelible Memory Neuron, 2000. 25(2): p. 269-78. 6. Goda, Y. and G.W. Davis, Mechanisms of synapse assembly and disassembly. Neuron, 2003. 40(2): p. 243-64. 7. Rao, A. and A.M. Craig, Activity regulates the synaptic localization of the NMDA receptor in hippocampal neurons. Neuron, 1997. 19(4): p. 801-12. 8. Gomperts, S.N., et al., Distinct roles for ionotropic and metabotropic glutamate receptors in the maturation of excitatory synapses. J Neurosci, 2000. 20(6): p. 2229-37. 9. Cottrell, J.R., et al., Distribution, density, and clustering of functional glutamate receptors before and after synaptogenesis in hippocampal neurons. J Neurophysiol, 2000. 84(3): p. 1573-87. 10. Mariani, J. and J.P. Changeux, Ontogenesis of olivocerebellar relationships. I. Studies by intracellular recordings of the multiple innervation of Purkinje cells by climbing fibers in the developing rat cerebellum. J Neurosci, 1981. 1(7): p. 696-702. 11. Hashimoto, K. and M. Kano, Postnatal development and synapse elimination of climbing fiber to Purkinje cell projection in the cerebellum. Neuroscience Research, 2005. 53(3): p. 221-228. 12. Adesnik, H. and M. Scanziani, Lateral competition for cortical space by layer-specific horizontal circuits. Nature, 2010. 464(7292): p. 1155-60. 13. Lowel, S. and W. Singer, Selection of intrinsic horizontal connections in the visual cortex by correlated neuronal activity. Science, 1992. 255(5041): p. 209-12. 14. Bavelier, D., et al., Removing brakes on adult brain plasticity: from molecular to behavioral interventions. J Neurosci, 2010. 30(45): p. 14964-71. 15. Lewis, T.L. and D. Maurer, Effects of early pattern deprivation on visual development. Optom Vis Sci, 2009. 86(6): p. 640-6. 16. Newport EL, Bavelier D, and N. HJN, Critical thinking about critical period: perspectives on a critical period for language acquisition, in Language, brain and cognitive development: essays in honor of Jacques Mehler, E. Dupoux, Editor. 2001, The MIT Press: Boston. p. 481–502. 17. Hubel, D.H. and T.N. Wiesel, The period of susceptibility to the physiological effects of unilateral eye closure in kittens. J Physiol, 1970. 206(2): p. 419-36. 18. Hubener, M., Mouse visual cortex. Curr Opin Neurobiol, 2003. 13(4): p. 413-20. 19. Hensch, T.K., Critical period plasticity in local cortical circuits. Nature Reviews Neuroscience, 2005. 6(11): p. 877-888. 20. Horton, J.C. and D.L. Adams, The cortical column: a structure without a function. Philos Trans R Soc Lond B Biol Sci, 2005. 360(1456): p. 837-62. 21. Shatz, C.J. and M.P. Stryker, Ocular dominance in layer IV of the cat's visual cortex and the effects of monocular deprivation. J Physiol, 1978. 281: p. 267-83.  121 22. Wiesel, T.N. and D.H. Hubel, Single cell responses in striate cortex of kittens deprived of vision in one eye. J. Neurophysiol, 1963. 26: p. 1003-1017. 23. Hubel, D.H., T.N. Wiesel, and S. LeVay, Plasticity of Ocular Dominance Columns in Monkey Striate Cortex. Philosophical Transactions of the Royal Society B: Biological Sciences, 1977. 278(961): p. 377-409. 24. Paoletti, P. and J. Neyton, NMDA receptor subunits: function and pharmacology. Curr Opin Pharmacol, 2007. 7(1): p. 39-47. 25. Kohr, G., NMDA receptor function: subunit composition versus spatial distribution. Cell Tissue Res, 2006. 326(2): p. 439-46. 26. Carroll, R.C. and R.S. Zukin, NMDA-receptor trafficking and targeting: implications for synaptic transmission and plasticity. Trends Neurosci, 2002. 25(11): p. 571-7. 27. Bear, M.F., et al., Disruption of experience-dependent synaptic modifications in striate cortex by infusion of an NMDA receptor antagonist. J Neurosci, 1990. 10(3): p. 909-25. 28. Roberts, E.B., M.A. Meredith, and A.S. Ramoa, Suppression of NMDA receptor function using antisense DNA block ocular dominance plasticity while preserving visual responses. J Neurophysiol, 1998. 80(3): p. 1021-32. 29. Daw, N.W., et al., Injection of MK-801 affects ocular dominance shifts more than visual activity. J Neurophysiol, 1999. 81(1): p. 204-15. 30. Roberts, E.B. and A.S. Ramoa, Enhanced NR2A subunit expression and decreased NMDA receptor decay time at the onset of ocular dominance plasticity in the ferret. J Neurophysiol, 1999. 81(5): p. 2587-91. 31. Sheng, M., et al., Changing subunit composition of heteromeric NMDA receptors during development of rat cortex. Nature, 1994. 368(6467): p. 144-7. 32. Vicini, S., et al., Functional and pharmacological differences between recombinant N- methyl-D-aspartate receptors. J Neurophysiol, 1998. 79(2): p. 555-66. 33. Erreger, K., et al., Subunit-specific gating controls rat NR1/NR2A and NR1/NR2B NMDA channel kinetics and synaptic signalling profiles. J Physiol, 2005. 563(Pt 2): p. 345-58. 34. Sobczyk, A., V. Scheuss, and K. Svoboda, NMDA receptor subunit-dependent [Ca2+] signaling in individual hippocampal dendritic spines. J Neurosci, 2005. 25(26): p. 6037- 46. 35. Gambrill, A.C. and A. Barria, NMDA receptor subunit composition controls synaptogenesis and synapse stabilization. Proc Natl Acad Sci U S A, 2011. 108(14): p. 5855-60. 36. Carmignoto, G. and S. Vicini, Activity-dependent decrease in NMDA receptor responses during development of the visual cortex. Science, 1992. 258(5084): p. 1007-11. 37. Monyer, H., et al., Heteromeric NMDA receptors: molecular and functional distinction of subtypes. Science, 1992. 256(5060): p. 1217-21. 38. Monyer, H., et al., Developmental and regional expression in the rat brain and functional properties of four NMDA receptors. Neuron, 1994. 12(3): p. 529-40. 39. Quinlan, E.M., D.H. Olstein, and M.F. Bear, Bidirectional, experience-dependent regulation of N-methyl-D-aspartate receptor subunit composition in the rat visual cortex during postnatal development. Proc Natl Acad Sci U S A, 1999. 96(22): p. 12876-80. 40. Cao, Z., et al., Postnatal development of NR1, NR2A and NR2B immunoreactivity in the visual cortex of the rat. Brain Res, 2000. 859(1): p. 26-37. 41. Erisir, A. and J.L. Harris, Decline of the critical period of visual plasticity is concurrent with the reduction of NR2B subunit of the synaptic NMDA receptor in layer 4. J Neurosci, 2003. 23(12): p. 5208-18.  122 42. Jiang, B., M. Trevino, and A. Kirkwood, Sequential development of long-term potentiation and depression in different layers of the mouse visual cortex. J Neurosci, 2007. 27(36): p. 9648-52. 43. Philpot, B.D., K.K. Cho, and M.F. Bear, Obligatory role of NR2A for metaplasticity in visual cortex. Neuron, 2007. 53(4): p. 495-502. 44. Cho, K.K., et al., The ratio of NR2A/B NMDA receptor subunits determines the qualities of ocular dominance plasticity in visual cortex. Proc Natl Acad Sci U S A, 2009. 106(13): p. 5377-82. 45. Liu, L., et al., Role of NMDA receptor subtypes in governing the direction of hippocampal synaptic plasticity. Science, 2004. 304(5673): p. 1021-4. 46. Massey, P.V., et al., Differential roles of NR2A and NR2B-containing NMDA receptors in cortical long-term potentiation and long-term depression. J Neurosci, 2004. 24(36): p. 7821-8. 47. Gambrill, A.C., G.P. Storey, and A. Barria, Dynamic Regulation of NMDA Receptor Transmission. Journal of Neurophysiology, 2010. 105(1): p. 162-171. 48. Bliss TV, Collingridge GL, and M. RG, Introduction. Long-term potentiation and structure of the issue. Philosophical Transactions of the Royal Society B: Biological Sciences, 2003. 358(1432): p. 607-611. 49. Lu, B. and A. Figurov, Role of neurotrophins in synapse development and plasticity. Rev Neurosci, 1997. 8(1): p. 1-12. 50. Pang, P.T. and B. Lu, Regulation of late-phase LTP and long-term memory in normal and aging hippocampus: role of secreted proteins tPA and BDNF. Ageing Res Rev, 2004. 3(4): p. 407-30. 51. Bramham, C.R., Local protein synthesis, actin dynamics, and LTP consolidation. Curr Opin Neurobiol, 2008. 18(5): p. 524-31. 52. Maffei, L., et al., Nerve growth factor (NGF) prevents the shift in ocular dominance distribution of visual cortical neurons in monocularly deprived rats. J Neurosci, 1992. 12(12): p. 4651-62. 53. Pizzorusso, T., et al., Role of neurotrophins in the development and plasticity of the visual system: experiments on dark rearing. Int J Psychophysiol, 2000. 35(2-3): p. 189-96. 54. Berardi, N. and L. Maffei, From visual experience to visual function: roles of neurotrophins. J Neurobiol, 1999. 41(1): p. 119-26. 55. Galuske, R.A., et al., Brain-derived neurotrophic factor reversed experience-dependent synaptic modifications in kitten visual cortex. Eur J Neurosci, 1996. 8(7): p. 1554-9. 56. Abidin, I., et al., Impaired GABAergic inhibition in the visual cortex of brain-derived neurotrophic factor heterozygous knockout mice. J Physiol, 2008. 586(7): p. 1885-901. 57. Lichtman, J.W. and H. Colman, Synapse elimination and indelible memory. Neuron, 2000. 25(2): p. 269-78. 58. Lechner, H.A. and J.H. Byrne, New perspectives on classical conditioning: a synthesis of Hebbian and non-Hebbian mechanisms. Neuron, 1998. 20(3): p. 355-8. 59. Turrigiano, G.G. and S.B. Nelson, Hebb and homeostasis in neuronal plasticity. Curr Opin Neurobiol, 2000. 10(3): p. 358-64. 60. Hata, Y. and M.P. Stryker, Control of thalamocortical afferent rearrangement by postsynaptic activity in developing visual cortex. Science, 1994. 265(5179): p. 1732-5. 61. Malenka, R.C. and M.F. Bear, LTP and LTDAn Embarrassment of Riches. Neuron, 2004. 44(1): p. 5-21. 62. Thiels, E., G. Barrionuevo, and T.W. Berger, Excitatory stimulation during postsynaptic inhibition induces long-term depression in hippocampus in vivo. J Neurophysiol, 1994. 72(6): p. 3009-16.  123 63. Raymond, C., LTP forms 1, 2 and 3: different mechanisms for the ‘long’ in long-term potentiation. Trends in Neurosciences, 2007. 30(4): p. 167-175. 64. Smith, G.B., A.J. Heynen, and M.F. Bear, Bidirectional synaptic mechanisms of ocular dominance plasticity in visual cortex. Philos Trans R Soc Lond B Biol Sci, 2009. 364(1515): p. 357-67. 65. Gordon, J.A., et al., Deficient plasticity in the primary visual cortex of alpha- calcium/calmodulin-dependent protein kinase II mutant mice. Neuron, 1996. 17(3): p. 491-9. 66. Taha, S., et al., Autophosphorylation of alphaCaMKII is required for ocular dominance plasticity. Neuron, 2002. 36(3): p. 483-91. 67. Hooks, B.M. and C. Chen, Critical periods in the visual system: changing views for a model of experience-dependent plasticity. Neuron, 2007. 56(2): p. 312-26. 68. Xiong, W., et al., Anisomycin activates p38 MAP kinase to induce LTD in mouse primary visual cortex. Brain Res, 2006. 1085(1): p. 68-76. 69. Heynen, A.J., et al., Molecular mechanism for loss of visual cortical responsiveness following brief monocular deprivation. Nat Neurosci, 2003. 6(8): p. 854-62. 70. Yoon, B.J., et al., Essential role for a long-term depression mechanism in ocular dominance plasticity. Proc Natl Acad Sci U S A, 2009. 106(24): p. 9860-5. 71. Daw, N., et al., LTP and LTD vary with layer in rodent visual cortex. Vision Res, 2004. 44(28): p. 3377-80. 72. Crozier, R.A., et al., Deprivation-induced synaptic depression by distinct mechanisms in different layers of mouse visual cortex. Proceedings of the National Academy of Sciences, 2007. 104(4): p. 1383-1388. 73. Tropea, D., A. Van Wart, and M. Sur, Molecular mechanisms of experience-dependent plasticity in visual cortex. Philos Trans R Soc Lond B Biol Sci, 2009. 364(1515): p. 341- 55. 74. Kaneko, M., et al., Tumor necrosis factor-alpha mediates one component of competitive, experience-dependent plasticity in developing visual cortex. Neuron, 2008. 58(5): p. 673- 80. 75. Turrigiano, G.G., et al., Activity-dependent scaling of quantal amplitude in neocortical neurons. Nature, 1998. 391(6670): p. 892-6. 76. Anggono, V., R.L. Clem, and R.L. Huganir, PICK1 loss of function occludes homeostatic synaptic scaling. J Neurosci, 2011. 31(6): p. 2188-96. 77. Turrigiano, G.G. and S.B. Nelson, Homeostatic plasticity in the developing nervous system. Nat Rev Neurosci, 2004. 5(2): p. 97-107. 78. Desai, N.S., et al., Critical periods for experience-dependent synaptic scaling in visual cortex. Nat Neurosci, 2002. 5(8): p. 783-9. 79. Waites, C.L., A.M. Craig, and C.C. Garner, Mechanisms of Vertebrate Synaptogenesis. Annu. Rev. Neurosci, 2005. 25: p. 251–74. 80. Meyer, M.P. and S.J. Smith, Evidence from in vivo imaging that synaptogenesis guides the growth and branching of axonal arbors by two distinct mechanisms. J Neurosci, 2006. 26(13): p. 3604-14. 81. Washbourne, P., J.E. Bennett, and A.K. McAllister, Rapid recruitment of NMDA receptor transport packets to nascent synapses. Nat Neurosci, 2002. 5(8): p. 751-9. 82. Sabo, S.L., R.A. Gomes, and A.K. McAllister, Formation of Presynaptic Terminals at Predefined Sites along Axons. Journal of Neuroscience, 2006. 26(42): p. 10813-10825. 83. Gerrow, K., et al., A Preformed Complex of Postsynaptic Proteins Is Involved in Excitatory Synapse Development. Neuron, 2006. 49(4): p. 547-562.  124 84. Ziv, N.E. and S.J. Smith, Evidence for a role of dendritic filopodia in synaptogenesis and spine formation. Neuron, 1996. 17(1): p. 91-102. 85. Ahmari, S.E., J. Buchanan, and S.J. Smith, Assembly of presynaptic active zones from cytoplasmic transport packets. Nat Neurosci, 2000. 3(5): p. 445-51. 86. Dalva, M.B., A.C. McClelland, and M.S. Kayser, Cell adhesion molecules: signalling functions at the synapse. Nat Rev Neurosci, 2007. 8(3): p. 206-20. 87. Scheiffele, P., et al., Neuroligin expressed in nonneuronal cells triggers presynaptic development in contacting axons. Cell, 2000. 101(6): p. 657-69. 88. Graf, E.R., et al., Neurexins induce differentiation of GABA and glutamate postsynaptic specializations via neuroligins. Cell, 2004. 119(7): p. 1013-26. 89. Chih, B., H. Engelman, and P. Scheiffele, Control of excitatory and inhibitory synapse formation by neuroligins. Science, 2005. 307(5713): p. 1324-8. 90. Varoqueaux, F., S. Jamain, and N. Brose, Neuroligin 2 is exclusively localized to inhibitory synapses. Eur J Cell Biol, 2004. 83(9): p. 449-56. 91. Prange, O., et al., A balance between excitatory and inhibitory synapses is controlled by PSD-95 and neuroligin. Proc Natl Acad Sci U S A, 2004. 101(38): p. 13915-20. 92. Levinson, J.N., et al., Neuroligins mediate excitatory and inhibitory synapse formation: involvement of PSD-95 and neurexin-1beta in neuroligin-induced synaptic specificity. J Biol Chem, 2005. 280(17): p. 17312-9. 93. McAllister, A.K., Dynamic Aspects of CNS Synapse Formation. Annual Review of Neuroscience, 2007. 30(1): p. 425-450. 94. Craig, A., E. Graf, and M. Linhoff, How to build a central synapse: clues from cell culture. Trends in Neurosciences, 2006. 29(1): p. 8-20. 95. Friedman, H.V., et al., Assembly of new individual excitatory synapses: time course and temporal order of synaptic molecule recruitment. Neuron, 2000. 27(1): p. 57-69. 96. Okabe, S., A. Miwa, and H. Okado, Spine formation and correlated assembly of presynaptic and postsynaptic molecules. J Neurosci, 2001. 21(16): p. 6105-14. 97. Fletcher, T.L., et al., The distribution of synapsin I and synaptophysin in hippocampal neurons developing in culture. J Neurosci, 1991. 11(6): p. 1617-26. 98. Fletcher, T.L., P. De Camilli, and G. Banker, Synaptogenesis in hippocampal cultures: evidence indicating that axons and dendrites become competent to form synapses at different stages of neuronal development. J Neurosci, 1994. 14(11 Pt 1): p. 6695-706. 99. Bamji, S.X., et al., BDNF mobilizes synaptic vesicles and enhances synapse formation by disrupting cadherin-beta-catenin interactions. J Cell Biol, 2006. 174(2): p. 289-99. 100. Bamji, S.X., et al., Role of beta-catenin in synaptic vesicle localization and presynaptic assembly. Neuron, 2003. 40(4): p. 719-31. 101. Ahmari, S.E. and S.J. Smith, Knowing a nascent synapse when you see it. Neuron, 2002. 34(3): p. 333-6. 102. Rao, A., E.M. Cha, and A.M. Craig, Mismatched appositions of presynaptic and postsynaptic components in isolated hippocampal neurons. J Neurosci, 2000. 20(22): p. 8344-53. 103. Holtmaat, A. and K. Svoboda, Experience-dependent structural synaptic plasticity in the mammalian brain. Nat Rev Neurosci, 2009. 10(9): p. 647-58. 104. Zhai, R.G., et al., Assembling the presynaptic active zone: a characterization of an active one precursor vesicle. Neuron, 2001. 29(1): p. 131-43. 105. Krueger, S.R., A. Kolar, and R.M. Fitzsimonds, The presynaptic release apparatus is functional in the absence of dendritic contact and highly mobile within isolated axons. Neuron, 2003. 40(5): p. 945-57.  125 106. Jin, Y. and C.C. Garner, Molecular Mechanisms of Presynaptic Differentiation. Annual Review of Cell and Developmental Biology, 2008. 24(1): p. 237-262. 107. Ruthazer, E.S., Stabilization of Axon Branch Dynamics by Synaptic Maturation. Journal of Neuroscience, 2006. 26(13): p. 3594-3603. 108. Kennedy, M.B., The postsynaptic density at glutamatergic synapses. Trends Neurosci, 1997. 20(6): p. 264-8. 109. Bresler, T., et al., Postsynaptic density assembly is fundamentally different from presynaptic active zone assembly. J Neurosci, 2004. 24(6): p. 1507-20. 110. Rao, A., et al., Heterogeneity in the molecular composition of excitatory postsynaptic sites during development of hippocampal neurons in culture. J Neurosci, 1998. 18(4): p. 1217-29. 111. Kim, E. and M. Sheng, PDZ domain proteins of synapses. Nat Rev Neurosci, 2004. 5(10): p. 771-81. 112. Bresler, T., et al., The dynamics of SAP90/PSD-95 recruitment to new synaptic junctions. Mol Cell Neurosci, 2001. 18(2): p. 149-67. 113. Prange, O. and T.H. Murphy, Modular transport of postsynaptic density-95 clusters and association with stable spine precursors during early development of cortical neurons. J Neurosci, 2001. 21(23): p. 9325-33. 114. Wenthold, R.J., et al., Trafficking of NMDA receptors. Annu Rev Pharmacol Toxicol, 2003. 43: p. 335-58. 115. Liao, D., N.A. Hessler, and R. Malinow, Activation of postsynaptically silent synapses during pairing-induced LTP in CA1 region of hippocampal slice. Nature, 1995. 375(6530): p. 400-4. 116. Isaac, J.T., R.A. Nicoll, and R.C. Malenka, Evidence for silent synapses: implications for the expression of LTP. Neuron, 1995. 15(2): p. 427-34. 117. Kerchner, G.A. and R.A. Nicoll, Silent synapses and the emergence of a postsynaptic mechanism for LTP. Nat Rev Neurosci, 2008. 9(11): p. 813-25. 118. Ashby, M.C. and J.T. Isaac, Maturation of a recurrent excitatory neocortical circuit by experience-dependent unsilencing of newly formed dendritic spines. Neuron, 2011. 70(3): p. 510-21. 119. Burrone, J., M. O'Byrne, and V. Murthy, Multiple forms of synaptic plasticity triggered by selective suppression of activity in individual neurons. Nature, 2002. 420(6914): p. 414-418. 120. Adesnik, H., et al., NMDA receptors inhibit synapse unsilencing during brain development. Proceedings of the National Academy of Sciences, 2008. 105(14): p. 5597- 5602. 121. Hall, B.J. and A. Ghosh, Regulation of AMPA receptor recruitment at developing synapses. Trends Neurosci, 2008. 31(2): p. 82-9. 122. Liao, D., et al., Regulation of morphological postsynaptic silent synapses in developing hippocampal neurons. Nat Neurosci, 1999. 2(1): p. 37-43. 123. Buchs, P.A. and D. Muller, Induction of long-term potentiation is associated with major ultrastructural changes of activated synapses. Proc Natl Acad Sci U S A, 1996. 93(15): p. 8040-5. 124. Kopec, C.D., et al., Glutamate receptor exocytosis and spine enlargement during chemically induced long-term potentiation. J Neurosci, 2006. 26(7): p. 2000-9. 125. De Roo, M., P. Klauser, and D. Muller, LTP promotes a selective long-term stabilization and clustering of dendritic spines. PLoS Biol, 2008. 6(9): p. e219. 126. Kopec, C.D., et al., GluR1 links structural and functional plasticity at excitatory synapses. J Neurosci, 2007. 27(50): p. 13706-18.  126 127. Matsuzaki, M., Factors critical for the plasticity of dendritic spines and memory storage. Neurosci Res, 2007. 57(1): p. 1-9. 128. Nakamura, Y., et al., PICK1 inhibition of the Arp2/3 complex controls dendritic spine size and synaptic plasticity. EMBO J, 2011. 30(4): p. 719-30. 129. Hsieh, H., et al., AMPAR removal underlies Abeta-induced synaptic depression and dendritic spine loss. Neuron, 2006. 52(5): p. 831-43. 130. Ripley, B., et al., Regulation of synaptic stability by AMPA receptor reverse signaling. Proc Natl Acad Sci U S A, 2011. 108(1): p. 367-72. 131. Bastrikova, N., et al., Synapse elimination accompanies functional plasticity in hippocampal neurons. Proc Natl Acad Sci U S A, 2008. 105(8): p. 3123-7. 132. Stevens, B., et al., The classical complement cascade mediates CNS synapse elimination. Cell, 2007. 131(6): p. 1164-78. 133. Reiter, H.O., D.M. Waitzman, and M.P. Stryker, Cortical activity blockade prevents ocular dominance plasticity in the kitten visual cortex. Exp Brain Res, 1986. 65(1): p. 182-8. 134. Stryker, M.P. and W.A. Harris, Binocular impulse blockade prevents the formation of ocular dominance columns in cat visual cortex. J Neurosci, 1986. 6(8): p. 2117-33. 135. Li, J., A. Erisir, and H. Cline, In vivo time-lapse imaging and serial section electron microscopy reveal developmental synaptic rearrangements. Neuron, 2011. 69(2): p. 273- 86. 136. Walsh, M.K. and J.W. Lichtman, In vivo time-lapse imaging of synaptic takeover associated with naturally occurring synapse elimination. Neuron, 2003. 37(1): p. 67-73. 137. Fu, M. and Y. Zuo, Experience-dependent structural plasticity in the cortex. Trends Neurosci, 2011. 34(4): p. 177-87. 138. Busetto, G., et al., Hebbian mechanisms revealed by electrical stimulation at developing rat neuromuscular junctions. J Neurosci, 2000. 20(2): p. 685-95. 139. Duxson, M.J., The effect of postsynaptic block on development of the neuromuscular junction in postnatal rats. J Neurocytol, 1982. 11(3): p. 395-408. 140. Buffelli, M., et al., Genetic evidence that relative synaptic efficacy biases the outcome of synaptic competition. Nature, 2003. 424(6947): p. 430-4. 141. Balice-Gordon, R.J. and J.W. Lichtman, Long-term synapse loss induced by focal blockade of postsynaptic receptors. Nature, 1994. 372(6506): p. 519-24. 142. Ichise, T., et al., mGluR1 in cerebellar Purkinje cells essential for long-term depression, synapse elimination, and motor coordination. Science, 2000. 288(5472): p. 1832-5. 143. Glynn, M.W., et al., MHCI negatively regulates synapse density during the establishment of cortical connections. Nat Neurosci, 2011. 14(4): p. 442-51. 144. Sanes, J. and J. Lichtman, Development of the vertebrate neuromuscular junction. Annu Rev Neurosci. , 1999. 22: p. 389-442. 145. An, M.C., et al., Acetylcholine negatively regulates development of the neuromuscular junction through distinct cellular mechanisms. Proc Natl Acad Sci U S A, 2010. 107(23): p. 10702-7. 146. Jansen, J.K. and T. Fladby, The perinatal reorganization of the innervation of skeletal muscle in mammals. Prog Neurobiol, 1990. 34(1): p. 39-90. 147. Favero, M., et al., Synapse formation and elimination: role of activity studied in different models of adult muscle reinnervation. J Neurosci Res, 2007. 85(12): p. 2610-9. 148. Balice-Gordon, R.J., et al., Gradual loss of synaptic cartels precedes axon withdrawal at developing neuromuscular junctions. Neuron, 1993. 11(5): p. 801-15. 149. Gan, W., Synaptic Segregation at the Developing Neuromuscular Junction. Science, 1998. 282(5393): p. 1508-1511.  127 150. Patton, B.L., Basal lamina and the organization of neuromuscular synapses. J Neurocytol, 2003. 32(5-8): p. 883-903. 151. Favero, M., et al., On the mechanism of action of muscle fibre activity in synapse competition and elimination at the mammalian neuromuscular junction. Eur J Neurosci, 2009. 29(12): p. 2327-34. 152. Bosman, L.W.J., et al., Homosynaptic Long-Term Synaptic Potentiation of the "Winner" Climbing Fiber Synapse in Developing Purkinje Cells. Journal of Neuroscience, 2008. 28(4): p. 798-807. 153. Kano, M., et al., Impaired synapse elimination during cerebellar development in PKC gamma mutant mice. Cell, 1995. 83(7): p. 1223-31. 154. Bosman, L.W. and A. Konnerth, Activity-dependent plasticity of developing climbing fiber-Purkinje cell synapses. Neuroscience, 2009. 162(3): p. 612-23. 155. Hashimoto, K. and M. Kano, Postnatal development and synapse elimination of climbing fiber to Purkinje cell projection in the cerebellum. Neurosci Res, 2005. 53(3): p. 221-8. 156. Letellier, M., et al., Synapse elimination in olivo-cerebellar explants occurs during a critical period and leaves an indelible trace in Purkinje cells. Proceedings of the National Academy of Sciences, 2009. 157. Chedotal, A., E. Bloch-Gallego, and C. Sotelo, The embryonic cerebellum contains topographic cues that guide developing inferior olivary axons. Development, 1997. 124(4): p. 861-70. 158. Harms, K.J., Synapse-Specific Regulation of AMPA Receptor Subunit Composition by Activity. Journal of Neuroscience, 2005. 25(27): p. 6379-6388. 159. Tarsa, L., Synaptophysin regulates activity-dependent synapse formation in cultured hippocampal neurons. Proceedings of the National Academy of Sciences, 2002. 99(2): p. 1012-1016. 160. Campenot, R.B., Local control of neurite development by nerve growth factor. Proc Natl Acad Sci U S A, 1977. 74(10): p. 4516-9. 161. Park, J., et al., Microfluidic compartmentalized co-culture platform for CNS axon myelination research. Biomed Microdevices, 2009. 11(6): p. 1145-53. 162. Park, J.W., et al., Microfluidic culture platform for neuroscience research. Nature Protocols, 2006. 1(4): p. 2128-2136. 163. Filler, A.G., et al., Tri-partite complex for axonal transport drug delivery achieves pharmacological effect. BMC Neurosci, 2010. 11: p. 8. 164. Taylor, A.M., et al., A microfluidic culture platform for CNS axonal injury, regeneration and transport. Nat Methods, 2005. 2(8): p. 599-605. 165. Gordon, J.A.S., M.P., Experience-dependent plasticity of binocular responses in the primary visual cortex of the mouse. J. Neurosci., 1996. 16: p. 3274-3286. 166. Zeitelhofer, M., et al., Transfection of cultured primary neurons via nucleofection. Curr Protoc Neurosci, 2009. Chapter 4: p. Unit4 32. 167. Gartner, A., L. Collin, and G. Lalli, Nucleofection of primary neurons. Methods Enzymol, 2006. 406: p. 374-88. 168. Lee, H.K., et al., NMDA induces long-term synaptic depression and dephosphorylation of the GluR1 subunit of AMPA receptors in hippocampus. Neuron, 1998. 21(5): p. 1151-62. 169. Otmakhov, N., et al., Forskolin-induced LTP in the CA1 hippocampal region is NMDA receptor dependent. J Neurophysiol, 2004. 91(5): p. 1955-62. 170. Lu, W., et al., Activation of synaptic NMDA receptors induces membrane insertion of new AMPA receptors and LTP in cultured hippocampal neurons. Neuron, 2001. 29(1): p. 243-54.  128 171. Whitesides, G.M., The origins and the future of microfluidics. Nature, 2006. 442(7101): p. 368-73. 172. Uesaka, N., et al., Activity dependence of cortical axon branch formation: a morphological and electrophysiological study using organotypic slice cultures. J Neurosci, 2005. 25(1): p. 1-9. 173. El-Husseini, A.E., et al., PSD-95 involvement in maturation of excitatory synapses. Science, 2000. 290(5495): p. 1364-8. 174. Craig AM, B.C., Huganir RL, Banker G., The distribution of glutamate receptors in cultured rat hippocampal neurons: postsynaptic clustering of AMPA-selective subunits. Neuron, 1993. 10(6): p. 1055-1068. 175. Ben-Ari Y, G.J., Tyzio R, Khazipov R., GABA: a pioneer transmitter that excites immature neurons and generates primitive oscillations. Physiol Rev., 2007. 87(4): p. 1215-84. 176. Ganguly K, S.A., Wong ST, Poo M., GABA itself promotes the developmental switch of neuronal GABAergic responses from excitation to inhibition. Cell, 2001. 105(4): p. 521- 32. 177. Ganguly, K., et al., GABA itself promotes the developmental switch of neuronal GABAergic responses from excitation to inhibition. Cell, 2001. 105(4): p. 521-32. 178. Ben-Ari, Y., et al., GABA: a pioneer transmitter that excites immature neurons and generates primitive oscillations. Physiol Rev, 2007. 87(4): p. 1215-84. 179. Chen, G., P.Q. Trombley, and A.N. van den Pol, GABA receptors precede glutamate receptors in hypothalamic development; differential regulation by astrocytes. J Neurophysiol, 1995. 74(4): p. 1473-84. 180. Swanwick, C.C., et al., Development of gamma-aminobutyric acidergic synapses in cultured hippocampal neurons. J Comp Neurol, 2006. 495(5): p. 497-510. 181. Pizzorusso, T., Reactivation of Ocular Dominance Plasticity in the Adult Visual Cortex. Science, 2002. 298(5596): p. 1248-1251. 182. McGee, A.W., et al., Experience-Driven Plasticity of Visual Cortex Limited by Myelin and Nogo Receptor. Science, 2005. 309(5744): p. 2222-2226. 183. Mathews, G.C., Selective changes in hippocampal GABAA receptors during status epilepticus. Epilepsy Curr, 2008. 8(6): p. 170-2. 184. Goodkin, H.P., et al., Subunit-specific trafficking of GABA(A) receptors during status epilepticus. J Neurosci, 2008. 28(10): p. 2527-38. 185. Calkin, P.A. and E.M. Barnes, Jr., gamma-Aminobutyric acid-A (GABAA) agonists down- regulate GABAA/benzodiazepine receptor polypeptides from the surface of chick cortical neurons. J Biol Chem, 1994. 269(2): p. 1548-53. 186. Tehrani, M.H. and E.M. Barnes, Jr., Agonist-dependent internalization of gamma- aminobutyric acidA/benzodiazepine receptors in chick cortical neurons. J Neurochem, 1991. 57(4): p. 1307-12. 187. Groc, L., et al., NMDA Receptor Surface Trafficking and Synaptic Subunit Composition Are Developmentally Regulated by the Extracellular Matrix Protein Reelin. Journal of Neuroscience, 2007. 27(38): p. 10165-10175. 188. Liu, X.B., Switching of NMDA Receptor 2A and 2B Subunits at Thalamic and Cortical Synapses during Early Postnatal Development. Journal of Neuroscience, 2004. 24(40): p. 8885-8895. 189. Sinagra, M., Reelin, Very-Low-Density Lipoprotein Receptor, and Apolipoprotein E Receptor 2 Control Somatic NMDA Receptor Composition during Hippocampal Maturation In Vitro. Journal of Neuroscience, 2005. 25(26): p. 6127-6136.  129 190. Ge, S., et al., Synaptic integration and plasticity of new neurons in the adult hippocampus. The Journal of Physiology, 2008. 586(16): p. 3759-3765. 191. Kolb, B., et al., Growth factor-stimulated generation of new cortical tissue and functional recovery after stroke damage to the motor cortex of rats. Journal of Cerebral Blood Flow &#38; Metabolism, 2006. 192. Reynolds, B. and W. S, Generation of neurons and astrocytes from isolated cells of the adult mammalian central nervous system. Science, 1992. 255(5052): p. 1707-1710. 193. Ge, S., et al., GABA regulates synaptic integration of newly generated neurons in the adult brain. Nature, 2006. 439(7076): p. 589-93. 194. Ge, S., et al., A Critical Period for Enhanced Synaptic Plasticity in Newly Generated Neurons of the Adult Brain. Neuron, 2007. 54(4): p. 559-566. 195. Lema, G.M.C., Modes and Models of GABAA Receptor Gating. The Journal of Physiology, 2006. 196. Vaughn, J.E., R.P. Barber, and T.J. Sims, Dendritic development and preferential growth into synaptogenic fields: a quantitative study of Golgi-impregnated spinal motor neurons. Synapse, 1988. 2(1): p. 69-78. 197. Vaughn, J.E., C.K. Henrikson, and J.A. Grieshaber, A quantitative study of synapses on motor neuron dendritic growth cones in developing mouse spinal cord. J Cell Biol, 1974. 60(3): p. 664-72. 198. Cull-Candy, S.G., Role of Distinct NMDA Receptor Subtypes at Central Synapses. Science's STKE, 2004. 2004(255): p. re16-re16. 199. Liao, D., R.H. Scannevin, and R. Huganir, Activation of silent synapses by rapid activity- dependent synaptic recruitment of AMPA receptors. J Neurosci, 2001. 21(16): p. 6008- 17. 200. Star, E.N., D.J. Kwiatkowski, and V.N. Murthy, Rapid turnover of actin in dendritic spines and its regulation by activity. Nat Neurosci, 2002. 5(3): p. 239-46. 201. Fukazawa, Y., et al., Hippocampal LTP is accompanied by enhanced F-actin content within the dendritic spine that is essential for late LTP maintenance in vivo. Neuron, 2003. 38(3): p. 447-60. 202. Malinow, R. and R.C. Malenka, AMPA receptor trafficking and synaptic plasticity. Annu Rev Neurosci, 2002. 25: p. 103-26. 203. Steward, O. and E.M. Schuman, Protein synthesis at synaptic sites on dendrites. Annu Rev Neurosci, 2001. 24: p. 299-325. 204. Merrill, M., et al., Activity-driven postsynaptic translocation of CaMKII. Trends in Pharmacological Sciences, 2005. 26(12): p. 645-653. 205. Strack S, C.S., Lovinger DM, Colbran RJ., Translocation of Autophosphorylated Calcium/Calmodulin-dependent Protein Kinase II to the Postsynaptic Density. J Biol Chem, 1997. 272(21): p. 13467-70. 206. Appleby, V.J., et al., LTP in hippocampal neurons is associated with a CaMKII-mediated increase in GluA1 surface expression. Journal of Neurochemistry, 2011. 116(4): p. 530- 543. 207. Hayashi, Y., Driving AMPA Receptors into Synapses by LTP and CaMKII: Requirement for GluR1 and PDZ Domain Interaction. Science, 2000. 287(5461): p. 2262-2267. 208. Lu W, I.K., Roche KW, Nicoll RA., Synaptic targeting of AMPA receptors is regulated by a CaMKII site in the first intracellular loop of GluA1. Proc Natl Acad Sci, 2010. 107(51): p. 22266-71. 209. Wenthold RJ, P.R., Blahos J II, Niedzielski AS., Evidence for multiple AMPA receptor complexes in hippocampal CA1/CA2 neurons. J Neurosci, 1996. 16(6): p. 1982-1989.  130 210. Groc, L., B. Gustafsson, and E. Hanse, AMPA signalling in nascent glutamatergic synapses: there and not there! Trends Neurosci, 2006. 29(3): p. 132-9. 211. Passafaro M, N.T., Sala C, Sheng M., Induction of dendritic spines by an extracellular domain of AMPA receptor subunit GluR2. Nature, 2003. 424(6949): p. 677-681. 212. Saglietti, L., et al., Extracellular Interactions between GluR2 and N-Cadherin in Spine Regulation. Neuron, 2007. 54(3): p. 461-477. 213. Ripley, B., et al., Regulation of synaptic stability by AMPA receptor reverse signaling. Proceedings of the National Academy of Sciences, 2010. 108(1): p. 367-372. 214. Chen L, C.N., Mower GD., Developmental changes in the expression of NMDA receptor subunits (NR1, NR2A, NR2B) in the cat visual cortex and the effects of dark rearing. Brain Res Mol Brain Res., 2000. 78((1-2)): p. 196-200. 215. Fink, C., Molecular mechanisms of CaMKII activation in neuronal plasticity. Current Opinion in Neurobiology, 2002. 12(3): p. 293-299. 216. Bayer, K.U., et al., Interaction with the NMDA receptor locks CaMKII in an active conformation. Nature, 2001. 411(6839): p. 801-5. 217. Ahmadian G, J.W., Liu L, Wyszynski M, Lee SH, Dunah AW, Taghibiglou C, Wang Y, Lu J, Wong TP, Sheng M, Wang YT., Tyrosine phosphorylation of GluR2 is required for insulin-stimulated AMPA receptor endocytosis and LTD. EMBO J., 2004. 23(5): p. 1040- 50. 218. Wang, Y., -Amino-3-hydroxy-5-methylisoxazole-4-propionic Acid Subtype Glutamate Receptor (AMPAR) Endocytosis Is Essential for N-Methyl-D-aspartate-induced Neuronal Apoptosis. Journal of Biological Chemistry, 2004. 279(40): p. 41267-41270. 219. Brebner, K., Nucleus Accumbens Long-Term Depression and the Expression of Behavioral Sensitization. Science, 2005. 310(5752): p. 1340-1343. 220. Rajan, I., S. Witte, and H.T. Cline, NMDA receptor activity stabilizes presynaptic retinotectal axons and postsynaptic optic tectal cell dendrites in vivo. J Neurobiol, 1999. 38(3): p. 357-68. 221. O'Rourke, N.A., H.T. Cline, and S.E. Fraser, Rapid remodeling of retinal arbors in the tectum with and without blockade of synaptic transmission. Neuron, 1994. 12(4): p. 921- 34. 222. Hua, J.Y., et al., Regulation of axon growth in vivo by activity-based competition. Nature, 2005. 434(7036): p. 1022-6. 223. Sawtell, N.B., et al., NMDA Receptor-Dependent Ocular Dominance Plasticity in Adult Visual Cortex. Neuron, 2003. 38(6): p. 977-985. 224. Lu HC, Gonzalez E, and C. MC., Barrel cortex critical period plasticity is independent of changes in NMDA receptor subunit composition. Neuron, 2001. 34(4): p. 619-34. 225. Bear, M.F. and R.C. Malenka, Synaptic plasticity: LTP and LTD. Curr Opin Neurobiol, 1994. 4(3): p. 389-99. 226. Yashiro, K. and B.D. Philpot, Regulation of NMDA receptor subunit expression and its implications for LTD, LTP, and metaplasticity. Neuropharmacology, 2008. 55(7): p. 1081-94. 227. Lisman J, S.H., Cline H., The molecular basis of CaMKII function in synaptic and behavioural memory. Nat Rev Neurosci., 2002. 3(3): p. 175-90. 228. Derkach, V., A. Barria, and T.R. Soderling, Ca2+/calmodulin-kinase II enhances channel conductance of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionate type glutamate receptors. Proc. Natl Acad. Sci. , 1999. 96: p. 3269-3274. 229. Rongo, C. and J.M. Kaplan, CaMKII regulates the density of central glutamatergic synapses in vivo. Nature, 1999. 402(6758): p. 195-9.  131 230. Kopec, C.D., et al., GluR1 Links Structural and Functional Plasticity at Excitatory Synapses. Journal of Neuroscience, 2007. 27(50): p. 13706-13718. 231. Fortin, D.A., et al., Long-term potentiation-dependent spine enlargement requires synaptic Ca2+-permeable AMPA receptors recruited by CaM-kinase I. J Neurosci, 2010. 30(35): p. 11565-75. 232. Hanse, E., et al., Glutamate synapse in developing brain: an integrative perspective beyond the silent state. Trends Neurosci, 2009. 32(10): p. 532-7. 233. Rittenhouse, C.D., et al., Monocular deprivation induces homosynaptic long-term depression in visual cortex. Nature, 1999. 397(6717): p. 347-50. 234. Zhu, Y., et al., Rap2-JNK removes synaptic AMPA receptors during depotentiation. Neuron, 2005. 46(6): p. 905-16. 235. Fu, Z., et al., Differential roles of Rap1 and Rap2 small GTPases in neurite retraction and synapse elimination in hippocampal spiny neurons. Journal of Neurochemistry, 2007. 100(1): p. 118-131. 236. Kirkwood, A., M.C. Rioult, and M.F. Bear, Experience-dependent modification of synaptic plasticity in visual cortex. Nature, 1996. 381(6582): p. 526-8. 237. Barria, A. and R. Malinow, NMDA Receptor Subunit Composition Controls Synaptic Plasticity by Regulating Binding to CaMKII. Neuron, 2005. 48(2): p. 289-301. 238. Peineau, S., et al., A systematic investigation of the protein kinases involved in NMDA receptor-dependent LTD: evidence for a role of GSK-3 but not other serine/threonine kinases. Mol Brain, 2009. 2: p. 22. 239. Gainey, M.A., et al., Synaptic Scaling Requires the GluR2 Subunit of the AMPA Receptor. Journal of Neuroscience, 2009. 29(20): p. 6479-6489. 240. Morishita, H., et al., Lynx1, a Cholinergic Brake, Limits Plasticity in Adult Visual Cortex. Science, 2010. 330(6008): p. 1238-1240. 241. Cline, H. and K. Haas, The regulation of dendritic arbor development and plasticity by glutamatergic synaptic input: a review of the synaptotrophic hypothesis. J Physiol, 2008. 586(6): p. 1509-17. 242. Cooke, S.F. and M.F. Bear, Visual experience induces long-term potentiation in the primary visual cortex. J Neurosci, 2010. 30(48): p. 16304-13. 243. Taha, S.A. and M.P. Stryker, Ocular dominance plasticity is stably maintained in the absence of alpha calcium calmodulin kinase II (alphaCaMKII) autophosphorylation. Proc Natl Acad Sci U S A, 2005. 102(45): p. 16438-42. 244. Harvey, C.D. and K. Svoboda, Locally dynamic synaptic learning rules in pyramidal neuron dendrites. Nature, 2007. 450(7173): p. 1195-200. 245. Matsuzaki, M., et al., Structural basis of long-term potentiation in single dendritic spines. Nature, 2004. 429(6993): p. 761-6. 246. Matsuzaki, M., et al., Dendritic spine geometry is critical for AMPA receptor expression in hippocampal CA1 pyramidal neurons. Nat Neurosci, 2001. 4(11): p. 1086-92. 247. Lee, S.J., et al., Activation of CaMKII in single dendritic spines during long-term potentiation. Nature, 2009. 458(7236): p. 299-304. 248. Perestenko, P.V. and J.M. Henley, Visualization of AMPAR Trafficking and Surface Expression. 2006. 249. Frick, A., J. Magee, and D. Johnston, LTP is accompanied by an enhanced local excitability of pyramidal neuron dendrites. Nat Neurosci, 2004. 7(2): p. 126-35. 250. Kerschensteiner, D., et al., Neurotransmission selectively regulates synapse formation in parallel circuits in vivo. Nature, 2009. 460(7258): p. 1016-20. 251. Ruthazer, E.S., Control of Axon Branch Dynamics by Correlated Activity in Vivo. Science, 2003. 301(5629): p. 66-70.  132 252. Kolb, B., et al., Growth factor-stimulated generation of new cortical tissue and functional recovery after stroke damage to the motor cortex of rats. J Cereb Blood Flow Metab, 2007. 27(5): p. 983-97. 253. Ge, S., et al., A critical period for enhanced synaptic plasticity in newly generated neurons of the adult brain. Neuron, 2007. 54(4): p. 559-66. 254. Toni, N., et al., Neurons born in the adult dentate gyrus form functional synapses with target cells. Nat Neurosci, 2008. 11(8): p. 901-7. 255. Fredj, N.B. and J. Burrone, A resting pool of vesicles is responsible for spontaneous vesicle fusion at the synapse. Nat Neurosci, 2009. 12(6): p. 751-8. 256. Aniksztejn, L. and Y. Ben-Ari, Novel form of long-term potentiation produced by a K+ channel blocker in the hippocampus. Nature, 1991. 349(6304): p. 67-9. 257. Bourne, J.N. and K.M. Harris, Coordination of size and number of excitatory and inhibitory synapses results in a balanced structural plasticity along mature hippocampal CA1 dendrites during LTP. Hippocampus, 2011. 21(4): p. 354-73. 258. Tominaga-Yoshino, K., et al., Repetitive induction of late-phase LTP produces long- lasting synaptic enhancement accompanied by synaptogenesis in cultured hippocampal slices. Hippocampus, 2008. 18(3): p. 281-93. 259. Plessy, C., et al., A resource for transcriptomic analysis in the mouse brain. PLoS One, 2008. 3(8): p. e3012. 260. Min, D.H. and M. Mrksich, Peptide arrays: towards routine implementation. Curr Opin Chem Biol, 2004. 8(5): p. 554-8. 261. Ibanez-Tallon, I., et al., Novel modulation of neuronal nicotinic acetylcholine receptors by association with the endogenous prototoxin lynx1. Neuron, 2002. 33(6): p. 893-903.     133 Appendix  A. Calibration Curves  Appendix A 1 : Concentration curves for fluorescein and crystal violet used for leak tests of microfluidic devices     y = 4.019x + 0.0145 R² = 0.9992 0 0.5 1 1.5 2 2.5 0 0.1 0.2 0.3 0.4 0.5 0.6 Ab so rb an ce  48 0 Concentration gradient (dilutions) of Fluorescein (1 = 100µM) y = 0.5396x ‐ 0.0037 R² = 0.9982 ‐0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.2 0.4 0.6 0.8 1 1.2 Ab so rb an ce  59 0 Concentration gradient (dilutions) of crytal violet (1 = 100µM)

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072583/manifest

Comment

Related Items