• No results found

Biochemical regulation of the CaMKII/GluN2B interaction and its role in the maintenance of synaptic strength

N/A
N/A
Protected

Academic year: 2021

Share "Biochemical regulation of the CaMKII/GluN2B interaction and its role in the maintenance of synaptic strength"

Copied!
161
0
0

Loading.... (view fulltext now)

Full text

(1)

BIOCHEMICAL REGULATION OF THE CAMKII/GLUN2B INTERACTION AND ITS ROLE IN THE MAINTENANCE OF SYNAPTIC STRENGTH

by

KELSEY BARCOMB

B.A. Neuroscience, Hamilton College 2009

A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment

of the requirements for the degree of Doctor of Philosophy

Pharmacology Program 2015

(2)

ii This thesis for the Doctor of Philsophy degree by

Kelsey Barcomb has been approved for the

Pharmacology Program by

Timothy Benke, Chair Richard Allen Mark Dell’Acqua

Paula Hoffman Matthew Kennedy K. Ulrich Bayer, Advisor

(3)

iii Barcomb, Kelsey (Ph.D., Pharmacology)

The Biochemical Regulation of CaMKII/GluN2B Binding and its Role in Maintaining Synaptic Strength

Thesis directed by Associate Professor K. Ulrich Bayer ABSTRACT

Ca2+/Calmodulin (CaM)-dependent protein kinase II (CaMKII) is a major regulator of long-term potentiation (LTP) in the hippocampus. During this process, CaMKII becomes activated and translocates to the activated synapse through an interaction with the GluN2B subunit of the NMDAR. This interaction persists after the initial stimulus has subsided and has therefore been proposed to act as a form of molecular memory. For this reason, the mechanisms that underlie the interaction and the physiological functions related to its induction and persistence are of great interest.

It was first shown here that the interaction is independent of enzymatic activity, but requires occupation of the ATP-binding pocket, adding to the model of CaMKII/GluN2B binding. This model is similar in many ways to the model of ischemia-related CaMKII clustering and the dependence on enzymatic activity was also determined for that interaction. Like GluN2B binding, clustering was independent of enzymatic activity.

Performing mutational analysis of CaMKII regulatory sites led to another interesting finding. A constitutively autonomous, nonstimulatable CaMKII mutant did not interact with GluN2B basally within cells, despite the fact that it did in vitro. Further, this mutant translocated to GluN2B upon Ca2+ stimulation, even though it is not itself able to bind CaM. These results indicate that Ca2+ regulates the CaMKII/GluN2B interaction through a mechanism that is in addition to its effects on CaMKII. Mechanistically, it was

(4)

iv suggested that this novel regulation by Ca2+/CaM is mediated by either phosphatase activity or a direct CaM/GluN2B interaction.

The function of CaMKII/GluN2B binding was also examined in the context of the maintenance of synaptic strength. It was shown that an inhibitor of the interaction reduced synaptic strength, but did so to a lesser extent in a knockin mouse that is incompetent for CaMKII/GluN2B binding. Interestingly, this effect appears to be dependent on structural rather than enzymatic effects of the kinase, suggesting that the CaMKII/GluN2B interaction acts to scaffold protein complexes at the PSD.

Taken together, the results of these experiments indicate that CaMKII/GluN2B binding is independent of enzymatic activity, has a multifaceted regulation by CaM, and at least partially regulates synaptic strength through structural functions.

The form and content of this abstract are approved. I recommend its publication. Approved: K. Ullrich Bayer

(5)

v

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my mentor, Ulli Bayer, for all of his help; Ulli was truly the best mentor I could have hoped for. The Bayer lab in general was a joy to work in and provided many wonderful labmates during my time in graduate school. In particular, Steve Coultrap was always extremely helpful and kind. My fellow student, Dayton Goodell, is also a great scientist and the frequent discussions I had with made me a better researcher. Heather Caballes trained me my first two years in grade school and always did so with patience and kindness. Vincent Zaegel provided great assistance in making constructs very quickly, significantly increasing the number of experiments I was able to do.

In addition to the superior lab environment, I enjoyed a larger community of scientists and friends in the Department of Pharmacology. Our local synapse group of the Bayer, Benke, Dell’Acqua, and Kennedy labs was always incredibly helpful. Every member of that group was a joy to work with, but in particular Ron Freund provided training on electrophysiological techniques and Brian Hiester and Kevin Woolfrey were always available for discussion and guidance. Tim Benke, Mark Dell’Acqua, and Matt Kennedy also served on my thesis committee, along with Paula Hoffman and Rich Allen, creating a committee that was always very helpful and encouraging. The Pharmacology Program Graduate Training Committee was additionally always very supportive of my graduate career, recognizing my accomplishments and funding my attendance at meetings. I would also like to acknowledge Nancy Zahniser, who was an inspiration and provided me real guidance in my first two years of graduate school.

Outside (and inside) of the lab I also had a really great group of friends who made my life in Denver really great. In particular, Jacki Rorabaugh and Anna Castano

(6)

vi were top-notch friends and confidants. My labmates, Steve Coultrap and Dayton Goodell, also were truly excellent friends and helped me out a lot outside of the lab.

Finally, my family has always been incredibly supportive of my career goals. I love them all very much and thank them for helping me follow my dreams.

(7)

vii TABLE OF CONTENTS

CHAPTER

I. INTRODUCTION ... 1

 

THE HIPPOCAMPUS, SYNAPTIC PLASTICITY, AND LEARNING AND MEMORY ... 1

 

NMDA RECEPTOR COMPOSITION ... 3

 

CALCIUM/CALMODULIN-DEPENDENT PROTEIN KINASE II ... 4

 

REGULATION OF SYNAPTIC PLASTICITY BY CAMKII ... 9

 

CAMKII AND THE NMDAR ... 18

 

STRUCTURAL ROLE OF CAMKII ... 28

 

OBJECTIVES ... 29

 

II. ENZYMATIC ACTIVITY OF CAMKII IS NOT REQUIRED FOR ITS INTERACTION WITH THE GLUTAMATE RECEPTOR SUBUNIT GLUN2B ... 30

 

INTRODUCTION ... 30

 

RESULTS ... 31

 

DISCUSSION ... 43

 

III. LIVE IMAGING OF ENDOGENOUS CAMKII IN NEURONS REVEALS THAT ISCHEMIA-RELATED AGGREGATION DOES NOT REQUIRE KINASE ACTIVITY ... 49

 

RESULTS ... 50

 

DISCUSSION ... 55

 

IV. NOVEL REGULATORY EFFECT OF CALMODULIN IN CAMKII/GLUN2B BINDING ... 59

 

INTRODUCTION ... 59

 

RESULTS ... 59

 

(8)

viii V. ELECTROPHYSIOLOGICAL EFFECTS OF TATCN21

ARE PARTIALLY MEDIATED BY CAMKII/GLUN2B BINDING ... 82

 

INTRODUCTION ... 82

 

RESULTS ... 85

 

DISCUSSION ... 92

 

VI. CONCLUSIONS ... 97

 

BIOCHEMICAL REGULATION OF CAMKII/GLUN2B BINDING ... 97

 

ENZYMATIC ACTIVITY OF CAMKII IS NOT REQUIRED FOR ITS INTERACTION WITH THE GLUTAMATE RECEPTOR SUBUNIT GLUN2B ... 97

 

LIVE IMAGING OF ENDOGENOUS CAMKII IN NEURONS REVEALS THAT ISCHEMIA-RELATED AGGREGATION DOES NOT REQUIRE KINASE ACTIVITY ... 102

 

NOVEL REGULATORY EFFECT OF CALMODULIN IN CAMKII/GLUN2B BINDING ... 103

 

ELECTROPHYSIOLOGICAL EFFECTS OF TATCN21 ARE PARTIALLY MEDIATED BY CAMKII/GLUN2B BINDING ... 112

 

CONCLUSION ... 113

 

VII. METHODS ... 115

 

CALMODULIN OVERLAY ... 115

 

CAMKII/GLUN2B IN VITRO BINDING ... 115

 

CAMKII SELF-ASSOCIATION ... 116

 

CELL CULTURE ... 117

 

ELECTROPHYSIOLOGY ... 118

 

FLUORESCENT MICROSCOPY AND IMAGE ANALYSIS ... 119

 

IMMUNOCYTOCHEMISTRY ... 121

 

WESTERN BLOT ANALYSIS ... 122

(9)

LIST OF FIGURES FIGURE

1.1: CaMKII structure and regulation………... 1.2: CaMKII interactions with the NMDAR cytoplasmic tails………... 2.1: Inhibition of CaMKII enzymatic activity by H7 and Sta…………... 2.2: H7 and Sta mimic nucleotide enhancement of CaMKII/GluN2B…... 2.3: H7 and Sta inhibit CaMKII enzymatic activity within cells... 2.4: H7 and Sta permit translocation of GFP-CaMKII in heterologous cells……... 2.5: H7 and Sta permit synaptic translocation of

CaMKII in primary hippocampal neurons………... 2.6: Sta interferes with fixation/immunostaining in neurons……...…….... 3.1: KN93 and TatCN21/TatCN19o reduce CaMKII aggregation…...…... 3.2: H7 and Sta support CaMKII aggregation by mimicking nucleotide…... 3.3. CaMKII clustering is normal in the presence of the CaMKII intrabody... 3.4. Clustering in neurons is independent of enzymatic activity………...…... 3.5. Current mechanistic model of CaMKII clustering………... 4.1: Autonomous CaMKII interacts with GluN2B basally in vitro…………... 4.2: CaM has a novel regulatory role in CaMKII/GluN2B binding…………... 4.3: GluN2B-cSh is sufficient for translocation of CaM-independent CaMKII…... 4.4: The CaMKII/GluN2B interaction is not Zn2+-dependent………... 4.5: Experimental evidence for a novel CaM/GluN2B interaction………... 4.6: Prevention of P-S1303 induces basal colocalization………... 4.7: Translocation is unaffected by inhibitors of PP1, PP2A, and CaN…... 4.8: SG blocks CaMKII translocation to GluN2B-WT but not GluN2B S1303A….... 4.9: SG blocks CaMKII clustering in neurons………... 5.1: 20 M tatCN21 induces a persistent reduction in basal transmission…...

8 20 33 34 36 37 41 42 51 52 53 55 57 60 62 64 65 67 71 72 75 77 86

(10)

x 5.2: TatCN21 reduces LTP maintenance………...…………...

5.3: TatCN19o is not much more potent towards CaMKII/GluN2B binding……….. 5.4: TatCN21 requires stimulation to reduce synaptic strength……...…..…… 5.5: TatCN21 differentially affects WT and GluN2BCaMKII mice……….…... 5.6: TatCN21 induces a presynaptic reduction………..……... 6.1: Unanswered questions about CaMKII at the PSD………....……... 6.2: Conserved PP2C sequences………...…... 87 88 89 91 93 101 110

(11)

xi LIST OF TABLES

TABLE

(12)

xii LIST OF ABBREVIATIONS

AC-2/AC-3 Autocamtide-2/Autocamtide-3

ADP/ATP Adenosine Diphosphate/ Adenosine Triphosphate

AIP Autoinhibitory Peptide

AMPAR α-Amino-3-Hydroxy-5-Methyl-4-Isoxazolepropionic Acid Receptor

ANOVA Analysis of Variance

Ant Antennapaedia

C-Tail Cytoplasmic C-terminal Tail

CA1/CA3 Cornu Ammonis 1/Cornu Ammonis 3

CaM Calmodulin

CaMKII Ca2/CaM-Dependent Protein Kinase II

CaN Calcineurin (or PP2B)

CBS Calmodulin-Binding Site

Ctl Control

CN CaMKII-N Derived Peptide

CycA Cyclosporin-A

DIV Days in vitro

fEPSP Field Excitatory Postsynaptic Potential

FV Fiber Volley

GABAR γ-Amino-Butyric-Acid Receptor GFP Green Fluorescent Protein

GluA1 AMPA-Type Glutamate Receptor Subunit 1 GluN1 NMDA-Type Glutamate Receptor Subunit 1 GluN2A NMDA-Type Glutamate Receptor Subunit 2A GluN2B NMDA-Type Glutamate Receptor Subunit 2B GluN2BΔCaMKII GluN2B(L1298Q/R1300A); CaMKII-Binding Mutant

Glut Glutamate

Gly Glycine

GST Glutathione S-Transferase

H7 (±)-1-(5-Isoquinolinesulphonyl)-2-methylpiperazine IC50 Half Minimal Inhibitory Concentration

Iono Ionomycin Kd Dissociation Constant Ki Inhibitory Constant Km Michaelis Constant KI Knockin KO Knockout LTD Long-Term Depression LTP Long-Term Potentiation mCer mCerulean mCh mCherry

mGluR Metabotropic Glutamate Receptor NMDAR N-Methyl-D-Aspartate Receptor

OA Okadaic Acid

PDZ PSD95/Discs Large/ZO-1 Homology PKA cAMP Dependent Protein Kinase

PKC Protein Kinase C

(13)

xiii

PP2A Protein Phosphatase 2A

PP2B Protein Phosphatase 2B (or Calcineurin)

PP2C Protein Phosphatase 2C

PPF/PPR Paired-Pulse Facilitation/Paired-Pulse Ratio PPM Mg2+/Mn2+-Dependent Protein Phosphatases

PSD Postsynaptic Density

PSD95 Postsynaptic Density Protein 95

RMANOVA Repeated Measures Analysis of Variance S-Site Substrate Binding Site

SAP97 Synapse Associated Protein 97`

SG Sanguinarine

Sta Staurosporine

T-Site T286 Interacting Site

TPEN N,N,N’,N’-Tetrakis(2-pyridylmethyl)ethylenediamine tCaMKII Truncated CaMKII; Fully Active Kinase

TARP Transmembrane AMPAR Regulatory Protein

Vmax Maximum Rate

WT Wild Type

(14)

1 CHAPTER ONE

INTRODUCTION

THE HIPPOCAMPUS, SYNAPTIC PLASTICITY, AND LEARNING AND MEMORY The hippocampal formation is an area of the brain that has long been associated with regulating learning and memory (Douglas 1967). Though the exact manner in which the interactions that an organism has with the world become a memory encoded in the brain is still not well understood, it is generally agreed that it involves processes of synaptic plasticity (Baudry et al. 2014; Stuchlik 2014). Synaptic plasticity is most simply a group of cellular mechanisms by which the connections between neurons are modified in response to activity. While forms of plasticity take place throughout the nervous system – including in the spinal cord (Chirila et al. 2014) – plasticity has been most extensively studied in relation to learning and memory in the hippocampus. The two canonical long-term plasticity processes, which are also crucial in learning and memory, are long-term potentiation (LTP) and long-term depression (LTD), which strengthen and weaken synapses, respectively (Malenka & Bear 2004; Luscher & Malenka 2012). These processes are certainly not restricted to the hippocampus or the regulation of memory (Cerovic et al. 2013; Chirila et al. 2014; Hunt & Castillo 2012; Massey et al. 2004), and vice versa hippocampal function also relies on other forms of plasticity (Abraham & Bear 1996; Hunt & Castillo 2012), however LTP in the hippocampus will be the focus of the current discussion.

LTP was first discovered in the hippocampus in the 1970s by electrical stimulation of hippocampal slices (Bliss & Gardner-Medwin 1973; Bliss & Lomo 1973; Schwartzkroin & Wester 1975). While, LTP is commonly evoked electrophysiologically by high frequency stimulation, it can also be induced by chemical stimulation (Makhinson et al. 1999). Since the original measurements of hippocampal LTP, decades of research

(15)

2 have coalesced to a thorough understanding of the general molecular mechanisms that underlie potentiation. In its basic form, the model of LTP is that a high frequency stimulus activates N-methyl D-aspartate receptors (NMDARs) allowing Ca2+ into the postsynaptic neuron and this Ca2+ stimulus activates a kinase signaling cascade through Ca2+/CaM-dependent protein kinase II (CaMKII), which results in an enhancement in α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) function (for reviews see (Malenka & Nicoll 1999; Coultrap & Bayer 2012; Luscher & Malenka 2012)). Though this description is heavily simplified, it describes the basic functions of some of the most important players in the process – AMPARs, NMDARs, and CaMKII.

The AMPAR is a ligand gated ion channel, or in other terms an ionotropic glutamate receptor (Dingledine et al. 1999; Traynelis et al. 2010). During basal activity of the synapse, AMPARs are active and the number of AMPARs expressed at the cell surface, as well as their relative conductance properties, are important determinants of synaptic strength. During LTP, the AMPAR plays a fairly passive role in LTP, acting as a target of regulation, while both the NMDAR and CaMKII play active roles in initiating, and possibly maintaining, LTP (Coultrap & Bayer 2012; Sanhueza & Lisman 2013). NMDARs are also ionotropic glutamate receptors, but they are basally inactive due to a Mg2+ block (Cull-Candy et al. 2001). This block is removed when the cell becomes depolarized, allowing Mg2+ to act as a coincidence detector of simultaneous pre- and postsynaptic activation. LTP induction provides such activation, opening NMDARs, which are permeable to Ca2+. Ca2+ signaling is one of the most important regulators of cellular signaling and activates a number of enzymes, in particular kinases and phosphatases (Browning et al. 1985; Turner et al. 1982; Ghosh & Greenberg 1995). While the proteins that are activated by Ca2+ depend on its concentration and spike frequency, LTP specifically results in the activation of CaMKII, which becomes highly stimulated in

(16)

3 response to high Ca2+ spike frequencies (Burgoyne 2007; Li et al. 2012). The resulting effects of CaMKII activation will be discussed in detail below.

NMDA RECEPTOR COMPOSITION

NMDARs are tetramers consisting of two GluN1 subunits and two GluN2/3 subunits (for reviews see (Cull-Candy et al. 2001; Traynelis et al. 2010; Dingledine et al. 1999; Paoletti et al. 2013)). GluN1 is monogenic, but has multiple splice variants, whereas there are four GluN2 genes (GluN2A-D) and two GluN3 genes (GluN3A-B); the expression patterns of the GluN2/3 subunits are developmentally and regionally regulated (Sheng et al. 1994; Monyer et al. 1994). The GluN2A and GluN2B subunits predominate in the hippocampus (Fritschy et al. 1998). At least nine combinations of subunits have been described, all with two GluN1 subunits and either two identical or two different GluN2 or GluN3 subunits (Paoletti et al. 2013; Sheng et al. 1994). The resulting NMDAR compositions differ in channel properties, drug sensitivity, and cation sensitivity (Gielen et al. 2009; Mony et al. 2009; Siegler Retchless et al. 2012; Glasgow et al. 2015; Vicini et al. 1998). Furthermore, NMDAR subunits have been differentially associated with plasticity (Gray et al. 2011; Berberich et al. 2007; Liu et al. 2004; Brigman et al. 2010) and disease progression (Martel et al. 2009). Hippocampal GluN2A- and GluN2B-containing receptors in particular are thought to play different roles in synaptic plasticity. Selective blockage of GluN2B containing receptors with ifenprodil blocks LTP in hippocampal slices (Barria & Malinow 2005), though other studies have seen only a partial reduction in LTP with another GluN2B-containing receptor blocker, Ro25-981 (Bartlett et al. 2007; Foster et al. 2010). It should be noted that triheteromeric receptors, which may predominate in the adult hippocampus (Rauner & Köhr 2011; Al-Hallaq et al. 2007; Gray et al. 2011), are not as significantly inhibited by ifenprodil

(17)

4 (Hansen et al. 2014). Furthermore, overexpression of GluN2B in the mouse forebrain enhances potentiation and behavioral memory (Tang et al. 1999).

NMDAR subunits also differ significantly in their intracellular domains (Sanz-Clemente, Nicoll, et al. 2013b; Ryan et al. 2008), which themselves differentially regulate neuronal signaling, plasticity, and behavior (Sprengel et al. 1998; Ryan et al. 2012; Martel et al. 2012). The GluN2 subunits in particular are characterized by long cytoplasmic C-tails, unique among the family of glutamate receptors (Ryan et al. 2008). While GluN2A and GluN2B are 69% identical in their extracellular and transmembrane domains, they differ significantly intracellularly, as their C-tails are only 29% identical (Ryan et al. 2008). The differences in the tail domains lead to different protein-protein interactions (Al-Hallaq et al. 2007; Sans et al. 2000; Q. Chen et al. 2007) and post-translational modifications (B.-S. Chen & Roche 2007; B.-S. Chen et al. 2006), which in turn alter trafficking and synaptic expression (Prybylowski et al. 2005; Bard et al. 2010; Chung et al. 2004). Furthermore, it is thought that the interactions that the tail regions have with other proteins in the PSD regulate synaptic strength. In particular GluN2B has an important interaction with CaMKII, which is the focus of the current study.

CALCIUM/CALMODULIN-DEPENDENT PROTEIN KINASE II

CaMKII is a serine/threonine kinase that is part of a family of Ca2+/CaM-regulated protein kinases (for review see (Soderling et al. 2001; Hanson & Schulman 1992b; Coultrap et al. 2011)). CaMKII has broad substrate specificity (Schulman et al. 1985) – resulting in an extensive list of substrate proteins (Yoshimura et al. 2000; Yoshimura et al. 2002; White et al. 1998) – and primarily uses the substrate recognition sequence R-X-X-S/T (Payne et al. 1983; Pearson et al. 1985). There are four isoforms of CaMKII that exist in mammals - α, β, γ, and δ - which are encoded by different genes (Tobimatsu et al. 1988; Tobimatsu & Fujisawa 1989; Hudmon & Schulman 2002), and all isoforms

(18)

5 have multiple splice variants (Brocke et al. 1999; Tombes & Krystal 1997; Tombes et al. 2003). Though the isoforms are independently coded they share ~85% homology (Tobimatsu et al. 1988; Tobimatsu & Fujisawa 1989) and are similarly regulated biochemically (Edman & Schulman 1994). They do, however, differ in expression – whereas CaMKII γ and δ are expressed throughout the body (Tobimatsu & Fujisawa 1989; Schworer et al. 1993), CaMKII α and β are almost exclusively expressed in the brain (Erondu & Kennedy 1985; Tobimatsu & Fujisawa 1989). Within the brain, expression of both α and β are developmentally regulated and vary between regions (Sahyoun et al. 1985; Burgin et al. 1990). CaMKIIα is particularly abundant in the postnatal hippocampus, constituting as much as 1-2% of total protein in that structure (Erondu & Kennedy 1985).

CaMKII Structure

CaMKII exists as a dodecameric complex (Bennett et al. 1983; Goldenring et al. 1983) consisting of two stacked hexameric rings with resulting dimensions of about 20 nm in diameter and 10 nm in height (Woodgett et al. 1983; Coultrap et al. 2012; Chao et al. 2011; Kolodziej et al. 2000; Gaertner et al. 2004; Stratton et al. 2013) (Figure 1.1A). Dodecamers can have a multitude of subunit compositions – including α-homomers – but an apparent α to β ratio of approximately 2-4:1 in the hippocampus predicts holoenzymes that contain on average 9 α- and 3 β-subunits (Bennett et al. 1983; Brocke et al. 1999). Each kinase subunit has a C-terminal association domain, serving to link the subunits into a holoenzyme (Shen & T. Meyer 1998), and an N-terminal kinase domain (Figure 1.1B). Between these two major domains exist the regulatory domain and the variable linker region. This latter region accounts for most of the variability between CaMKII isoforms and splice variants (Bayer et al. 2002). The regulatory domain maintains the kinase in an inactive state basally through an

(19)

6 interaction between T286 (residue numbers throughout will refer to those on the α-isoform; in β, γ, and δ this residue is T287) and a hydrophobic pocket on the kinase domain in an area known as the T-site for its interaction with “T”-286 (Yang & Schulman 1999). The regulatory domain also contains the CaM binding site of the kinase; the minimal region for CaM binding is within residues 290-310 (Hanley et al. 1987; Payne et al. 1988; Putkey & Waxham 1996). This region, like most known CaM binding sites, is a basic amphiphilic α-helix (Bennett & Kennedy 1987).

CaMKII Regulation and Autonomous Activity

Binding of CaM to CaMKII causes a dissociation of the regulatory domain from the kinase domain, exposing the substrate-binding domain for substrate access (Figure 1.1C). When two neighboring subunits within a holoenzyme are bound by CaM, rapid autophosphorylation can occur at T286 by an intraholoenzyme and intersubunit interaction (Hanson et al. 1994; Rich & Schulman 1998). Because T286 participates in the basal interaction between the regulatory domain and the T-site of the kinase domain, its phosphorylation introduces a negative charge that occludes the interaction. Thus, T286 phosphorylation prevents reassociation of the regulatory domain with the kinase domain and maintains the kinase in an active state, even after Ca2+/CaM dissociation (Saitoh & Schwartz 1985; Lai et al. 1986; Miller & Kennedy 1986; Schworer et al. 1986; Lou et al. 1986; Fong et al. 1989; Hanson et al. 1989; Miller et al. 1988). This activity state – termed autonomy – produces activity that is far less than Ca2+/CaM-stimulated activity (~15-20% of maximum), though both stimulation and autonomy are required for the full activation of the kinase (Coultrap et al. 2010; Coultrap et al. 2012). Autonomy has an important role in the induction of LTP, which will be discussed further below.

Autophosphorylation of T286 has the further effect of changing CaMKII from one of the lowest affinity CaM-binding proteins to one of the highest (Kd ≈ 15 nM and ≤ 20 pM

(20)

7 respectively), a function known as CaM-trapping (T. Meyer et al. 1992). Because T286 autophosphorylation requires CaM-binding to neighboring subunits, the probability of autophosphorylation depends on Ca2+ concentration, allowing the kinase to effectively decode Ca2+ oscillation frequency (Hanson et al. 1994; T. Meyer et al. 1992; De Koninck & Schulman 1998); at low Ca2+ spike frequencies the probability of autophosphorylation is low and the binding rate of CaM is less than its off-rate (~1 sec-1), resulting in minimal activity within a holoenzyme, at high Ca2+ spike frequencies the likelihood of autophosphorylation is also high and the binding rate of CaM is greater than its off-rate (~0.00009 sec-1) (Putkey & Waxham 1996) and CaMKII is able to remain Ca2+ -stimulated and autophosphorylate. Both T286 autophosphorylation and Ca2+-frequency decoding are functions of CaMKII that are attributed to the holoenzyme nature of the kinase (Yamauchi et al. 1989; De Koninck & Schulman 1998). The ability of CaMKII to decode both the frequency and amplitude of postsynaptic Ca2 influx likely contributes to various forms of synaptic plasticity, including both LTP and LTD. Interestingly, both the probability of T286 autophosphorylation and the CaM activation constant vary between CaMKII isoforms/splice variants, as those parameters are functions of the composition of the variable linker region (Bayer et al. 2002; Gaertner et al. 2004). Differences in the length of the variable linker region likely determine the relative positioning of the kinase domains within the holoenzyme in relation to each other, thus affecting their ability to undergo intraholoenzyme autophosphorylation.

CaMKII autonomy can also be generated by other post-translational modifications on the regulatory domain (including nitrosylation (Coultrap & Bayer 2014; Coultrap, Zaegel, et al. 2014b), oxidation (Erickson et al. 2008), and O-linked glycosylation (Erickson et al. 2013)) and by GluN2B binding (Bayer et al. 2001). None of these autonomy-inducing modifications are as well understood as P-T286, however.

(21)

8 Figure 1.1: CaMKII structure and regulation

(A) CaMKII exists as a dodecameric holoenzyme with 12 subunits bound through their association domains in a central hub (yellow) with kinase domains (blue) radiating out. The predicted structure of this complex is a pair of stacked hexameric rings with a diameter of 20 nm and a height of 10 nm. (B) CaMKII has an N-terminal kinase domain (blue) and C-terminal association domain (yellow), which are connected by a regulatory domain (REG; green) and variable linker region (purple). Within the regulatory domain lies T286, the phosphorylation site that regulates autonomous activity, and the CaM-binding site (CBS, red). The CBS further includes T305/6, autophosphorylation sites that negatively regulate CaM binding. Numbers on the structure diagram refer to the amino acid sequence of CaMKIIα; “P” indicates a phosphorylation site. (C) Magnification of a kinase domain from a holoenzyme (shown from a “top-down” schematic view on the left) under basal conditions shows the block of the substrate binding domain (S-site; orange) by the regulatory domain, which binds to the T-site (white) through an interaction by T286. Ca2+-stimulation leads to activation of the kinase by Ca2+/CaM; CaM association with the CBS (red) relieves the interaction between the regulatory domain and the kinase domain, making the T- and S-sites accessible. When two neighboring kinase subunits are bound by CaM, autophosphorylation occurs by an intersubunit, intraholoenzyme reaction at T286. This phosphorylation maintains partial dissociation of the regulatory domain from the kinase domain and therefore imparts Ca2+/CaM-independent, autonomous activity. When CaM dissociates from the CBS, the kinase can undergo an additional autophosphorylation reaction T305/6, which prevents further CaM binding. It should be noted that although the diagram shows both sites of phosphorylation, they are not necessarily present at the same time – the kinase can be phosphorylated at T286 alone or at all three sites. P-T306 has also been observed without concomitant P-T286, though the reaction is much slower and it is unclear to what extent it occurs in vivo.

CaMKII has additional autophosphorylation sites that are not directly related to inducing autonomous activity, including T305+T306 (T305/6), S314 (Patton et al. 1990; Hanson & Schulman 1992a), and T253 (Skelding et al. 2010; Migues et al. 2006; Gurd

+Ca2+ BASAL CaM STIMULATED P-T286 AUTONOMY P P-T305/6 NO STIMULATION P P AUTOPHOSPHORYLATION ~10 nm ~20 nm A B C P

KINASE DOMAIN VARIABLE ASSOCIATION DOMAIN

T286 T305/6

REG CBS

P P

(22)

9 et al. 2008). The circumstances under which the latter two modifications are promoted and their physiological consequences are not well understood.

T305/6 phosphorylation negatively regulates CaM binding (Hanson & Schulman 1992a; Patton et al. 1993) and has been associated with regulating the stimulation thresholds for generating LTP and LTD (Elgersma et al. 2002). As T305/6 phosphorylation blocks CaM binding, CaM binding also blocks T305/6 phosphorylation. Because only a low level of T306 phosphorylation is observed basally (Hanson & Schulman 1992a), the T305/6 phosphorylation therefore occurs after CaM dissociation from P-T286 autonomous kinase.

REGULATION OF SYNAPTIC PLASTICITY BY CAMKII

Soon after it was discovered and charaterized in the late 1970s and early 1980s (Bennett et al. 1983; Grab et al. 1981; Kennedy et al. 1983; Yamauchi & Fujisawa 1980; Schulman & Greengard 1978), CaMKII was suspected to function as a memory molecule in the brain (Lisman & Goldring 1988). Since then, CaMKII has been widely studied in the field of synaptic plasticity across multiple organisms and brain areas. CaMKII has a particularly established role in the regulation of LTP in CA3 to CA1 synapses of the hippocampus (for reviews see (Coultrap & Bayer 2012; Hell 2014; Lisman et al. 2012)), which will be the specific type of LTP referred to throughout this document unless otherwise specified.

CaMKII and LTP

CaMKII was one of the first proteins to be shown to regulate LTP (Malinow et al. 1989) and is one of the most important proteins in the process (for review see (Luscher & Malenka 2012)). Specific inhibitors of CaMKII prevent the induction of LTP when applied prior to the LTP stimulus, including KN62 (Barria, Derkach, et al. 1997a; Ito et al. 1991), KN93 (Barria & Malinow 2005), auto-inhibitory peptide (AIP, (Lengyel et al.

(23)

10 2004)), and tatCN21 (Buard et al. 2010). In agreement with the connection between LTP and memory, these inhibitors also impair behavior on memory tasks (Wolfman et al. 1994; Tan & Liang 1996; Buard et al. 2010). Likewise general kinase inhibitors (H7 (Malinow et al. 1988) and staurosporine (Matthies et al. 1991)) and CaM inhibitors (calmidazolium (Fukunaga et al. 1995)) block LTP induction.

While inhibition of CaMKII blocks LTP, expression of a fully active truncated CaMKII (tCaMKII) mimics LTP; this CaMKII-induced potentiation displays the same hallmarks as electrophysiologically induced LTP (e.g. increased AMPAR currents and EPSC amplitudes) and occludes further potentiation (Hayashi et al. 2000; Lledo et al. 1995; Pettit et al. 1994; Poncer et al. 2002). Mutant mouse models of CaMKII have further solidified the role of the kinase in LTP. In particular, a CaMKIIα KO mouse displays greatly diminished LTP (Elgersma et al. 2002; Hinds et al. 1998; Silva, Stevens, et al. 1992b), as well as severe behavioral impairments in hippocampal-dependent spatial and contextual memory tasks (Elgersma et al. 2002; Gordon et al. 1996; Silva, Paylor, et al. 1992a).

Mechanism of CaMKII in LTP

CaMKII has a number of important functions in LTP, both enzymatic and structural. One of the most important targets of CaMKII at the PSD is the GluR1 subunit of the AMPAR. In response to LTP stimuli, CaMKII phosphorylates GluR1 at S831 (Barria, Derkach, et al. 1997a; Barria, Muller, et al. 1997b; Mammen et al. 1997; McGlade-McCulloh et al. 1993), which increases its single-channel conductance (Derkach et al. 1999; Kristensen et al. 2011). S831 is, however, also a substrate for PKC (Roche et al. 1996), though it is unclear what the relative contribution is of the two kinases in phosphorylating this site after LTP. CaMKII also regulates AMPAR cell surface expression (Hayashi et al. 2000; Opazo et al. 2010; Appleby et al. 2011), which

(24)

11 is likely regulated by phosphorylation of Transmembrane AMPAR Regulatory Proteins (TARPs) such as stargazin by CaMKII (Tsui & Malenka 2006; Opazo et al. 2010); TARPs are auxiliary subunits of AMPARs that regulate their surface expression (Jackson & Nicoll 2011). Though AMPARs are an important functional readout of synaptic strength, they are not the only targets whose function is modified by CaMKII after LTP. In particular, CaMKII also regulates protein turnover, both through protein synthesis – by phosphorylation of transcription factors such as CPEB (Atkins et al. 2005; Atkins et al. 2004) – and protein degradation – by phosphorylation and synaptic recruitment of the proteasome (Bingol et al. 2010; Djakovic et al. 2009).

Role of autophosphorylation in LTP

An LTP stimulus also results in autophosphorylation of CaMKII at T826 and autonomous kinase activity (Fukunaga et al. 1993; Fukunaga et al. 1995; Lengyel et al. 2004). The importance of T286 phosphorylation has been shown with mouse models. T286A and T286D mice have deficits in both LTP and behavioral memory (Cho et al. 1998; Giese et al. 1998; Mayford et al. 1996; Mayford et al. 1995; Bach et al. 1995; Lengyel et al. 2004). Additionally, performance on the Morris Water Maze is positively correlated with autonomous CaMKII activity (Tan & Liang 1996). However, while T286 phosphorylation after LTP induction was initially thought to last for at least an hour (Makhinson et al. 1999; Fukunaga et al. 1995; Barria, Muller, et al. 1997b), high resolution imaging has suggested that it does outlast the initial Ca2+-stimulus, but only for minutes (Lee et al. 2009). Further, it has been shown that the persistent increase in P-T286 can at least partially be attributed to increased P-T286 in the cell bodies rather than the dendrites (Ouyang et al. 1999). Though these findings may miss low levels of persistent P-T826 at the PSD that are below the detection limit, functional studies have confirmed that autonomous activity is not required for the post-induction expression or

(25)

12 maintenance of LTP. CaMKII inhibitors that target both stimulated and autonomous activity inhibit LTP when applied before the stimulus, but have no effect on the level of potentiation when applied post-stimulation (Buard et al. 2010; Otmakhov et al. 1997; H. X. Chen et al. 2001). Further, the autonomy-deficient T286A KI mouse, when trained to a criterion level, has normal recall of the task despite requiring more training, suggesting that its deficits are in learning rather than memory (Irvine et al. 2005). Importantly, however, maximal CaMKII activity requires both stimulation and autophosphorylation (Coultrap et al. 2012) and the consequent “stimulated autonomy” is required for the regulation of synaptic strength by CaMKII (Barcomb et al. 2014). Therefore, despite early hypotheses that CaMKII autonomy was a form of molecular memory, it is now thought that its importance is in mediating the full activation of the kinase during LTP induction.

On the other hand, CaMKII is still thought to be a memory molecule, but through structural regulation of the synapse, rather than enzymatic. This role of CaMKII may still involve T286 autophosphorylation, as autophosphorylation may itself regulate the duration of CaMKII association at the PSD. A phosphomimetic T286D mutant has enhanced persistence at the PSD (Shen et al. 2000) and phosphatase activity decreases CaMKII content in the PSD (Yoshimura & Yamauchi 1997). Further, some binding partners of CaMKII at the PSD require T286 phosphorylation, such as GluN1 (Leonard et al. 1999).

The T305/6 phosphorylation site also regulates LTP, but seems to be more important in mediating the threshold at which LTP is induced. A non-phosphorylatable T305/6AV mutant mouse displays altered plasticity thresholds whereby LTP can be induced with lower stimulation protocols than for WT (Elgersma et al. 2002). These thresholds appear to be important in encoding memory, as the T305/6AV mouse is

(26)

13 impaired in both spatial and cued memory tasks (Elgersma et al. 2002). Furthermore, expression of T286D+T305/6A and T286D+T305/6D enhance and diminish synaptic strength, respectively, in hippocampal slices (Pi et al. 2010) and dissociated cultures (Barcomb et al. 2014), showing that T305/6 phosphorylation has a role in regulating the directionality of plasticity. The T305D mouse is also severely impaired, though it is not valid to attribute impairments to the phosphomimetic residue rather than the inability of the kinase to be activated in general (Elgersma et al. 2002). Phosphorylation at T305/6 has also been suggested to negatively regulate the association of CaMKII with the NMDAR and therefore the concentration of CaMKII at the PSD (Shen et al. 2000; Leonard et al. 2002). Indeed, the T305/6AV mouse has an increase in total CaMKII associated with the PSD, which may conribute to the increased response of synapses to smaller stimuli (Elgersma et al. 2002).

Role of presynaptic CaMKII in LTP

Though the focus of CaMKII function in LTP has been on the postsynaptic side, the kinase also has important functions presynaptically (for review see (Z.-W. Wang 2008)). CaMKII regulates vesicular release through phosphorylation of synapsin I (Llinás et al. 1991) and possibly synaptotagmin (Verona et al. 2000). Changes in vesicular release have been associated with alterations in post-tetanic potentiation (PTP) and paired-pulse facilitation (PPF), both of which are altered in the CaMKIIα KO mice (Silva, Stevens, et al. 1992b; Chapman et al. 1995). Conditional knockout of CaMKII on the presynaptic side also results in mild impairments in release probability in response to repetitive stimulation (Hinds et al. 2003). At least one study has argued that these short-term changes may significantly affect behavioral memory (Silva et al. 1996). Furthermore, presynaptic injection of a non-permeable peptide inhibitor of CaMKII blocks LTP and, likewise, presynaptic injection of constitutively active CaMKII induces LTP

(27)

14 when paired with a weak tetanus (Ninan & Arancio 2004). Mechanistically, some functions of CaMKII in presynaptic plasticity are mediated structurally rather than enzymatically; synaptic augmentation, synaptic fatigue, and the number of docked vesicles in presynaptic active zones are all impaired in CaMKIIα KO mice, but not in T305D KI mice (T305D being an inactivating mutation), suggesting that a structural interaction of CaMKII is regulating these forms of short-term presynaptic plasticity (Hojjati et al. 2007). Further, the CaMKII regulation of presynaptic Cav2.1 channels has been attributed to a non-enzymatic function of the kinase (Jiang et al. 2008). It should be noted that a number of studies have focused on presynaptic CaMKII in motor neurons of invertebrates (He et al. 2000; Jin & Hawkins 2003; Shakiryanova et al. 2007; Shakiryanova et al. 2011), though there is a dearth of information from the mammalian hippocampus. Thus, CaMKII clearly plays a role in plasticity presynaptically, though it has not been well enough studied to fully determine its function and physiological implications.

Role of CaMKIIα versus CaMKIIβ in LTP

Another outstanding question in the LTP field is the role of CaMKIIβ in the process. In CaMKIIα KO mice, the expression of CaMKIIβ is intact, with some studies indicating an increase in either the total expression of CaMKIIβ (Coultrap, Freund, et al. 2014a) or PSD-localized CaMKIIβ (Elgersma et al. 2002). While the CaMKIIα mouse is severely impaired in LTP, it is not entirely eliminated; the residual LTP is thought to be mediated by CaMKIIβ, though this hypothesis has not been formally tested experimentally. Evidence for this effect of CaMKIIβ comes from the T305D mutant mouse, which expresses a non-activatable dominant/negative form of CaMKIIα. Unlike the CaMKIIα KO mouse, the T305D mutant does not have enhanced levels of CaMKIIβ

(28)

15 at the PSD and is likewise more severely impaired than the α-KO in both LTP and behavioral memory abilities (Elgersma et al. 2002).

On the other hand, CaMKIIα and β each have unique traits that may differentiate their function in LTP. The mRNA of CaMKIIα, but not CaMKIIβ, is expressed both in the cell body and in neuronal dendrites, the latter allowing rapid translation of CaMKIIα at dendritic polyribosomes after LTP (Ouyang et al. 1999; Giovannini et al. 2001). LTP is associated with an increase in CaMKIIα protein levels in a protein synthesis dependent manner (Ouyang et al. 1999) and mutating the dendritic targeting domain in the 3’UTR of CaMKIIα alone impairs late-phase LTP and memory consolidation (Miller et al. 2002). While CaMKIIα translation is enhanced by NMDAR activation, CaMKIIβ translation is oppositely regulated – decreased neuronal activity leads to an increase in CaMKIIβ protein levels, suggesting a role for CaMKIIβ in homeostatic plasticity (Thiagarajan et al. 2002; Groth et al. 2011).

CaMKIIα and β also have unique targeting domains and CaMKIIβ has a significantly slower translocation to synapses after an LTP stimulus (Shen & T. Meyer 1999). CaMKIIβ has an interaction with F-actin that is not shared with CaMKIIα (O'Leary et al. 2006; Shen et al. 1998). Likewise, CaMKIIβ has been shown to be important in synapse formation and spine maturity (Fink et al. 2003; Okamoto et al. 2007). Thus it seems that CaMKIIβ cannot substitute for all functions of CaMKIIα in LTP, and vice versa, α cannot substitute for all functions of β.

A CaMKIIβ KO mouse has impaired LTP at lower stimulation protocols and impaired contextual fear conditioning (Borgesius et al. 2011). This deficit is not present in a CaMKIIβ-A303R mutant, which is catalytically inactive, suggesting a structural role of CaMKIIβ in LTP (Borgesius et al. 2011). Consistent with this hypothesis, CaMKIIβ KO,

(29)

16 but not CaMKIIβ-A303R, has decreased synaptic targeting of CaMKIIα, indicating that CaMKIIβ has a role in regulating the subcellular localization of CaMKIIα and that this targeting is important for LTP (Borgesius et al. 2011).

CaMKII in LTP maintenance

The role of CaMKII in the maintenance phase of LTP has been disputed for decades (Sanhueza & Lisman 2013; H. X. Chen et al. 2001). The persistent association and activation of CaMKII for potentially hours after LTP induction suggests in theory that CaMKII may play a role in maintaining that potentiation. However, most studies have not found an effect on post-induction synaptic strength when inhibitors of CaMKII activity are applied after LTP has been induced (Otmakhov et al. 1997; H. X. Chen et al. 2001; Ito et al. 1991; Buard et al. 2010), though it should be noted that some studies have seen a reduction (T. P. Feng 1995; J. H. Wang & Kelly 1996). One persistent theory as to why CaMKII inhibitors may not decrease LTP maintenance is that CaMKII plays a structural rather than enzymatic role in the process (Erondu & Kennedy 1985; Giovannini et al. 2001). As modeled in Figure 1.2, CaMKII is a multimeric protein that can bind to multiple proteins at once (Robison, Bass, et al. 2005b; Colbran & Brown 2004), therefore it may act as a scaffold at the PSD while cytoskeletal remodeling is occurring and also recruit additional proteins to the targeted synapse. CaMKII inhibitors would only reverse this process if they were able to effectively compete with high affinity protein-protein interactions. One such inhibitor, tatCN21, appears to be able to both compete off CaMKII associations with the NMDAR and reverse LTP maintenance (Sanhueza et al. 2011). This effect of tatCN21 has been further tested for specificity of mechanism, which will be the focus of Chapter 5.

(30)

17 CaMKII and LTD

Until recently, little attention had been paid toward the role of CaMKII in long-term depression (LTD), despite early evidence from the CaMKIIα KO mouse displaying impaired LTD (Stevens et al. 1994) and inhibition of LTD by KN62 (Stanton & Gage 1996). This LTD impairment has been confirmed in a subsequent CaMKIIα KO mouse and new model for differential CaMKII activity during LTP and LTD has been proposed, as described below (Coultrap, Freund, et al. 2014a).

Interestingly, both NMDAR-dependent LTP and LTD are characterized by a post-induction increase in T286 phosphorylation (Coultrap, Freund, et al. 2014a; Marsden et al. 2010). This shared mechanism leads to a conundrum: how do two opposing processes share an underlying mechanism? It has been proposed that the differential stimuli – strong but brief for LTP and weak but prolonged for LTD – lead to different substrate specificities of the kinase. Indeed, CaMKII has different substrate types that respond differently to the autonomous activity state (Coultrap et al. 2012; Ishida & Fujisawa 1995). GluA1 S831 is considered a regular substrate and is phosphorylated during LTP but not LTD, while GluA1 S567 is a different type of substrate that is phosphorylated equally well by both stimulated and autonomous CaMKII and is phosphorylated during LTD but not LTP (Coultrap, Freund, et al. 2014a). This latter phosphorylation site has been shown to regulate removal of AMPARs from the synapse (W. Lu et al. 2010), which is a component of LTD. Further, LTD-like stimulation induces CaMKII translocation to inhibitory synapses and this translocation is dependent on T286 phosphorylation and is associated with an increase in GABAAR surface expression (Marsden et al. 2010).

LTD signaling also appears to be regulated by T305/6 phosphorylation, as overexpression of a T286D mutant (likely hyperphosphorylated at T305/6 (Barcomb et

(31)

18 al. 2014)) or a T286D+T305/6D mutant reduces AMPAR EPSCs and occludes further LTD in hippocampal slice cultures (Pi et al. 2010). For the T286D mutant, blocking NMDARs or inhibiting PP2B rescued the depression in AMPAR EPSCs, implicating activity-dependent LTD signaling (Pi et al. 2010).

LTD can be also induced by a mGluR-dependent mechanism. Inhibition of CaMKII with tatCN21 blocks mGluR LTD (Bernard et al. 2014). Further, CaMKII T286 phosphorylation is upregulated and CaMKII activity mediates protein synthesis during mGluR LTD (Mockett et al. 2011).

CAMKII AND THE NMDAR

The NMDAR functions as a Ca2+-permeable ion channel and is important for many forms of plasticity, including LTP and LTD. This role for the NMDAR relates to CaMKII in that it provides a Ca2+ source for CaMKII activation; both LTP-induced T286 phosphorylation and synaptic translocation of CaMKII are blocked under most conditions by the NMDAR channel blocker, APV (Barria, Muller, et al. 1997b; Fukunaga et al. 1995; Thalhammer et al. 2006; Shen & T. Meyer 1999). CaMKII additionally has important interactions with the cytoplasmic C-tails of the individual NMDAR subunits. There is some evidence that these interactions lead to the modification of channel kinetics, as phosphorylation of S1303 on GluN2B, a CaMKII site, increases NMDAR currents and coexpression of CaMKII with the NMDAR in heterologous cells decreases receptor desensitization (Liao et al. 2001; Sessoms-Sikes et al. 2005). However, the majority of research has centered on the function of stable interactions between CaMKII and the NMDAR subunits. The GluN2 subunits in particular have long C-tails (~300 amino acids) that are studded with protein-protein interaction sites and post-translational modifications (for review see (Sanz-Clemente, Nicoll, et al. 2013b)). CaMKII most strongly interacts with GluN2B, though interactions with GluN1 and GluN2A occur as well (Figure 1.2).

(32)

19 Biochemical Regulation of CaMKII binding to the NMDAR Subunits

GluN1

GluN1 interacts with CaMKII on the C0 domain of its C-tail (Leonard et al. 1999; Leonard et al. 2002). This interaction requires CaMKII autophosphorylation at T286, but is impaired by autophosphorylation of T305/6 (Leonard et al. 1999). While CaM is not required for CaMKII/GluN1 binding, it does mediate the interaction in cells by displacing α-actinin from GluN1; basally both apo-CaM and α-actinin are bound to GluN1 and upon Ca2+ binding to CaM, α-actinin is displaced and CaMKII is recruited (Merrill et al. 2007). Further, α-actinin is competitive with both CaM and CaMKII for GluN1 binding (Leonard et al. 2002).

GluN2A

CaMKII has been shown to have a stable interaction with the cytoplasmic C-tail of GluN2A between residues 1389-1461 (Gardoni, Schrama, et al. 2001b; Gardoni et al. 2006; Gardoni, Bellone, et al. 2001a; Gardoni et al. 1999), though this interaction is of an affinity that is at least an order of magnitude less than that of the other NMDAR subunits and requires more sensitive methods to measure (Leonard et al. 2002). This site is immediately N-terminal to a PDZ domain that binds PSD-95, and CaMKII competes with PSD-95 for GluN2A binding (Gardoni, Schrama, et al. 2001b). The interaction of CaMKII with the NMDAR is enhanced after LTP (Leonard et al. 1999; Strack, Barban, et al. 1997a) and, interestingly, there is also a reduction in the amount of PSD-95 that co-immunoprecipitates with GluN2A/2B after an LTP stimulus (Gardoni, Schrama, et al. 2001b). CaMKII/GluN2A is also negatively regulated by phosphorylation of GluN2A on S1416 by PKC (Gardoni, Bellone, et al. 2001a). GluN2A is as well a CaMKII substrate at S1289, though this phosphorylation does not affect the CaMKII/GluN2A interaction (Gardoni, Schrama, et al. 2001b).

(33)

20 Figure 1.2: CaMKII interactions with the NMDAR cytoplasmic tails

CaMKII has interactions with GluN1, GluN2A, and GluN2B, indicated by purple boxes, with low affinity interactions denoted by shading. GluN1 interacts with CaMKII through its C0 domain (between residues 840-860). This binding site also contains interaction sites for α-actinin and CaM, which are competitive and promotional towards CaMKII binding, respectively. This interaction requires T286 phosphorylation. GluN2A has a low affinity interaction with CaMKII between residues 1389 and 1461. This region overlaps with a PDZ domain (orange), and likewise CaMKII is competitive with PSD95 for GluN2A binding. CaMKII phosphorylates GluN2A at S1291. The alignment of the homologous region on GluN2B (GluN2B 1407-1479 + PDZ domain) is shown below, highlighting the PDZ domains in orange. GluN2B has two interaction sites with CaMKII, one low affinity site between residues 839 and 1133 and a high affinity site between residues 1290 and 1309. The sequence of this latter region is expanded to the right of the GluN2B schematic, showing the similarities between GluN2B and CaMKII (top) and GluN2A (bottom) in the homologous regions, phosphorylation sites are highlighted in red. The homology between GluN2B(1290-1309) and CaMKII(273-293) relates to their shared interaction with the T-site of CaMKII, and the mechanisms that regulate both CaMKII/GluN2B binding and CaMKII clustering. For GluN2A, the “IN” residues underlined in the sequence are thought to be an insertion that prevents CaMKII binding, despite the surrounding homology to GluN2B. Deletion of the “IN” insert induces an interaction between CaMKII and GluN2A, while insertion of “IN” into GluN2B reduces its binding to CaMKII. In both alignments, solid lines identify identical residues and dashed lines/colons indicate conservative substitutions.

273HRSTVASCMHRQETVDCLKKF293 CaMKII!

::|| ::| : :!

1290AQKKNRNKLRRQHSYDTFVDL1310 GluN2B!

||||||:::| !

1276ALQFQKNKLKINRQHSYDNILDK1298 GluN2A!

840 860 1290 1309 839 1120 P S1303 1276 1298 1389 1461 PSD95 P S1291 !-Act CaM

GluN1

GluN2A

GluN2B

P

CaMKII Binding Site PDZ Domain

CaMKII Phosphorylation Site CaMKII Competitor CaMKII Co-Binder 1389HSLPSQAVNDSYLRSSLR----STASYCSRDSRGHSDVYI1424 GluN2A! :: :| : :| : | |: |! 1407PTVAGASKTRPDFRALVTNKPVVSALHGAVPGRFQKDICI1446 GluN2B! ! 1425SEHVMPYAANKNNMYSTPRVLNSCSNRRVYKKMPSIESDV1464 GluN2A! | : | ||::| || :|| |: ||||||! 1447GNQSNPCVPNNK----NPRAFNGSSNGHVYEKLSSIESDV1482 GluN2B! ! 1407 1479

(34)

21 The homologous region on GluN2B (residues 1407-1479) does not bind CaMKII (Strack & Colbran 1998), despite a high degree of homology in the region just adjacent to the PDZ domain (see Figure 1.2). However, further N-terminal the sequences differ more significantly. The exact binding site of GluN2A has not been well defined, and in particular the residues contributing to the interaction have not been determined. Thus, it is unclear where in the alignment of this region the divergence between GluN2A and GluN2B is important.

A second area of the GluN2A C-tail – residues 1279-1298 – is homologous to the CaMKII binding site of GluN2B (Figure 1.2). Though this region is 45% identical in sequence, it does not appreciably bind to CaMKII (Strack, McNeill, et al. 2000a). Furthermore, making GluN2B more “GluN2A-like” significantly reduces its interaction with CaMKII (Strack, McNeill, et al. 2000a).

GluN2B

CaMKII has a high affinity interaction with GluN2B (Kd = 138 nM) on its cytoplasmic C-tail within residues 1290-1309 (Strack & Colbran 1998; Strack, McNeill, et al. 2000a). There is also a second, lower affinity interaction site within residues 839-1120, which is dependent on P-T286 (Bayer et al. 2001); the role and possible physiological relevance of this binding site remain unknown. The high-affinity site on GluN2B is homologous to the autoinhibitory region of CaMKII and binds to the same area of the site (Figure 1.2; (Bayer et al. 2001)). Other than GluN2B, other known T-site interactors include CaMKII itself (discussed below) and CaMKII-N – an endogenous CaMKII inhibitory protein (Chang et al. 1998; Chang et al. 2001). Other proteins have been suggested to be T-site binders such as connexin-36 (Alev et al. 2008), densin-180 (Jiao et al. 2011), and voltage gated calcium channel β-subunits (Grueter et al. 2008).

(35)

22 Table 1.1: Effect of mutations and other manipulations on CaMKII/GluN2B binding

EFFECT RATIONALE REFERENCE

CaMKII MUTANTS AND MODIFICATIONS

CaMKII (1-326) ê Trans. in neurons Requires holoenzyme Bayer et al., 2006

K42R No trans. in HC

Nucleotide binding mutants

Strack et al., 2000

K42M ê in vitro ê Trans .in HC

ê Trans. in neurons

O’Leary et al., 2011 T286A

Normal in vitro Normal Trans. in HC

Normal Trans. in neurons P-T286 not required or deleterious

Bayer et al., 2001 O’Leary et al., 2011

T286D Normal in vitro Normal trans in HEK293 Raveendran et al., 2009

F98K E139R

ê in vitro

ê Trans. in HEK293 ê Trans. in neurons (normal persistent phase)

S-Site Mutants; required

for transient binding Bayer et al., 2006

A302R No Trans. in HC Prevent stimulation by

CaM (can be overcome by addition of T286D)

Strack et al., 2000 T305/6D

P-T305/6 ê in vitro O’Leary et al., 2011

I205K No Trans

T-site mutants; required for persistent binding

Bayer et al., 2001 I205K W237K ê in vitro ê Trans. in HC ê Trans. in neurons (no persistent phase)

Bayer et al., 2006

GluN2B MUTANTS AND MODIFICATIONS*

L1298D/G R1300A/Q R1300E/Q ê in vitro No Trans. in HC CaMKII-Binding Site

Mutants Strack et al., 2000

S1303A

Normal in vitro Normal trans in HC, or

é basal interaction in HC S1303 phosphorylation is

inhibitory

Strack et al., 2000 O’Leary et al., 2011

Barcomb et al., 2014+

S1303D

P-S1303 ê or normal in vitro Normal trans. in HC

Strack et al., 2000 O’Leary et al., 2011 Raveendran et al., 2009

GluN”2A-like” ** ê in vitro No Translocation GluN2A does not preserve

binding

Strack et al., 2000

GluN2B+IN *** ê in vitro Mayadevi et al., 2002

GluN2B (Δ1289-1296) ê in vitro CaMKII-Binding Domain Mutant Strack et al., 2000

INHIBITOR/COMPOUND/PROTEIN

Syntide-2 No effect Non-competitive inhibitor Strack et al., 2000 Bayer et al., 2006

Autocamtide-2 ê in vitro Competitive at T-site Strack et al., 2000

Autocamtide-3 Bayer et al., 2006

High [ATP] ê in vitro Inhibitory phosphorylation Strack et al., 2000

High [ADP] é in vitro Nucleotide enhancement O’Leary et al., 2011

Microcystin é PSD retention in vitro Maintenance of pT286 Strack et al., 2000

PP1 ê PSD retention in vitro Decreased pT286 Strack et al., 2000

H7

Staurosporine

Normal in vitro Normal trans. in HC Normal trans. in neurons

Enzymatic activity not

required Barcomb et al., 2014+

Trans. = Translocation; HC = Heterologous Cells * in vitro experiments use only the C-terminal domain

** GluN”2A-like” = A1290Q/Q1291F/K1292Q/R1295K/N1296L/R1299N

*** GluN2B+IN = GluN2B with residues “IN” inserted at residue 1300 (see Figure 1.2)

(36)

23 Because CaMKII/GluN2B binding plays an important role in LTP induction (Barria & Malinow 2005; Halt et al. 2012; Zhou et al. 2007), the biochemical regulation of the interaction has been the focus of a number of studies, which are described in detail below and summarized in Table 1.1. This research combines the assessment of binding

in vitro with the measurement of protein colocalization within cells. The latter technique

is useful as CaMKII translocates to GluN2B upon stimulation, allowing for the determination of factors regulating basal versus stimulated interactions. CaMKII translocation to GluN2B can be mimicked in heterologous cells, allowing for analysis of the interaction without the confounding factors of other synaptic interactions (Bayer et al. 2001; Strack, McNeill, et al. 2000a), though it should be noted that CaMKII does not translocate to either GluN1 or GluN2A in heterologous cells (Strack & Colbran 1998).

GluN2B interacts with the T-site of CaMKII, therefore the kinase must be active in order to expose the binding site and permit the interaction. Activation can be either by Ca2+/CaM stimulation or T286 phosphorylation (Bayer et al. 2001; Strack, McNeill, et al. 2000a). Contrary to some early studies (Strack & Colbran 1998; Strack, McNeill, et al. 2000a), T286 phosphorylation is not required for the interaction to occur (Bayer et al. 2001; Shen & T. Meyer 1999). Previous results suggesting that the interaction requires P-T286 may be explained by the absence of nucleotides (i.e. ATP) in the binding reaction with non-phosphorylated CaMKII, as nucleotides are required for the full expression of the interaction (O'Leary et al. 2011; Robison, Bartlett, et al. 2005a).

As T286 is homologous to S1303 on GluN2B (Figure 1.2) and T286 phosphorylation reduces the interaction of the kinase with the T-site (Yang & Schulman 1999), it is also expected that phosphorylation of S1303 – a CaMKII substrate (Omkumar et al. 1996) – would reduce the CaMKII GluN2B interaction. While it has been shown that S1303 is involved in the interaction (Strack, McNeill, et al. 2000a),

(37)

24 mixed results have been found for its effects on binding after phosphorylation. Strack et al., (2000) found that the interaction of CaMKII with a phosphomimetic S1303D mutant or with P-S1303 was reduced in vitro, though translocation was normal within heterologous cells. However, this study used coexpression with GluN1 for the translocation assay and GluN1 also binds CaMKII (Leonard et al. 1999). O’Leary et al., (2011) found that the interaction was significantly reduced in vitro, but did not test the interaction within cells. Raveendran et al., (2009) saw a reduced interaction in vitro for both WT and T286D CaMKII with either the S1303D mutant or P-S1303 GluN2B, but only in the absence of Ca2+/CaM. These results seem to imply that phosphorylation at S1303 is inhibitory towards the CaMKII/GluN2B interaction, but does not completely eliminate it.

As phosphorylation of CaMKII and GluN2B both have a partial role in regulating CaMKII/GluN2B binding (O'Leary et al. 2011), phosphatase activity likely contributes to the induction and/or persistence of the interaction (Strack, McNeill, et al. 2000a). However, while CaMKII phosphorylation at T286 can promote the interaction by making the T-site accessible in the absence of Ca2+/CaM, GluN2B phosphorylation at S1303 can directly hinder the interaction. Thus, the potential effects of phosphatase activity are not entirely straightforward. Further complicating the situation, T305/6 phosphorylation has also been suggested to regulate PSD retention of CaMKII (Shen et al. 2000) and reduce CaMKII/GluN2B binding (O'Leary et al. 2011). The role of these phosphorylation sites and the regulation of binding by phosphatase activity will be discussed further in Chapter 5.

The holoenzyme nature of CaMKII may also regulate its interaction with GluN2B, as monomeric CaMKII has not been seen to translocate to synapses in neurons (Bayer et al. 2006). Furthermore, some studies have found that monomeric CaMKII does not

(38)

25 bind to GluN2B in vitro (Strack, McNeill, et al. 2000a; Bayer et al. 2006), though at least one study did measure an appreciable interaction (Robison, Bass, et al. 2005b). Thus, while the current model indicates that the kinase domain is sufficient for the interaction (Bayer et al. 2006), it is possible that the holoenzyme nature of the kinase allows it to bind multiple GluN2B subunits, increasing the effective affinity.

Extrasynaptic CaMKII clustering

Many of the biochemical factors regulating CaMKII/GluN2B binding are very similar to those regulating interactions between CaMKII holoenzymes, referred to here as CaMKII clustering. These large CaMKII clusters are formed under ischemic conditions (Dosemeci et al. 2000; Tao-Cheng et al. 2001) and thought to be involved in regulating cell death after stroke or other ischemic insults (Ashpole & Hudmon 2011; Vest et al. 2010). Clustering interactions occur between the T-site of a CaMKII kinase domain and the regulatory domain of another subunit on another holoenzyme (Hudmon et al. 2005; Hudmon et al. 1996) – the regulatory domain bearing homology to the GluN2B C-tail (Figure 1.2). Phosphorylation of T286, the homologous residue to S1303, negatively regulates clustering (Hudmon et al. 2005; Vest et al. 2009). Also like GluN2B binding, CaMKII clustering requires occupation of the ATP binding pocket (Vest et al. 2009), but does not require enzymatic activity. This elucidation of this latter regulatory effect is the subject of Chapter 3.

Regulation of NMDAR Surface Expression by CaMKII

The relative surface expression of GluN2A- and GluN2B-containing NMDARs is regulated by cellular activity and, in particular, CaMKII kinase activity. GluN2A synaptic expression is regulated by the association of GluN2A with Synapse Associated Protein 97 (SAP97), and SAP97 is regulated by CaMKII. CaMKII phosphorylates SAP97 at two different sites, S39 and S232; phosphorylation of S39 induces release of SAP97/GluN2A

(39)

26 from the ER and phosphorylation of S232 releases GluN2A into the synapse (Mauceri et al. 2004; Mauceri et al. 2007). Coincidentally, the T305D mouse has decreased synaptic, but not total GluN2A (Park et al. 2008). Conversely, the surface expression of GluN2B is decreased by CaMKII activity. Specifically, CaMKII targets casein kinase 2 (CK2) to GluN2B, and CK2 phosphorylates GluN2B at S1480, which disrupts the interaction between GluN2B and MAGUK proteins (Sanz-Clemente, Gray, et al. 2013a; Sanz-Clemente et al. 2010; Chung et al. 2004). Therefore, CaMKII activity enhances GluN2A synaptic expression and decreases GluN2B synaptic expression.

Activity Induced Translocation of CaMKII to GluN2B and LTP

In response to activation by high Ca2+ concentration – such as during an LTP stimulus – CaMKII translocates to the post-synaptic density (PSD) in an NMDAR-dependent manner, which has been detected by both biochemistry (Leonard et al. 1999; Strack, Choi, et al. 1997b) and imaging (Gleason et al. 2003; Otmakhov et al. 2004; Shen & T. Meyer 1999). While global glutamate stimulation of neurons induces global translocation of CaMKII to synaptic sites (Bayer et al. 2006), a localized stimulus, such as glutamate uncaging, induces translocation to the stimulated synapse(s) (Zhang et al. 2008; Rose et al. 2009), indicating that CaMKII exhibits input specificity in its selective translocation (Zhang et al. 2008). Furthermore, it also requires local Ca2+ entry at the spine, and translocation does not occur with a global Ca2+ increase (Thalhammer et al. 2006). The translocation is mediated by the GluN2B subunit of the NMDAR (Bayer et al. 2001; Strack, Robison, et al. 2000b; Barria & Malinow 2005; Halt et al. 2012), and increases the amount of CaMKII that co-immunoprecipitates with the NMDAR (Gardoni, Schrama, et al. 2001b).

Synaptic translocation is a rapid process, reaching a maximum within 100 seconds, and leaves CaMKII tightly bound at the PSD with little protein turnover for at

References

Related documents

The increasing availability of data and attention to services has increased the understanding of the contribution of services to innovation and productivity in

Generella styrmedel kan ha varit mindre verksamma än man har trott De generella styrmedlen, till skillnad från de specifika styrmedlen, har kommit att användas i större

I regleringsbrevet för 2014 uppdrog Regeringen åt Tillväxtanalys att ”föreslå mätmetoder och indikatorer som kan användas vid utvärdering av de samhällsekonomiska effekterna av

Närmare 90 procent av de statliga medlen (intäkter och utgifter) för näringslivets klimatomställning går till generella styrmedel, det vill säga styrmedel som påverkar

Den förbättrade tillgängligheten berör framför allt boende i områden med en mycket hög eller hög tillgänglighet till tätorter, men även antalet personer med längre än

På många små orter i gles- och landsbygder, där varken några nya apotek eller försälj- ningsställen för receptfria läkemedel har tillkommit, är nätet av

Det har inte varit möjligt att skapa en tydlig överblick över hur FoI-verksamheten på Energimyndigheten bidrar till målet, det vill säga hur målen påverkar resursprioriteringar

Detta projekt utvecklar policymixen för strategin Smart industri (Näringsdepartementet, 2016a). En av anledningarna till en stark avgränsning är att analysen bygger på djupa