Synaptic Energetics: Consumption of ATP in the Presynaptic and Postsynaptic Terminal and the Interaction of ATP with the Presynaptic Protein Synapsin

Thesis submitted in partial fulfillment of the requirements for the degree of “DOCTOR OF PHILOSOPHY”

Alexandra Stavsky

Submitted to the Senate of Ben-Gurion University of the Negev

05.10.2019

Beer-Sheva

Synaptic Energetics: Consumption of ATP in the Presynaptic and Postsynaptic Terminal and the Interaction of ATP with the Presynaptic Protein Synapsin

Thesis submitted in partial fulfillment of the requirements for the degree of “DOCTOR OF PHILOSOPHY”

Alexandra Stavsky

Submitted to the Senate of Ben-Gurion University of the Negev

Approved by the advisor ______Approved by the Dean of the Kreitman School of Advanced Graduate Studies ______

05.10.2019

Beer-Sheva

This work was carried out under the supervision of Dr. Daniel Gitler and Prof. Yael Amitai In the Department Physiology and Cell biology Faculty of Health Science

Research-Student's Affidavit when Submitting the Doctoral Thesis for Judgment

I, Alexandra Stavsky, whose signature appears below, hereby declare that (Please mark the appropriate statements):

X I have written this Thesis by myself, except for the help and guidance offered by my Thesis Advisors.

X The scientific materials included in this Thesis are products of my own research, culled from the period during which I was a research student.

X This Thesis incorporates research materials produced in cooperation with others, excluding the technical help commonly received during experimental work. Therefore, I am attaching another affidavit stating the contributions made by myself and the other participants in this research, which has been approved by them and submitted with their approval.

Date: 05.11.2019 Student's name: Alexandra Stavsky Signature:

Affidavit stating the contributions of the other participants in this research

Dr. Yoav Shulman, a former doctoral student in our lab, and Prof. Inna Slutsky, our collaborator from Tel-Aviv University, contributed to this research. I state their contribution in the body of the thesis (in the Methods and Results sections) where their data presented with their agreement: Dr. Yoav Shulman carried out all the intracellular electrophysiological recordings in autaptic cultures and measured synaptic puncta intensity or CI of Syb2. Prof. Inna Slutsky and her lab members (mentioned in the paper) measured the FR of SypI-AT1.03 in their confocal system during electrical stimulation.

Alexandra Stavsky Signature: Date: 08.11.2019

I hereby declare that the author’s above statement about my contribution to this research is true and accurate, and I agree with the presentation of these results in her thesis.

Prof. Inna Slutsky Signature: Date: 11.11.2019

Dr. Yoav Shulman Signature: Date: 13.11.2019

Acknowledgments

First I would like to thank Dr. Daniel Gitler. Thank you for teaching me everything! How to think, reconsider, question and solve problems. Thank you for your endless patience during my work in the lab. Thank you for what you have thought me. You showed me what passion for science and knowledge is.

Prof. Yael Amitai, thank you for your elegant and incredibly important inputs.

Huge thanks to Prof. Israel Sekler. Thank you for the small project you gave me the opportunity to grow into this huge topic and took me into the mitochondria world. Thank you for always pushing me forward and believing in me. Thank you for teaching me what pure optimism in science looks like and the importance of it.

Prof. Ilya Fleidervish, thank you for sharing with me your electrophysiology knowledge. You gave me permission to use your set-up, without any experience, and play with it.

Dr. Joy Kahn, thank you for teaching me, for your inputs, for your care and help. Thank you for guiding me, in research and in life. Thank you for the enlightening and fascinating talks we had. You were my voice of reason in the darkest times… I couldn't be where I am and who I am without you.

To all my department members, who became my very good friends a long time ago:

Arik, Ivana, Tomer, Moshe, Oron, Leenor, Milos, Marcel, Marko, Zaga – thank you for being there for me, for the laughs and the fun we had.

Thanks to Ohad Stoler – for teaching me, counseling and helping with everything. For the late dinners and the passion for sweets and food in general.

Special thanks to Maya Rozenfeld and Shirel Arguetti – you're amazing, and I am very lucky to have you both in my life…

Thanks to Shamchal whom I met just recently at the beginning of my writing… You added color to my life and made writing of this thesis into a much easier task. What will I do without you??

To my lab members:

Eden – Thank you for your creativity and unique way of thinking. Thank you for the thoughts and inputs. Thank you for teaching all of us on the importance of good food and drinks…

Yoav – Thank you for everything you thought me, for the talks and the laughs.

Shoham – Thank you for your humor and unique sarcasm, thank you for all the fun time we had while teaching together – I'll miss that a lot!

Yaara – Thank you for your questions and curiosity. For listening and helping me. Thank you for pushing me forward and keeping my feet on the ground.

And Merav! Thank you for being there with me for all those years. For the times we spent talking about science and life, for the laughs and cries, for your words and your deeds. You've been my mentor, my guide and my friend. I couldn't have reached this stage without you!

Last but not least, I'm forever grateful to my parents, Ella Stavsky and Oleg Vilkov. You are my solid cornerstone, my voice of reason and my love. Thanks to my sister Vita and brother Miron, my grandmother Marra and my nieces, Eva and Olivia for making my day much better…

Table of content

ACKNOWLEDGMENTS ...... 5 TABLE OF CONTENT ...... 7 LIST OF TABLES ...... 11 LIST OF FIGURES ...... 12 ABBREVIATIONS ...... 14 SUMMARY ...... 16 INTRODUCTION ...... 18 1 SYNAPTIC TRANSMISSION...... 18 2 SYNAPTIC VESICLE POOLS ...... 19

2.1 THE READILY RELEASABLE POOL ...... 19 2.2 THE RECYCLING POOL ...... 20 2.3 THE RESTING POOL ...... 20 2.4 THE SUPERPOOL AND VESICLE MOBILITY ...... 20 3 THE SYNAPSIN FAMILY ...... 21

3.1 FUNCTIONAL IMPLICATIONS OF THE DELETION OF ALL SYNAPSINS ...... 22 3.2 THE ROLE OF ATP BINDING IN SYNAPSIN ACTIVITY ...... 22 4 SYNAPTIC CALCIUM HOMEOSTASIS ...... 23 5 SYNAPTIC PLASTICITY ...... 25

5.1 SHORT-TERM SYNAPTIC PLASTICITY ...... 25 5.2 LONG-TERM SYNAPTIC PLASTICITY ...... 25 5.3 THE ROLE OF SYNAPSINS IN SYNAPTIC PLASTICITY ...... 27 6 SYNAPTIC MITOCHONDRIA ...... 29

6.1 MITOCHONDRIAL PROVISION OF ATP ...... 29 6.2 MITOCHONDRIAL CALCIUM SIGNALING ...... 31 6.3 MITOCHONDRIAL CALCIUM INFLUX MECHANISM ...... 32 6.4 MITOCHONDRIAL CALCIUM EFFLUX MECHANISM: THE MITOCHONDRIAL SODIUM/CALCIUM EXCHANGER ……………………………………………………………………………………………………………33 6.5 THE ROLE OF MITOCHONDRIA IN SYNAPTIC PLASTICITY ...... 34 RESEARCH OBJECTIVES ...... 36 RESEARCH SIGNIFICANCE ...... 38 MATERIALS AND METHODS ...... 39

7 SOLUTIONS, MATERIALS AND ANTIBODIES ...... 39 8 MICE …………………………………………………………………………………………………………….. 42 9 NEURONAL CULTURES ...... 43

9.1 PRIMARY DENSE HIPPOCAMPAL CULTURES...... 43 9.2 AUTAPTIC CULTURES ...... 43 10 ACUTE HIPPOCAMPAL SLICES...... 44 11 MOLECULAR CONSTRUCTS AND VIRAL PREPARATIONS ...... 44 12 WESTERN BLOT ANALYSIS ...... 45 13 IMMUNOCYTOCHEMISTRY ...... 45 14 FLUORESCENCE MICROSCOPY ...... 46 15 NEURONAL CULTURE FIELD-STIMULATION ...... 46 16 SEMI-QUANTITATIVE SYNAPTIC IMMUNOFLUORESCENCE ...... 46 17 SYNAPTIC 'CLUSTERING INDEX' ...... 47 18 SYNAPSE WIDTH ANALYSIS ...... 47 19 CALCULATION OF SYNAPTIC TARGETING FACTOR ...... 48 20 MITOCHONDRIAL MEMBRANE POTENTIAL MEASUREMENTS ...... 49 21 SYPHY: MEASURING VESICLE CYCLING ...... 49 22 CALCIUM IMAGING ...... 50

22.1 MITOCHONDRIAL CALCIUM...... 50 22.2 PRESYNAPTIC CYTOPLASMIC CALCIUM MEASUREMENTS ...... 51 23 FM DYE LOADING AND UNLOADING ...... 51 24 FLUORESCENCE RECOVERY AFTER PHOTOBLEACHING (FRAP) ...... 52 25 FLUORESCENCE RESONANCE ENERGY TRANSFER (FRET) IMAGING AND ANALYSIS ……………………………………………………………………………………………………………………………….53

25.1 THE SENSITIZED EMISSION METHOD (BEN-GURION UNIVERSITY) ...... 53 25.2 ACCEPTOR-PHOTOBLEACHING METHOD (TEL-AVIV UNIVERSITY)...... 54 25.3 FRET RATIO CALCULATION (TEL-AVIV UNIVERSITY) ...... 54 25.4 ATEAM1.03-SYPI CALIBRATION ...... 55 26 ELECTROPHYSIOLOGY ...... 55

26.1 ELECTROPHYSIOLOGICAL RECORDINGS FROM AUTAPTIC CULTURES ...... 55 26.2 FIELD-POTENTIAL RECORDINGS ...... 56 26.3 PROBABILITY OF RELEASE ...... 56 26.4 LONG-TERM PLASTICITY INDUCTION VIA HIGH-FREQUENCY STIMULATION ...... 56

27 STATISTICS ...... 57 RESULTS ...... 58 1 ATP BINDING TO SYNASPSIN IIA REGULATES USAGE AND CLUSTERING OF VESICLES IN TERMINALS OF HIPPOCAMPAL NEURONS ………………………………………….58

1.1 MUTATING AN ATP-BINDING SITE IN SYNAPSIN IIA DOES NOT DISRUPT ITS STRUCTURE ...... 58 1.2 MUTATING THE ATP-BINDING SITE AFFECTS THE ASSOCIATION OF SYNAPSIN IIA WITH THE VESICLE CLUSTER ………………………………………………………………………………………………………59 1.3 MUTATING THE ATP BINDING SITE IN SYNAPSIN IIA ENHANCES VESICLE CLUSTERING AT THE TERMINAL ………………………………………………………………………………………………………………61 1.4 ATP BINDING BY SYNAPSIN IIA DOES NOT AFFECT VESICLE REDISTRIBUTION OR INTER-SYNAPTIC SHARING ………………………………………………………………………………………………………………..64 1.5 MUTATING THE ATP-BINDING SITE ENHANCES BASAL EVOKED SYNAPTIC RELEASE BUT DOES NOT AFFECT SPONTANEOUS RELEASE …………………………………………………………………………………..66 1.6 ATP BINDING BY SYNAPSIN IIA IS ESSENTIAL FOR ITS ABILITY TO SUSTAIN SYNAPTIC RELEASE DURING PERIODS OF INTENSE ACTIVITY ………………………………………………………………………….67 1.7 INTERACTION BETWEEN ATP BINDING BY SYNAPSIN IIA AND ITS PHOSPHORYLATION STATE ..... 70 1.8 SYNAPSES CONTAINING MITOCHONDRIA CONTAIN A HIGHER DENSITY OF SYNAPTIC VESICLES IN WT BUT NOT IN SYNAPSIN TKO NEURONS ……………………………………………………………………..72 2 THE EFFECT OF NCLX ON THE PROPERTIES OF SYNAPTIC VESICLE CLUSTERS IN THE PRESYNAPTIC TERMINALS ……………………………………………………………………………………75

2.1 DELETION OF NCLX ENHANCES THE SIZE OF THE PRESYNAPTIC VESICLE CLUSTER ...... 75 2.2 NCLX DELETION INCREASES THE STABILITY OF THE PRESYNAPTIC TOTAL VESICLE POOL ...... 76 2.3 THE RCP FRACTION IS HIGHER IN NCLX-KO NEURONS ...... 77 2.4 ENDOGENOUS SYNAPSIN I, BUT NOT SYNAPSIN II, IS OVEREXPRESSED IN NCLX-KO PRESYNAPTIC TERMINALS …………………………………………………………………………………………………………….79 3 EFFECT OF NCLX DELETION ON SYNAPTIC TRANSMISSION AND PLASTICITY ...... 81

3.1 MITOCHONDRIA IN NCLX KO NEURONS SEQUESTER MORE CALCIUM AT REST AND ARE PARTIALLY DEPOLARIZED ………………………………………………………………………………………………………….81 3.2 CALCIUM SHUTTLING THROUGH THE MITOCHONDRIA IS ALTERED IN THE ABSENCE OF NCLX .... 83 3.3 NCLX DELETION AFFECTS CYTOPLASMIC CALCIUM LEVELS IN PRESYNAPTIC BOUTONS DURING SYNAPTIC ACTIVITY ………………………………………………………………………………………………….. 84 3.4 SYNAPTIC RELEASE IS LOWER IN NCLX-KO NEURONS DUE TO A DECREASE IN THE PROBABILITY OF RELEASE ……………………………………………………………………………………………………………….. 85 3.5 NCLX-KO EXHIBIT HIGHER FREQUENCY FACILITATION IN THE HIPPOCAMPUS SCHAFFER COLLATERAL PATHWAY …………………………………………………………………………………………….. 87 3.6 LONG-TERM POTENTIATION IS IMPAIRED IN NCLX-KO ...... 88 4 THE RELATIONSHIP BETWEEN SYNAPTIC ATP LEVELS, MITOCHONDRIA AND SYNAPSINS ………………………………………………………………………………………………………… 90

4.1 CONSTRUCTION AND USE OF A SYNAPTIC ATP SENSOR ...... 90 4.2 ACCEPTOR-PHOTOBLEACHING AND SENSITIZED EMISSION METHODS FOR DETERMINATION OF FRET DURING SYNAPTIC ACTIVITY ………………………………………………………………………………..91 4.3 CALIBRATION OF SYPI-ATEAM1.03 AND SYPI-ATEAM1.03NL ...... 94 4.4 INTENSE SYNAPTIC ACTIVITY CONSUMES ATP IN THE PRESYNAPTIC TERMINAL ...... 95

4.5 [ATP]SYN LEVELS ARE LOWER IN SYNAPSIN TKO NEURONS, BUT CAN BE RESCUED BY THE REINTRODUCTION OF SYNAPSIN IIA ……………………………………………………………………………….97

4.6 CORRELATION OF [ATP]SYN AND MITOCHONDRIA IN RESTING NEURONS ...... 98 DISCUSSION ...... 100 1 ATP-BINDING TO SYNAPSIN IIA ...... 100 2 THE ROLE OF SYNAPSINS AND MITOCHONDRIA IN REGULATING PRESYNAPTIC [ATP] ……………………………………………………………………………………………………………….. 102 3 THE ROLE OF NCLX IN SV CLUSTERING ...... 105 4 THE ROLE OF NCLX IN MODULATING CALCIUM HOMEOSTASIS, NEURONAL TRANSMISSION AND PLASTICITY ………………………………………………………………………. 107 CONCLUSIONS ...... 111 REFERENCES ...... 112 131 ...... תקציר

List of tables

Table 1: Reagents and chemical compounds ...... 39 Table 2: Fluorescent dyes ...... 40 Table 3: Solutions ...... 41 Table 4: Antibodies for immunocytochemistry ...... 41

List of figures

Figure 1 - Ultrastructural examination of the dispersal of SVs in synapsin TKO neurons...... 22 Figure 2 – Schematic view of synaptic mitochondria...... 31 Figure 3 - Measuring synaptic width...... 48 Figure 4 – FM1-43 dye experiment...... 52 Figure 5 – Schematic illustration of a FRAP of whole presynaptic terminals...... 53 Figure 6 - Mutating the ATP-binding site in synapsin IIa does not overtly alter its structure...... 59 Figure 7 - Mutating the ATP-binding site affects the association of synapsin IIa with the vesicle cluster...... 61 Figure 8 - The K270Q mutation increases the capability of synapsin IIa to cluster vesicles at the presynaptic terminal...... 63 Figure 9 - Mutating the ATP-binding site does not affect vesicle redistribution at rest, during synaptic activity, or during depletion of cellular ATP...... 65 Figure 10 - Effect of synapsin IIa and its mutant on basal synaptic properties...... 67 Figure 11 - Mutating the ATP-binding site annuls the effect of synapsin IIa on synaptic depression...... 69 Figure 12 - Mutating the ATP-binding site in the C domain of synapsin IIa attenuates phosphorylation in site 1 within the A domain...... 71 Figure 13 – Higher density of synaptic vesicles occurs in the presence of mitochondria in the presynaptic terminals...... 73 Figure 14 – Deletion of NCLX increases the synaptic vesicle cluster size in presynaptic terminals...... 76 Figure 15 – FRAP experiments reveal higher synaptic vesicle stability in NCLX-KO neurons...... 77 Figure 16 – NCLX-KO neurons exhibit a higher RcP fraction...... 78 Figure 17 - Synapsin I but not synapsin II overexpress in NCLX-KO...... 79 Figure 18 - Mitochondrial calcium levels and rates are affected by the deletion of NCLX in resting conditions...... 82

Figure 19 – NCLX deletion decreases calcium levels in the mitochondria during neuronal activity...... 83 Figure 20 - Cytoplasmic calcium levels reached during stimulation are different in NCLX-KO and WT synaptic boutons...... 85 Figure 21 - Deletion of NCLX decreases synaptic release and the initial probability of release...... 86 Figure 22 - NCLX-KO exhibit higher frequency facilitation...... 88 Figure 23 - NCLX-KO slices fail to exhibit hippocampal Schaffer-collateral LTP...... 89 Figure 24 - The synaptic FRET-based ATP sensor ATeam1.03-SypI...... 91 Figure 25 – ATeam1.03 FRET measurements using the acceptor photobleaching method...... 92 Figure 26 – Spectral FRET determinations...... 93 Figure 27 - ATP affinity calibration of AT1.03 and AT1.03NL...... 95

Figure 28 - Monitoring activity-dependent changes in [ATP]syn using a synaptic FRET- based ATP sensor...... 96

Figure 29 - Decline in [ATP]syn due to blockage of ATP synthesis...... 97 Figure 30 – Basal FR is higher in WT than in TKO neurons...... 98

Figure 31 - A correlation between resting [ATP]syn and mitochondria...... 99

Abbreviations

ACSF artificial serebrospinal fluid AMPAR α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid APV DL-2-Amino-5-phosphonopentanoic acid Ara-C Cytosine β-D-arabinofuranoside hydrochloride ATP adenosine triphosphate CaM calmodulin CaMKI calcium/ calmodulin-dependent kinase I CaMKII calcium/ calmodulin-dependent kinase II cAMP Cyclic adenosine monophosphate CI clustering index CREB cAMP response element-binding protein CS cover slips DDW double distilled water DIV days in vitro DKO double knock out DMEM Dulbecco's Modified Eagle Medium DNAse I Deoxyribonuclease I DNQX 6,7-Dinitroquinoxaline-2,3(1H,4H)-dione E-LTP early LTP EPSC excitatory postsynaptic currents ER/SR endoplasmic/sarcoplasmic reticulum FBS Fetal Bovine Serum FCCP Carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone fEPSP field excitatory postsynaptic potentiation FR FRET ratio FRAP Fluorescence Recovery After Photobleaching FRET Fluorescence resonance energy transfer FWHM full width at half maximum G green channel GABA gamma-aminobutyric acid GAD Glutamic acid decarboxylase N-[2-[[3-(4-Bromophenyl)-2-propenyl]amino]ethyl]-5-isoquinoline- H89 sulfonamide HEK human embryionic kidney cells HFS high-frequency stimulation IF Immunofluoresence IP3 Inositol triphosphate KO knock-out L-LTP late LTP LTP long-term potentiation MCU mitochondrial calcium uniporter MF-LTP Mossy fibers long-term potentiation

14 mPTP mitochondrial permeability transition pore NCKX potassium-dependent sodium/calcium exchanger NCLX Sodium\calcium\lithium exchanger NCX Sodium\calcium exchanger NMDAR N-methyl-D-aspartate receptors NOS nitricoxide synthase PDL Poly-D-Lysine PFA Para-formaldehyde PKA/PKC protein kinase A / C

Pr probability of release PTP post-tetanic potentiation R red channel RcP recycling pool RP Resting pool RRP readily releasable pool SC-LTP Schaffer Collateral long-term potentiation STP short-term potentiation SV synaptic vesicle Syb2 synaptobrevin 2 SynIIa synapsin Iia sypHy synaptopHluorin SypI synaptophysin I TKO triple knock out TMRM tetramethylrhodamine methyl ester TPP+ tetraphenylphosphonium TTX Tetrodotoxin VGCC voltage gated calcium channel vGlut1 vesicular glutamate transporter 1 WT wild-type

Summary

Synaptic transmission is an energy consuming process. Although it was long assumed that the energetic demands of neurotransmission should impact the cellular energy balance, it was only recently shown that intense synaptic activity causes a transient drop in the ATP concentration within the presynaptic terminal of central nervous system neurons. Mitochondria are considered to be a major supplier of ATP in neurons via oxidative phosphorylation, consequently, it is not surprising that approximately half of the presynaptic terminals harbor mitochondria. However, it has been assumed that this fact should affect synaptic transmission and short term plasticity in individual synapses by producing differences in the dynamics of synaptic ATP. Synapsins are abundant neuronal phospho-proteins that associate with the surface of synaptic vesicles. They possess a well-defined ATP-binding site of undetermined function. We hypothesized that in addition to being an energy source, ATP may directly modulate synapsins' function and synapsin-dependent neurotransmission. To examine our hypothesis, we produced a mutation (K270Q) in synapsin IIa that prevents ATP binding, and reintroduced the mutant into cultured mouse hippocampal neurons devoid of all synapsins. Our results indicate that ATP-binding to synapsin IIa plays a key role in modulating its function and in defining its contribution to hippocampal short-term synaptic plasticity.

We also investigated the relationship between mitochondria, synaptic vesicle clustering and synaptic ATP levels, and whether this connection is synapsin- dependent, thus advancing our understanding of the complex and important topic of synaptic energetics. We found that presynaptic terminals containing mitochondria have larger SV clusters but only in the presence of synapsins. We also found that presynaptic terminals containing mitochondria exhibit higher ATP levels, but also only in the presence of synapsins. These results could indicate that synapsins act as ATP- binding or buffering proteins, and imply complex regulation of the various synapsin isoforms by ATP.

Calcium is a pivotal player in synaptic transmission. Factors regulating presynaptic cytoplasmic calcium transients can influence release probability, vesicle recycling and energy consumption. Mitochondria have an important effect on these processes because they can sequester large quantities of calcium and release it back to the cytoplasm. The mitochondrial calcium steady-state is determined by the balance of calcium influx and efflux mechanisms. The mitochondrial sodium/lithium/calcium exchanger (NCLX) is a major pathway of calcium efflux. We sought to explore the effect of NCLX deletion on synaptic properties, mitochondrial and synaptic calcium homeostasis, synaptic activity and plasticity. Our results suggest that NCLX is an important player in regulating mitochondrial calcium homeostasis and consequently cytoplasmic calcium, and that it thus plays a key role in regulating synaptic activity and plasticity. Moreover, NCLX might affect basal presynaptic properties regulated by synapsin I, as synapsins are regulated not only by ATP and phosphorylation but also by calcium. This project has been published in part (Shulman, Stavsky et al., 2015), and other parts are being prepared for submission.

Introduction

The nervous system confers every living organism with the ability to perceive and appropriately respond to external signals. At any given moment, a multitude of different stimuli are received, processed and integrated by our brain. Chemical communication between the neurons, which is a cornerstone of this extremely complex system, is executed at specialized interface sites between the neurons, the synapses (Alabi and Tsien 2012). The combined action of synapses is responsible for shaping higher brain functions. However, synapses are not static, and are continuously modified by adaptive processes, some of which are responsible for learning and memory. Identifying the basic mechanisms underlying neurotransmitter release and its modification is therefore important for understanding how information is transferred within the brain, and is one of the major aims of neuroscience (Südhof 2004). In this thesis I present my research on basic mechanisms that govern the availability of synaptic vesicles, the cellular organelles that contain the neurotransmitters, the signal molecules.

1 Synaptic transmission

Synaptic activity starts when an action potential, which is initiated at the axon hillock, propagates along the axon and reaches the presynaptic terminal. In response to the ensuing depolarization, voltage gated calcium channels (VGCCs) allow the influx of calcium ions into the cytosol of the presynaptic terminal (Kuromi et al. 2004; Mochida 2018). A prominent feature of most presynaptic terminals is the presence of large synaptic vesicle (SV) clusters next to the active zone (De Robertis and Bennett 1955; Siksou et al. 2007). Synaptotagmin, a central component of the neuronal calcium- sensing mechanism, detects the calcium changes and triggers the fusion of SVs with the presynaptic plasma membrane, releasing their neurotransmitter content into the synaptic cleft (exocytosis; Klenchin and Martin, 2000; Stein et al., 2007; Jahn and Fasshauer, 2012). The latter interacts with postsynaptic receptors and produces a postsynaptic signal (Südhof 2004, 2013; Yao et al. 2009). Following exocytosis, and tightly coupled to it, starts the retrieval of both the membrane and protein

18 components of the SVs and their reconstitution (endocytosis; Cremona and De Camilli, 1997; Chanaday et al., 2019). The total SV population is divided in a hierarchical manner into pools which differ according to their readiness for fusion. This functional division complements endocytosis in its aim to supply SVs for exocytosis during periods of intense synaptic activity. While some vesicles are immediately ready for fusion, others require additional molecular steps in order to make them ready for release. In this manner, a large population of SVs is maintained in the presynaptic terminal, which can serve as a buffer mechanism from which SVs can be recruited during periods of necessity. The availability of SVs for release is a key determinant of pre-synaptic function and serves as an important locus for regulating synaptic strength (Pan and Zucker 2009; Regehr 2012).

2 Synaptic vesicle pools

In pioneering work, it was proposed that the SVs in the presynaptic terminals have different functions (Birks and MacIntosh 1961). In recent years this concept was elaborated, based on the observation that not all vesicles are similar in their location or in their readiness for release in response to stimulation; the vesicles in the presynaptic terminal were therefore divided into three main functional pools: the readily releasable pool, the recycling pool and the resting pool (Denker and Rizzoli 2010). Although not conclusive, the sub-categorization of the vesicle clusters and their molecular and functional heterogeneity offers an important platform for explaining basic principles of synaptic function and modulation (Alabi and Tsien 2012; Chamberland and Tóth 2016; Fowler and Staras 2015; Neher 2015).

2.1 The readily releasable pool

The readily releasable pool (RRP) is defined functionally as that pool which contains the vesicles that are the first to undergo fusion during synaptic activity, meaning they have the highest fusion probability (Kaeser and Regehr 2017). This pool is located specifically and directly at the active zone (Denker and Rizzoli 2010). In hippocampal neurons, it is estimated that the RRP size is ~5% out of the total vesicle pool, and they are typically released after just a few seconds of electrical stimulation delivered at 10-

40 Hz (Alabi and Tsien 2012; Fowler and Staras 2015). The size of this pool is changeable and responds to modulation by signaling pathways. For example, it can be enlarged by the activation of protein kinase C, due to acceleration of SV refilling. Conversely, long-term depression may cause a reduction in the RRP size. The RRP is a subgroup of the recycling pool (Fowler and Staras 2015), which is defined next.

2.2 The recycling pool

The recycling pool (RcP) is defined functionally as the group of SVs that participate in stimulus-evoked recycling, meaning that these are the vesicles that undergo exo- and endocytosis in response to moderate but continuous stimulation (Fowler and Staras 2015). As the RRP is depleted, vesicles from the RcP are mobilized to replace them (Gitler et al. 2008), i.e. the RRP is refilled. The RcP constitutes a fraction of the total pool of vesicles. Its relative size is highly variable, constituting anything from approximately 10-20% out of the total pool (Alabi and Tsien 2012; Fowler and Staras 2015) and up to 90%, depending on signaling cascades, such as the balance between CdK5 and calcineurin activity (Kim and Ryan 2013).

2.3 The resting pool

The resting pool (RP) is defined as the rest of the SVs found in the presynaptic terminal, aside from the RcP vesicles (Fowler and Staras 2015). These SVs are resistant to release, even during intense stimulation, but can be recruited to the RcP as its SVs are depleted (Denker and Rizzoli 2010). The RP is the largest pool within the intra-synaptic total pool, and contains 50-85% of its vesicles (Alabi and Tsien 2012). The RcP and RP vesicles are spatially intermixed within the terminal, yet they have different mobility rates – the RcP vesicles are highly mobile, while the movement of the RP vesicles is restricted (Denker and Rizzoli 2010).

2.4 The superpool and vesicle mobility

The superpool consists of the vesicle population that is not restricted to the synapse. These are extra-synaptic vesicles that dynamically traffic anterogradely and retrogradely along the axons between adjacent synapses. They comprise approximately 4% out of the total pool (Darcy et al. 2006; Fowler and Staras 2015).

This pool includes vesicles from the RcP and RP of many adjacent synaptic terminals (Orenbuch et al. 2012a; Staras et al. 2010).

Synaptic transmission and the division of the SVs into distinct morphological and functional pools are crucial for the function of the central nervous system, thus these processes are under tight regulation by many different proteins. Among these proteins, we will introduce the synapsin protein family.

3 The synapsin family

Synapsins are the most abundant phosphoproteins that are associated with the membrane of the SVs. In mammals, at least eight different neuronal synapsin isoforms exist, which are created by the alternative splicing of three genes (Cesca et al. 2010). They are proposed to cluster SVs based on their dual interaction with SVs and cytoskeletal elements (Cesca et al. 2010). The association of synapsins with these components is modulated by phosphorylation (Chi et al. 2003; Jovanovic et al. 2001), which has been shown to modulate synaptic transmission (Hosaka et al. 1999). In addition to their association with the SVs and to an actin-based cytoskeletal meshwork, synapsins tend to homo- and hetero-dimerize (Gitler et al. 2004b; Hosaka et al. 1999) thus creating the presynaptic vesicle cluster. Synapsin to synapsin interactions appear to control the fraction of SVs available for release, and thereby regulate the long-term efficacy of neurotransmission (Greengard et al. 1993) and selectively control the mobility of resting pool vesicles (Orenbuch et al. 2012b). Recently, a model was proposed for the formation of a liquid phase within the presynaptic terminal that is composed of synapsins, and in which SVs are reversibly embedded (Milovanovic et al. 2018). Synapsins play an important role in maintaining transmission during periods of strong synaptic activity (Gitler et al. 2004b), as well as in certain forms of short term plasticity (see chapter 5 on synaptic plasticity). The study of synapsin proteins is therefore instrumental for the understanding of the molecular mechanisms modulating synaptic plasticity, and may contribute to the clarification of the cellular dysfunctions

21 underlying different pathologies, such as epilepsy (Garcia et al. 2004), Alzheimer's disease (Scheff et al. 2015), schizophrenia (Dyck et al. 2009) and others.

3.1 Functional implications of the deletion of all synapsins

The role of the synapsins in defining synaptic structure and function is revealed upon their deletion (Ferreira et al. 1995; Hosaka and Südhof 1999; Rosahl et al. 1995; Ryan et al. 1996; Shupliakov et al. 2011). Synapsin triple knockout (TKO) mice are viable, but suffer from abnormalities such as seizures and anomalous reflexes. Furthermore, they exhibit deficits in spatial learning and memory (Gitler et al. 2004b). Synapsin TKO neurons contain significantly fewer SVs at sites close to the synaptic active zone, which is consistent with the acceleration of synaptic depression that is observed when synapsin TKO excitatory neurons are stimulated repeatedly (Gitler et al. 2004b; Orenbuch et al. 2012b) However, the deletion of the synapsins does not affect excitatory synaptic transmission when evoked by single stimuli (Gitler et al. 2004b, 2008; Song and Augustine 2016). In addition, in synapsin TKO neurons, SV mobility is enhanced, resulting in the redistribution of the SVs into the axon and the partial dissolution of the SV cluster (Figure 1) (Fornasiero et al. 2012; Orenbuch et al. 2012c).

Figure 1 - Ultrastructural examination of the dispersal of SVs in synapsin TKO neurons. Three-dimensional reconstruction of synapses from WT and synapsin TKO neurons, with SVs represented as brown spheres. In synapsin TKO neurons the vesicle cluster is dispersed and vesicles are redistributed into the axon. Adapted from Orenbuch et al. 2012b.

3.2 The role of ATP binding in synapsin activity

It is now established that brain function, and specifically synaptic activity, is a major consumer of energy and of ATP (Rangaraju et al. 2014). It was reported that the synapsins bind ATP, and crystallization studies determined a well-defined ATP binding site within its conserved C-domain (Hosaka and Südhof 1998a). Although all three synapsin isoforms have been shown to bind ATP, they are differently regulated by

22 calcium: synapsin I is calcium activated, synapsin II is calcium independent (Esser et al. 1998; Hosaka and Südhof 1998a) and synapsin III is calcium inhibited (Hosaka and Südhof 1998b). Although the specific role of synapsins in presynaptic terminals is still not fully known, their function may be versatile, depending on the calcium fluxes in the nerve terminals during neuronal activity. This is part of the theory that synapsins evolved to regulate synaptic transmission in a calcium-dependent manner. Determination of the role of ATP binding by synapsin IIa is discussed in chapters 1 and 4.

4 Synaptic calcium homeostasis

Calcium is a highly versatile intracellular signal that operates over a wide temporal range to regulate many different cellular processes. Calcium, as a signal carrier, regulates functions that are common to all eukaryotic cells such as metabolism, enzyme phosphorylation and dephosphorylation, motility, exocytosis, gene expression and programmed cell death (Brini et al. 2014; Carafoli et al. 2001). Other important properties that are controlled by calcium are cell specific; neurons depend on calcium for the control of processes that specific to them. Calcium, besides being the initiator of synaptic transmission, is also its regulator. Processes of exo- and endocytosis of SVs are highly dependent not only on the presence of calcium but on calcium concentrations (Wu and Wu 2014). Moreover, calcium is essential for the processes of short-term plasticity (STP) and long-term potentiation (LTP) which are linked to learning and the formation and consolidation of memory (see synaptic plasticity section). The versatility of calcium as an intracellular messenger depends on its spatial and temporal properties. Spatial properties refer to the idea that calcium enters to the cytoplasm via different pathways (from the extracellular matrix or from internal organelles) thus creating a specific site where calcium changes are localized, referred to as calcium nano- or microdomains (Augustine et al. 2003; Llinas et al. 1992; Shahrezaei and Delaney 2005). This allows a precise localization of the signal but also enhances the sensitivity of the cell to the transient, especially in pre- and postsynaptic

23 terminals. Temporal properties refer to the fact that most calcium signals are introduced into the cells as brief transients. Calcium-based signaling system uses 'on' reactions that introduce calcium into the cell and 'off' reactions that remove it from the cytoplasm. Under resting conditions, free cytosolic calcium levels in neurons are maintained around 50-100 nM, whereas the extracellular free calcium concentration is within the mM range (Berridge et al. 2003; Collins et al. 2001). This 10,000-fold concentration gradient makes calcium stand out compared to other more abundant cations, when it leads to a significant increase (to low µM concentrations) in cytoplasmic calcium. This mechanism of cytoplasmic calcium increase can be achieved by the influx of calcium from the extracellular space via opening of specific ion channels, which include the membrane VGCCs and several ligand-gated ion channels, such as glutamate receptors (Berridge et al. 2003). Another pathway to increase cytosolic calcium is calcium release from internal store organelles like the ER (Foskett et al. 2007) or the mitochondria (Szabadkai and Duchen 2008). The subsequent rise in intracellular calcium results in various calcium-dependent intracellular events. However, prolonged and uncontrolled cytosolic calcium elevation can cause calcium overload which in turn can be toxic and trigger cell death (Szydlowska and Tymianski 2010). Thus, many regulatory mechanisms operate in the cell in order to terminate the calcium signaling events and rapidly restore the calcium to its resting levels. The main mechanisms involve cell membrane protein families (e.g. ATP-driven calcium pumps and sodium/calcium exchangers) (Strehler and Treiman 2005). Another system to assist in maintaining calcium homeostasis is the presence of calcium-binding proteins, for example calbindin, parvalbumin and calretinin (Lukas and Jones 1994; Matthews et al. 2013; Müller et al. 2007), that buffer the calcium ions by binding them. Another mechanism involves calcium channels in cellular organelles such as the endoplasmic-reticulum and mitochondria. Acting as a high capacity calcium buffer, mitochondria maintain cytosolic calcium homeostasis: mitochondria take up calcium from the cytoplasm and can release it back (MacAskill et al. 2010), thus they can modulate both the amplitude and the spatial patterns of calcium signals. The importance of calcium handling by mitochondria will be elaborated on below.

5 Synaptic plasticity

The strength of synaptic connections is modulated at various timescales in a process termed synaptic plasticity (Madison et al. 1991; Malenka and Bear 2004). These dynamic processes of the neuronal system underlie the ability of the brain to change and to behaviorally adapt to a continuously changing environment.

5.1 Short-term synaptic plasticity

Short-term synaptic plasticity, which occurs on a timescale of milliseconds to minutes, regulates the activity of neural networks and information processing in the nervous system (Jackman and Regehr 2017; Regehr 2012; Regehr and Abbott 2004; Zucker and Regehr 2002). Short-term plasticity typically represents a presynaptic mechanism in neurotransmitter release and manifests as an enhancement or decrement in post-synaptic potentials (PSPs) (del Castillo and Katz 1954; Zucker and Regehr 2002). Short-term plasticity can result in synaptic enhancement through three processes: facilitation, augmentation, and post-tetanic potentiation (PTP), which vary in their duration. The molecular mechanisms mediating the various forms of short-term plasticity are still a topic of debate, but all of them are calcium dependent. Katz and Miledi proposed that residual calcium remaining in the synapse after an action potential acts to enhance synaptic transmission (Charlton and Bittner 1978; RENGEL 1992), a model that is supported by more recent studies (Atluri and Regehr 1996; Korogod et al. 2005; Regehr et al. 1994). Augmentation of PSPs is caused by an increase in the probability of vesicle release (Pr), a phenomenon directly affected by presynaptic calcium levels (Jackman and Regehr 2017; Stevens and Wesseling 1999): residual calcium accumulates during high frequency activity thus increasing the Pr. Thus, the rate of calcium clearance from the synapse can determine whether facilitation or PTP occurs. Newer models describe how calcium channel positioning changes in vesicle readiness, or the activation of alternative endogenous calcium sensors contribute to the enhancement of synaptic transmission (Böhme et al. 2018; Gustafsson et al. 2019; Jackman and Regehr 2017; Neher and Brose 2018).

5.2 Long-term synaptic plasticity

To this day LTP is believed to be one of the primary mechanisms giving rise to synaptic plasticity in the brain and remains an uncontested cellular model of learning and memory. In this part we will describe some of the molecular mechanisms related to LTP that are calcium-dependent and thus will be relevant to this thesis topic.

Long-term synaptic modifications have long been postulated to occur in response to the simultaneous activation of both pre- and postsynaptic neurons (Hebb 1952). First experiments on plasticity showed that brief but strong tetanic stimulation of afferent fibers results in a long-lasting increase in synaptic strength or efficacy (Bliss and Lømo 1973). Although LTP occurs at excitatory synapses throughout the brain, the LTP that occurs at the Schaffer collateral–CA1 synapses (SCs) and the Mossy fiber-CA3 synapses (MF) of the hippocampus is to this day the most robust and widely studied forms of this phenomenon. The enhanced synaptic connectivity can arise from either an increase in the presynaptic release of glutamate or from an increase in the sensitivity of the postsynaptic side to glutamate. The best characterized form of presynaptic LTP is described in the Mossy fiber-CA3 synapses, in which repetitive activation of MFs with high frequency stimulation (HFS) triggers a long-lasting increase in presynaptic Pr followed by enhanced synaptic release (Gundlfinger et al. 2010; Nicoll and Schmitz

2005). The increased Pr has been shown to be connected with an increase in presynaptic calcium levels that activates secondary messengers, leading to enhanced presynaptic cyclic adenosine monophosphate (cAMP; Weisskopf et al., 1994) and activation of protein kinase A (PKA; Fourcaudot et al., 2008). The PKA then phosphorylated different presynaptic substrates that promote enhanced synaptic release (Castillo 2012). The discovery of silent synapses and their ability to be unsilenced during LTP provided a postsynaptic mechanism for LTP (Kerchner and Nicoll 2008; Liao et al. 1995). Since then, many more studies have solidified the idea that LTP can be postsynaptic. In the postsynaptic mechanism for LTP, calcium influx into postsynaptic terminals is also necessary for generating and sustaining LTP (Lynch et al. 1983). It is generally accepted that the induction of LTP at the SCs requires activation of postsynaptic N-methyl-D- aspartate receptors (NMDARs) by synaptically-released glutamate that occurs concurrently with adequate postsynaptic depolarization. This results in the relief of the voltage-dependent block of the NMDAR by extracellular magnesium, which in turn allows the entry of calcium into the postsynaptic dendrite spine, producing a rise in the concentration of intracellular calcium. This increase of free cytosolic calcium levels changes the activity of many proteins – effector molecules, for example nitricoxide synthases (NOS; Lu et al., 1999; Maltsev et al., 2019; Zhang et al., 2019)

26 and CaMKII (Chang et al. 2019; Lisman et al. 2014). CaMKII, the main protein of the postsynaptic density, is a calcium/calmodulin-activated enzyme. One of its main functional properties is its ability to phosphorylate itself. This reaction alters the enzyme’s function so that its activity becomes independent of calcium/calmodulin after its initial activation (Bliss and Collingridge 1993; Saitoh and Schwartz 1985) . This property makes CaMKII a good candidate for the initial storage of long-term synaptic memory at the early LTP (E-LTP) stages. For example, phosphorylation of AMPA receptors by CaMKII increases conductance of the channels and number of channels embedded into the neuronal membrane (Chater and Goda 2014; Herring and Nicoll 2016; Shi et al. 1999). Further cascades are activated in order to maintain the LTP for a longer period of time, a phenomenon termed consolidation which is a part of the late LTP (L-LTP; Rosenberg et al., 2014; Lynch et al., 2015). During consolidation, a series of cellular and molecular events take place, including changes in the levels of gene transcription and protein translation, and interfering with any of these key steps prevents the formation of a long-term memory (Rosenberg et al. 2014). Late phase of LTP involves activity of kinases activated during E-LTP, the cascade of MAP-kinases (Martin et al. 1997), CaMKII (Vigil and Giese 2018), cAMP-dependent PKA and different isoforms of protein kinase C (PKC; Jalil et al., 2015). Signals are integrated by a family of extracellularly- regulated kinases which phosphorylate many cytoplasmic and nuclear proteins including transcription factors and elements, such as the cAMP response element binding protein (CREB; Barco et al., 2002; Ying et al., 2002) leading to the synthesis of new mRNAs and proteins. Although the NMDARs are considered a critical entry point for the calcium involved in triggering LTP via CaMKII, other sources of calcium have been shown to participate in this process, when NMDARs are blocked (Alford et al. 1993; Perkel et al. 1993), suggesting that the source for the increase in intracellular calcium may not be exclusively via NMDARs but also through other intra- or extracellular mechanisms.

5.3 The role of synapsins in synaptic plasticity

As synapsins are the most abundant phospho-proteins in the brain and due to their ability to regulate neurotransmitter release by controlling SV availability for release,

27 they are relevant candidates to mediate short and/or long presynaptic plasticity. All synapsins have similar N-terminal domains that can be phosphorylated by CaMKI and PKA. In additions, synapsin I has a phosphorylation site at its C-terminus that can be phosphorylated by CaMKll (Südhof et al. 1989). Early studies with pharmacological LTP inducers showed an increase in the phosphorylation of synapsin I at its CaMKII sites (Browning and Dudek 1992; Parfitt et al. 1991). Additional evidence supported this claim in two different studies. One of them directly measured an increase in mRNA levels of phosphorylated synapsin I in rat dentate gyrus after LTP induction (Morimoto et al. 1998), and the other showed similar results via western-blot analysis (Nayak et al. 1996). Behavioral tests sugges some learning deficits only when synapsin II, but not synapsin I, is absent (Silva et al. 1996). In contrast, a study on synapsin I KO (Rosahl et al. 1993; Silva et al. 1996) and synapsin I and II double KO (DKO) (Spillane et al. 1995) mice concluded that synapsins I and II are not required for either MF LTP (which is thought to be induced and expressed presynaptically) or NMDA receptor-dependent LTP in SCs pathway (which is initially induced postsynaptically), and suggested that they are probably required in STP rater then LTP. PTP is believed to involve fast activity-dependent recruitment of reserve SVs (Zucker 1999). Because synapsins play a major role in assembling and maintaining the reserve pool (Benfenati et al. 1999; Greengard et al. 1993) and control recruitment of SVs from the reserve pool to the RRP during periods of intense activity (Gitler et al. 2004b) they might participate in generation of PTP. Although in synapsin I-null mice there was no apparent effect on PTP (Rosahl et al. 1993, 1995), decreased PTP together with long- lasting depression has been found in synapsin II or synapsin I and II DKO mice (Humeau et al. 2001; Rosahl et al. 1995; Silva et al. 1996). Synapsin I and II had been shown to participate in the process of facilitation. As calcium accumulate during high frequency stimulations, it induces cAMP and CaMK- dependent phosphorylation of synapsins, which, in turn, elevates the vesicular Pr. A study on synapsin I-null mice showed an enhanced facilitation and decreased depression (Rosahl et al. 1993, 1995; Silva et al. 1996). On the other hand, deletion of both synapsin I and II elicited decreased facilitation and increased depression (Humeau et al. 2001; Rosahl et al. 1995). These results could be explained by the

28 decline in vesicle numbers in synapsin-deficient synapses as suggested by these groups, but other explanations like decreased synaptic connections (Barbieri et al. 2018; Vasin et al. 2014) could also be sufficient. Although not conclusive, synapsins participate in synaptic plasticity, mainly in STP. In this work I did not looked at the role of synapsins in synaptic plasticity but the application of the information we found can be helpful in understanding the mechanism behind synapsin-dependent plasticity.

6 Synaptic mitochondria

Maintaining electrochemical gradients and releasing and recycling synaptic vesicles are all intensely energy-demanding processes (Harris et al. 2012). In addition, these processes are all regulated by calcium signaling (Jahn and Fasshauer 2012; Klenchin and Martin 2000). Thus, neurons are in need of mechanisms where local energy usage can be spatially matched to energy production and calcium buffering to tightly control synaptic activity, both spatially and temporally. Mitochondria are ideally suited to support this spatial variability in metabolic demand and calcium buffering because they generate ATP (via oxidative phosphorylation) and take up calcium into the mitochondrial matrix from the cytosol and vice versa. In order to supply these needs, mitochondria are strategically placed in pre- and postsynaptic terminals (Harris et al. 2012; MacAskill et al. 2010; Waters and Smith 2003).

6.1 Mitochondrial provision of ATP

Mitochondria are well-known as the cellular "power house". Most cellular energy, in the form of ATP, is produced in the process of oxidative phosphorylation within the mitochondria (Gautheron 1984), and it is considered to be the main mechanism providing energy for neuronal activity (Hall et al. 2012; Harris et al. 2012), whether presynaptic (Rangaraju et al. 2014) or postsynaptic (Rangaraju et al. 2019a). An increase in mitochondrial matrix calcium concentration upregulates the activity of at least three enzymes of the Krebs cycle as well as the F1F0-ATPase, thus causing increased production of ATP and NADH (McCormack et al. 1990). Synaptic mitochondria provides cellular energy in this form to fuel many active processes important for synaptic transmission, such as reversal of ion movements following the

29 opening of ion channels, exocytosis at the initiation of synaptic transmission, endocytosis and SVs cycling, and support phosphorylation reactions of diverse proteins and molecules (Attwell and Laughlin 2001; Murthy and Camilli 2003). Because of the highly specialized architecture of neurons, ATP diffusion from the central soma to the distal synapses is thought to be too slow to meet these energy requirements (Kuiper et al. 2008). Although mitochondria are located predominantly in neuronal somata and primary dendrites (Chavan et al. 2015), they can be transported along cytoskeletal elements to the distant neuronal processes and thus supply ATP locally. Evidence for the importance of mitochondrial localization and function can be derived from abnormalities observed in disease conditions (Rangaraju et al. 2019b). Due to the fact that only half of the presynaptic terminals contain mitochondria within them (Chavan et al. 2015), the role of mitochondrially-derived ATP in synaptic transmission is still debated. On the one hand, mitochondria can be specifically retained in regions of the axon where the energy demand is high, including synapses (Nguyen et al. 1997) and in particular near the active zones where SVs are released (Perkins et al. 2010; Rowland et al. 2000). Moreover, it has been shown that motile mitochondria passing through boutons dynamically influence synaptic vesicle release, mainly by altering ATP homeostasis in the axons (Sun et al. 2013). On the other hand, others have shown that synaptic vesicle cycling is similar in boutons without mitochondria as in those with mitochondria and that mitochondrially derived ATP is rapidly dispersed in axons, thereby maintaining near normal levels of ATP even in boutons lacking mitochondria (Pathak et al. 2015). Another source for ATP in neurons is glycolysis (Ashrafi and Ryan 2017). Glycolysis provides cells with a rapid source of ATP for a short period of time, particularly in the absence of oxygen. Mitochondrial oxidative phosphorylation produces ATP at a slower rate, but has a higher yield (32 ATP molecules vs 2 produced with glycolysis; Pfeiffer et al., 2001). Upon abolishing different stages of this process, glycolysis has been shown to take part in different stages of the synaptic transmission (Jang et al. 2016; Rangaraju et al. 2014). Furthermore, crosstalk between these two processes has been proposed, so that the main source of energy depends on the cell type, its location and specific energetic demands at a particular time point (Agrawal et al. 2018; Sobieski et al. 2017; Yellen 2018). The relative importance of glycolysis and mitochondrial

30 oxidative phosphorylation in synaptic function remains unclear. In this study, I did not directly examine the glycolysis pathways, and focused on the contribution of mitochondrially-derived ATP to synaptic function.

Figure 2 – Schematic view of synaptic mitochondria. Adapted from Devine and Kittler, 2018.

6.2 Mitochondrial calcium signaling

In addition to the role of mitochondria as a local energy powerhouse, mitochondria play a key role in calcium signaling at these terminals. By using the negative mitochondrial membrane potential created via the oxidative phosphorylation chain, mitochondria can also buffer cytoplasmic calcium and therefore play an important role alongside endogenous intracellular calcium buffers in shaping local calcium levels (Billups and Forsythe 2002; Kang et al. 2008; Lee et al. 2000). The idea that mitochondria control cellular calcium signaling first arose with the discovery that isolated, respiring mitochondria can accumulate significant amounts of calcium (Duchen 1999; Nicholls and Budd 2000). In addition, mitochondria are capable of releasing calcium back to the cytoplasm (Carafoli 1974) implying that the interaction between cytoplasmic calcium levels and mitochondrial activity is bidirectional. On the one hand, accumulation of calcium by mitochondria can activate calcium release from the endoplasmic-reticulum (ER) via activation of the inositol triphosphate (IP3)

31 pathway (Szabadkai and Duchen 2008), reduce the amplitude of cytosolic calcium increase evoked by membrane calcium channel opening and prolong cytosolic calcium elevations (Herrington et al. 1996; Jae et al. 2008). On the other hand, cytosolic calcium regulates mitochondrial metabolism (McCormack et al. 1990), as previously mentioned, and their movement along cytoskeletal elements (Sheng and Cai 2012). The role of mitochondrial calcium handling in presynaptic terminals remains unclear. Some studies showed that the localization of the mitochondria within the presynaptic terminals had little to no effect on synaptic transmission (Waters and Smith 2003) while more recent evidence supports the claim that mitochondrial calcium influx is important for variability in neurotransmission (Ashrafi et al. 2019) and that mitochondria rapidly sequester cytoplasmic calcium, significantly buffering cytoplasmic calcium and thus influencing neurotransmitter release on millisecond time scales (Billups and Forsythe 2002). A possible explanation for the contradicting affect could be that mitochondria are essential during prolonged or instance stimulation (Kang et al. 2008; Vos et al. 2010). Dysfunction of mitochondria, caused by mitochondrial calcium imbalance, has been associated with various neurodegenerative diseases (Britti et al. 2018; Devine and Kittler 2018; Duchen 2000). In order to maintain mitochondrial calcium homeostasis, a strict balance between calcium influx and efflux mechanisms should be maintained. For the past years, many researches have made important discoveries regarding mitochondrial calcium signaling, such as molecular identification of the major players involved in mitochondrial calcium shuttling, on which we will elaborate in the next section.

6.3 Mitochondrial calcium influx mechanism

The mitochondrial calcium uniporter (MCU) is the best characterized mitochondrial calcium uptake pathway. MCU is a highly selective calcium channel (Kirichok et al. 2004) which allows calcium ions to flow into the mitochondrial matrix using the electrical driving force created by the steep negative mitochondrial membrane potential (ΔΨm) of about -180 mV, which is established by the respiratory chain (Mitchell and Moyle 1967). The uniporter, that was identified in 2011 and published in two independent studies, is ubiquitously expressed in mammalian tissue

(Baughman et al. 2011; De Stefani et al. 2011). MCU was identified as a ~40 kDa protein with two transmembrane domains, which create a pore for calcium influx (Chaudhuri et al. 2013; De Stefani et al. 2011). Calcium uptake is performed by both the pore-forming complex MCU and regulatory subunits, for example MICU1 (Perocchi et al. 2010). One of the most important regulatory features of the uniporter is its apparent gating by extra-mitochondrial calcium concentrations. This was first suggested in a study on isolated liver mitochondria (Bragadin et al. 1979) and later through studies in intact cell lines (Igbavboa and Pfeiffer 1988; Montero et al. 2001). These studies showed that elevated calcium concentrations in the extra-mitochondrial medium enhances calcium accumulation within the mitochondrial matrix via the MCU, thus illustrating the allosteric regulation of MCU by cytosolic calcium levels. Beside its regulation by cytosolic calcium, the MCU is regulated by the ATP/ADP ratio in the cytoplasm and its activity is decreased upon high energy demand (Litsky and Pfeiffer 1997) and may also depend on the mitochondrial membrane potential (Igbavboa and Pfeiffer 1988). After the sequestration and buffering in the mitochondrial matrix, calcium is extruded from mitochondria via sodium-dependent or independent pathways (Drago et al. 2011) , as is explained next.

6.4 Mitochondrial calcium efflux mechanism: the mitochondrial sodium/calcium exchanger

The topic of mitochondrial calcium extrusion was first revealed by Carafoli and his colleagues back in the 1970's. Their finding claimed that mitochondria possess a component which allows calcium release from their matrix, because they observed that addition of sodium or lithium (but not potassium) to the solution outside isolated mitochondria causes release of calcium (Carafoli 1974). These findings suggested that sodium/calcium exchange is the main pathway of calcium removal from mitochondria. Existence of a functional sodium/calcium antiporter in mitochondria was known for many years, but the molecular identity of the mitochondrial sodium/calcium exchanger NCLX was revealed only in 2010 (Palty et al. 2010). A leading previous study researched a putative gene, at that time known as FLJ22233, that mediates sodium/calcium exchange (Ohana et al. 2004). The same gene was cloned by the

Lytton group and was considered to be a novel member of potassium-dependent sodium/calcium exchanger (NCKX) subtype of the sodium/calcium exchanger (NCX) superfamily (Cai and Lytton 2004). However, in a follow-up study it was shown that it mediates sodium/calcium anti-portage in a potassium-independent manner. In agreement with this finding, subsequent phylogenetic analysis found that it is the only member of the calcium exchangers subtypes of the NCX superfamily (Ohana et al. 2004). Remarkably, this new transporter mediated lithium-dependent calcium exchange and was termed NCLX (Sodium/Calcium/Lithium exchanger). In summary, localization and functional analyses concluded that NCLX is the long-sought mitochondrial sodium/calcium exchanger (Palty et al. 2010). Up to now, it is known that NCLX is a major mitochondrial calcium extrusion pathway that exchanges 3 sodium ions for 1 calcium ion (Baysal et al. 1994; Jung et al. 1995). Activity of NCLX is ubiquitously found in most cell types and tissues studied so far and is particularly robust in excitable cells, whereas the activity of the proton/calcium exchanger is primarily found in non-excitable cells (Nita et al. 2015). Another calcium efflux pathway is the mitochondrial permeability transition pore (mPTP), although its physiological relevance is still debated. Various factors, including high mitochondrial calcium levels and mitochondrial depolarization, trigger mPTP opening. If the pore opening is sustained, it facilitates solute entrance into the mitochondrial matrix, promoting mitochondrial swelling and decreased mitochondrial membrane potential, causing cellular dysfunction, release of apoptotic factors and eventually cell death (Duchen 2000; Halestrap and Davidson 1990).

6.5 The role of mitochondria in synaptic plasticity

As previously described, due to the metabolic/energetic and calcium buffering abilities of the mitochondria, they are great candidates to regulate and modify not only basic synaptic transmission but also to participate in functional plasticity in the brain. It has been suggested by several studies that changes in mitochondria occur during synaptic activation and LTP (Hirabayashi et al. 2017; Rangaraju et al. 2019a; Smith et al. 2016). For instance, it was shown that mitochondrial energy production and the activity of calcium channels changes during induction of LTP, and that mitochondrial gene expression is enhanced (Todorova and Blokland 2017). Several studies showed

34 the importance in mitochondria calcium handling in relation to LTP. Mitochondrial calcium uptake is persistently increased after the induction of LTP in the hippocampus due to increased calcium pump activity (Stanton and Schanne 1986). Additionally, recordings from neuromuscular junctions showed that PTP is blocked when mitochondrial calcium uptake is inhibited (Narita et al. 2000) suggesting that PTP might result from the slow release of calcium from mitochondria. Furthermore, the role of NCLX has been explored in the hippocampal mossy-fiber (MF) synapses and in the giant synapse of the calyx of held, where application of the then considered NCLX blockers tetraphenylphosphonium (TPP+) or CGP37157 reduced or diminished PTP (Jae et al. 2008; Lee et al. 2000). These results are not conclusive because in addition to the blockers' initially reported effect on NCLX (Luciani et al. 2007; Neumann et al. 2011) are now known to affect also the plasma membrane NCX, the sarcoplasmic reticulum calcium channels and mitochondrial potassium channels (Lukyanetz et al. 2009). For that reason we are in an excellent position to address this question.

Research objectives

Synaptic energetics and calcium homeostasis are to this date a central topic of research regarding basic synaptic properties, synaptic transmission and synaptic plasticity. In this project I addressed four main questions related to these topics: 1. How does ATP-binding to synapsin IIa affect its activity? We hypothesized that activity-dependent spatio-temporal alterations in the concentration of ATP within synapses ([ATP]syn) modulate synapsin function. We aimed to examine the consequences of ATP binding by the synapsins, by reintroducing into synapsin-null neurons synapsin IIa mutated at an amino-acid essential for ATP- binding. Specifically, we decided to probe vesicle clustering in resting synapses, vesicle utilization during intense synaptic activity and the effects on synapsin phosphorylation. Our results indicate that ATP-binding by synapsin IIa is profoundly significant for synapsin-dependent synaptic plasticity and for defining the resting structure of the presynaptic terminal. Because the mitochondria in neuronal cells provide most of the needed cellular energy in the form of ATP, we were also interested in checking whether the position of the mitochondria in relation to the presynaptic terminal could alter SV clustering in a synapsin-dependent way, as synapsins are ATP binding proteins and the binding of ATP modulates their activity. Our results suggest that mitochondria influence SV clustering and that this effect is synapsin-dependent. 2. Does mitochondrial calcium efflux by NCLX affect presynaptic properties? Does this effect depend on synapsin? Binding of ATP by some of the synapsins is regulated by cytoplasmic calcium levels. In order to investigate the role of synapsins in synaptic function, we studied the effect of mitochondrial calcium efflux mechanism via NCLX on the properties of synapses and its dependence on the synapsins. We measured clustering and mobility of synaptic vesicles and their division into functional pools in NCLX deficient neurons. Furthermore, we examined the effect of deleting NCKX on the expression of endogenous synapsins. Our results indicate that NCLX affects basal presynaptic

36 properties and that this effect could be mediated by altering the expression of synapsins. 3. How does NCLX deletion impact synaptic activity and plasticity? Many studies emphasize that mitochondria are essential for both synaptic activity and plasticity, because they are required not only for the supply of energy but also for the regulation of cytoplasmic calcium concentration. We decided to investigate the role of NCLX in mitochondrial and presynaptic calcium homeostasis, by imaging free calcium in the mitochondrial matrix and in the cytoplasm of presynaptic terminals in resting neurons as well as during the induction of synaptic activity. We examined how deletion of NCLX affects synaptic release and synaptic plasticity, short-term (facilitation) and long-term plasticity. Our results indicate that NCLX is an important player in regulating mitochondrial calcium homeostasis and consequently cytoplasmic calcium, and that it thus plays a key role in regulating synaptic activity and plasticity. 4. Do synapsins affect synaptic ATP levels? Finally, we measured ATP during neuronal activity and whether synapsins have an effect of local ATP changes during resting conditions and during neuronal activity. Synapsins have been shown to bind ATP but the role of this fact is still unknown. The homologous C domain that binds ATP is similar to that of an ATP-utilizing enzymes, such as glutathione synthetase (Esser et al. 1998; Hosaka and Südhof 1998a), but the function of this domain remains unclear. We examined whether they impact the maintenance of resting ATP levels within presynaptic terminals.

Research significance

To this date, there is still debate in the scientific community over the importance of mitochondrially-derived ATP and local calcium buffering properties in axonal processes. In this work we attempt to shed more light on the importance of mitochondrial calcium, specifically calcium efflux mechanism, in basal synaptic properties, activity and plasticity. My results expand our knowledge on how mitochondrial calcium overloads may occur, a phenomenon linked to many neurodegenerative disorders, and on how this mitochondrial dysfunction may affect synaptic release. We also investigated the role of ATP-binding to synapsins and conversely, the role of synapsins as possible ATP-buffering proteins. This is particularly interesting because synapsins are abundant presynaptic proteins and have been functionally linked to various pathologies. By introducing new methods and tools to the lab I was able to investigate these questions.

Materials and Methods

7 Solutions, materials and antibodies

Table 1: Reagents and chemical compounds

Material Company Carbonyl cyanide-4-(trifluoromethoxy) Santa Cruz Biotechnology phenylhydrazone (FCCP) Oligomycin Sigma-Aldrich, Rehovot, Israel 2-Deoxy-d-Glucose Sigma-Aldrich, Rehovot, Israel Skim milk, powdered Sigma-Aldrich, Rehovot, Israel Dulbecco's modified eagle medium (DMEM) Biological Industries, Beit-Haemek, Israel L-Glutamine Biological Industries, Beit-Haemek, Israel Penicillin Streptomycin Biological Industries, Beit-Haemek, Israel Sodium pyruvate Biological Industries, Beit-Haemek, Israel MEM Eagle (non-essential amino acids Biological Industries, Beit-Haemek, Israel solution) Trypsin/EDTA Biological Industries, Beit-Haemek, Israel Neurobasal-A medium Thermo-Fisher Scientific, Waltham, MA Fetal Bovine Serum (FBS) Biological Industries, Beit-Haemek, Israel B-27 supplement Thermo-Fisher Scientific, Waltham, MA Glutamax Thermo-Fisher Scientific, Waltham, MA Gentamicin Biological Industries, Beit-Haemek, Israel Papain Worthington, Lakewood, NJ Poly-D-Lysine (PDL) Sigma-Aldrich, Rehovot, Israel 12mm #1 glass Coverslips Bar Naor Ltd, Ramat Gan, Israel DL-2-Amino-5-phosphonopentanoic acid Sigma-Aldrich, Rehovot, Israel (APV) 6,7-Dinitroquinoxaline-2,3(1H,4H)-dione Alomone Labs, Jerusalem, Israel (DNQX) (+)-bicuculline Alomone Labs, Jerusalem, Israel (+)-MK801 Sigma-Aldrich, Rehovot, Israel Tetrodotoxin (TTX) Alomone Labs, , Jerusalem, Israel

Hank’s Balanced Salt Solution Biological Industries, Beit Haemek, Israel HEPES Biological Industries, Beit Haemek, Israel Bafilomycin A Enzo Life Sciences, Farmingdale, NY N-[2-[[3-(4-Bromophenyl)-2- Enzo Life Sciences, Farmingdale, NY propenyl]amino]ethyl]-5-isoquinoline- sulfonamide (H89) Picrotoxin Enzo Life Sciences, Farmingdale, NY T4 ligase Thermo Fisher Scientific, Waltham, MA Restriction enzymes Thermo Fisher Scientific, Waltham, MA Q5 polymerase New England Biolabs, Ipswich, MA Benzonase Thermo Fisher Scientific, Waltham, MA Saponin Fluka, Charlotte, NC ATP Sigma-Aldrich, Rehovot, Israel Para-formaldehyde (PFA) EMS, Hatfield, PA Immumount Thermo Fisher Scientific, Waltham, MA Isoflurane Piramal Critical Care, Bethlehem, PA Agarose Sigma-Aldrich, Rehovot, Israel Glacial Acetic Acid Bio-lab ltd, Jerusalem, Israel Rat tail collagen BD biosciences, San Jose, CA Deoxyribonuclease I (DNAse I) Sigma-Aldrich, Rehovot, Israel Cytosine β-D-arabinofuranoside Sigma-Aldrich, Rehovot, Israel hydrochloride (Ara-C) ADVASEP-7 CyDex, Lenexa, KS Boehringer Complete Protease Inhibitor Roche Molecular Biochemicals, Basel, Mixture Switzerland Phosphate buffered saline; PBS Biological Industries, Beit-Haemek, Israel Borosilicate glass pipettes Sutter Instruments, Novato, CA

Table 2: Fluorescent dyes

Material Company Tetramethylrhodamine methyl ester Sigma-Aldrich, Rehovot, Israel (TMRM) Synaptogreen C4 (FM1-43) Biotium, Hayward, CA

Table 3: Solutions

Name Composition (in mM)

Standard extracellular NaCl 150, KCl 3, Glucose 20, HEPES 10, CaCl2 2, MgCl2 2, pH saline adjusted to 7.35 with NaOH, 315 mOsm ATP-depleting saline the same as standard extracellular saline except 20 mM glucose was replaced with 2-deoxy-D-glucose and 1 µm oligomycin was added

Intracellular solution for potassium gluconate 113, NaCl 15, MgCl2 6, HEPES 20, EGTA 10 permeabilized cells

Hyperkalemic extracellular NaCl 63, KCl 90, Glucose 20, HEPES 10, CaCl2 2, MgCl2 2, pH solution adjusted to 7.35 with NaOH, 315 mOsm

Hypercalcemic NaCl 150, KCl 3, Glucose 20, HEPES 10, CaCl2 5, pH adjusted to extracellular solution 7.35 with NaOH, 315 mOsm

Alkalinizing extracellular NaCl 100, KCl 3, NH4Cl 50, Glucose 20, HEPES 10, CaCl2 2, MgCl2 solution 2, pH adjusted to 7.35 with NaOH, 313 mOsm

Intracellular solution for potassium gluconate 113, NaCl 15, MgCl2 6, HEPES 20, EGTA 2, autaptic recordings Na2ATP 5, NaGTP 0.3, pH adjusted to 7.3 with KOH, 295 mOsm

ACSF solution 124 NaCl, 3 KCl, 2 CaCl2, 2 MgSO4, 1.25 NaH2PO4, 26 NaHCO3,

and 10 glucose; pH 7.4 when bubbled with 95% O2/CO2

Sucrose based ACSF 252 Sucrose, 5 KCl, 1 CaCl2, 3 MgSO4, 26 NaHCO3, 1.25 NaH2PO4 solution and 10 Glucose; pH 7.3 when bubbled with 95% O2/CO2

Tyrode's solution NaCl 145, KCl 3, Glucose 15, HEPES 10, CaCl2 1.2, MgCl2 1.2, pH adjusted to 7.4 with NaOH WB Lysis buffer Ca2+/Mg2+ PBS supplemented with 0.1% Triton X-100a and 1:50 Boehringer Complete Protease Inhibitor Mixture

Table 4: Antibodies for immunocytochemistry

Name Company Dilution Primary Antibodies Rabbit polyclonal anti-Synaptobrevin 2 Synaptic Systems 1:600 Guinea pig polyclonal anti-Synaptophysin I Synaptic Systems 1:1000

Rabbit polyclonal anti phospho-S9 synapsin I Phosphosolutions 1:200 Mouse monoclonal anti Glutamic acid Developmental Studies 1:1000 decarboxylase of 65 KDa (GAD-6) Hybridoma Bank, DSHB Goat polyclonal anti vesicular glutamatergic Synaptic Systems 1:1000 transporter 1 (vGlut 1) Mouse monoclonal anti Synapsin 1 Synaptic Systems 1:1000 Rabbit polyclonal anti Synapsin 2 Synaptic Systems 1:1000 Rabbit polyclonal anti β3-tubulin Synaptic Systems 1:1000 Secondary Antibodies Donkey anti-mouse IgG labeled with Northern R&D Systems 1:1000 Lights 637 donkey anti-rabbit IgG labeled with Northern R&D Systems 1:1000 Lights 557 Affinity-purified goat anti-rabbit IgG, labeled Jackson ImmunoResearch 1:100 with Cy3 Laboratories goat anti-guinea pig labeled with DyLight 649 Jackson ImmunoResearch 1:1000 Laboratories Donkey anti goat IgG, labeled with AlexaFluor Abcam 1:1000 647

8 Mice

Synapsin triple knockout (TKO) mice (Gitler et al., 2004a), back-crossed onto the C57Bl6 background (Boido et al. 2010), were grown at the Ben-Gurion University mouse facility. NCLX knock-out (NCLX-KO; Slc8b1em1J) mice, back-crossed onto the C57BL6 background, were generated and obtained from Jackson Laboratory (Bar Harbor, ME). C57BL6 wild-type (WT) controls were obtained from Harlan Laboratories (Ein Kerem, Israel). Animals were treated in accordance to the guidelines of the Ben- Gurion University Institutional Committee for Ethical Care and Use of Animals in Research.

9 Neuronal cultures

9.1 Primary dense hippocampal cultures

Primary dense hippocampal cultures from P0-P2 pups of either sex were performed essentially as described previously (Gitler et al. 2004b; Tevet and Gitler 2016). Briefly, postnatal day 0-2 pups were decapitated, the brains quickly removed, hippocampi were dissected, sliced manually, and kept on ice in Hank’s Balanced Salt Solution supplemented with 20mM HEPES (termed HBSS) at pH 7.4. Hippocampus pieces were incubated for 20 minutes at room temperature (RT) in digestion solution consisting of

5ml HBSS, CaCl2 1.5mM, EDTA 0.5mM and 100 units of papain activated with cysteine. The brain fragments were then triturated gently two times using fire-polished Pasteur pipettes. Each time, dispersed cells were collected from the suspension after allowing debris to sink by gravity. Cells were counted using a hematocrit chamber, dyeing dead cell with trypan blue. Cells were seeded at a density of 80,000-100,000 cells per well on 12mm #1 glass coverslips (CS) coated with poly-D-Lysine. Initially, cells were plated in plating medium consisting of Neurobasal-A medium supplemented with 2% B27, 2 mM Glutamax I, 5% defined FBS and 1 μg/ml gentamicin. After 24 hours, the plating medium was replaced with serum-free culture medium that consisted of Neurobasal-

A, 2 mM Glutamax I and 2% B27. Cultures were maintained at 37°C in a 5% CO2 humidified incubator for 12-15 days prior to staining and imaging, or 10-14 days for electrophysiological recordings.

9.2 Autaptic cultures

Autaptic hippocampal cultures were prepared as described above. Coverslips were prepared in advance, placed in a Petri dish covered with a sheet of parafilm and were covered with 0.15% agarose in sterile double distilled water (DDW) and dried overnight in a biological hood. In the following day, coverslips were sprayed using an atomizer with a substratum of 480 g/ml poly-D-Lysine, 20mM glacial acetic acid and 465 g/ml rat tail collagen diluted in sterile DDW. After the spraying of micro islands, coverslips were dried overnight and kept for a few weeks. Microisland coverslips were suspended above feeder glial cultures grown in 12 wells plates 7-10 days prior to the neuronal culture in order to support the autaptic neurons. The neurons were plated

43 at a density of 40-45 thousand cells per well. To form raised bumps on which to place the coverslips, the plastic at the bottom of the well was melted gently in a few places with a clean welding iron. Glial cultures were prepared from the cortex similarly to the neuronal cultures, except 2.5mg Deoxyribonuclease I that was added to the digestion solution. 100 µM of Cytosine β-D-arabinofuranoside hydrochloride (Ara-C) was added at 5-7 days in vitro (DIV) to prevent glial overgrowth.

10 Acute hippocampal slices

P18-P21 mice from either sex were anesthetized with Isoflurane followed by decapitation. Their brain was rapidly removed and placed into ice-cold oxygenated sucrose-based ACSF. Transverse slices (300 µm) were cut on a vibratome (Leica) and placed into a holding chamber containing oxygenated ACSF at room temperature. Sliced were incubated for at least an hour prior to the recordings.

11 Molecular constructs and viral preparations

Neurons were infected using adeno-associated viral particles (AAV) carrying cDNA inserts. Most cDNAs encoding proteins of interest were coupled to a genetically- encoded fluorescent protein, which served as a reporter for verification of viral infection. This procedure is used routinely in the lab and is described thoroughly in Tevet and Gitler, 2016. Briefly: cDNAs of interest were subcloned by restriction/ligation into a plasmid containing adeno-associated virus 2 (AAV2) inverted terminal repeats flanking a cassette consisting of the 1.1 kb cytomegalovirus enhancer/chicken β-actin promoter or the 0.47 kb human synapsin 1 promoter (hSyn, Kügler et al., 2003), the woodchuck post-transcriptional regulatory element and the bovine growth hormone polyA signal. PCR assisted site-directed mutagenesis to produce the synapsin mutations was performed using mutagenesis primers, followed by digestion using the restriction enzyme DpnI. Final products were verified by sequencing. Viral particles were produced in HEK cells using both the pD1 and pD2 helper plasmids (Groh et al. 2008), which encode the rep/cap proteins of AAV1 and AAV2, respectively.

Primary cultures of hippocampal neurons were infected at 5 DIV and incubated for at least 7 days before imaging. Virus titer was set to produce 75-90% infection efficiency.

EGPF, ECFP, EYFP and DsRed2-Mito were from Clontech (Mountain View, CA); TagBFP from Evrogen (Moscow, Russia); Venus was a kind gift from Atsushi Miyakawi (Nagai et al., 2002); ATeam1.03, ATeam1.03NL and ATeam1.03R122K/R126K were a kind gift from Hiromi Imamura (Imamura et al. 2009; Tsuyama et al. 2013). SypHy-2X was from Leon Lagnado. mCherry from Roger Tsien; Synaptophysin I was from Thomas Kuner and GCaMP6f was from Ofer Yizhar. 2MT-GCaMP6m was a kind gift from Israel Sekler.

12 Western blot analysis

HEK-T cells were transfected with EGPF-SynIIa or EGFP-SynIIa-K270Q expression plasmids using the calcium-phosphate precipitation method. After 24 hours, cells were trypsinized, washed twice in PBS, resuspended in 300µl of lysis buffer, and homogenized. The homogenate was cleared by centrifugation at 1500 xg for 5 min at 4°C and the supernatant was frozen and stored at -80°C for future use. Protein concentrations were determined using the Bradford assay (Bio-Rad). Equal protein quantities were subjected to SDS-PAGE gel electrophoresis and immunoblot analysis was performed using an anti-synapsin II antibody (1:2500) PBS/5% skim milk, followed by incubation with a secondary anti-rabbit-IgG antibody coupled to HRP (1:5000). Detection was done with the EZ-ECL Chemiluminescence Detection kit for HRP (Biological Industries) on an ImageQuant LAS 4000 digital imaging system (General Electric Health- care Bio-Sciences).

13 Immunocytochemistry

12-14 DIV cultured neurons were fixed with 4% para-formaldehyde (EMS, Hatfield, PA) in PBS for 10 minutes, rinsed with PBS, permeabilized with 0.1% Triton X-100 in PBS for 2 minutes, blocked with 5% powdered skim-milk in PBS for 1 hour, rinsed, incubated with the primary antibody for 1 hour, rinsed, incubated with the secondary antibody for 1 hour, rinsed, and mounted in immumount. All steps were performed at RT.

14 Fluorescence microscopy

Fluorescence measurements were performed on a Nikon TiE inverted microscope driven by the NIS-elements software package (Nikon). The microscope was equipped with an Andor sCMOS camera (Oxford Instruments), a 40X 0.75 NA Super Fluor objective, a 60X 1.4 NA oil-immersion apochromatic objective (Nikon), a perfect-focus mechanism (Nikon), and EGFP, EYFP, and Cy3 TE-series optical filter sets (Chroma) as well as BFP and Cy5 filter sets (Semrock). For live imaging experiments, images were acquired using either regular triggered acquisition or the fast time lapse acquisition mode of our camera (for calcium imaging). The fast time lapse acquisition mode allows streaming the images to the on-camera memory chip followed by readout to the computer. Thus, the actual frame rate depends on the exposure time, binning, readout mode, sensor mode and the readout speed of the camera, all of which are computer-processing independent. This enables faster imaging with a very accurate inter-image interval, specifically suited for fast calcium fluxes.

15 Neuronal culture field-stimulation

Cultured neurons on CSs were placed in a stimulation chamber (RC-49MFSH, Warner Instruments) and stimulated at an intensity of 10 V/cm using a stimulus isolation unit (SIU-102B, Warner Instruments) monitored by an oscilloscope (TDS 1001B, Tektronix). Stimulation duration and frequency was manipulated by an isolated pulse stimulator (2100, A-M Systems).

16 Semi-quantitative synaptic immunofluorescence

To allow semi-quantitative comparison of immunostaining intensity of synapses, experimental and control groups were processed under identical conditions using the same reagents. Synapses were detected semi-automatically using an in-house iterative algorithm based on serially decreasing thresholds as described previously (Bergsman et al. 2006) and implemented in NIS-elements software (Nikon). Fluorescence values for each synapse were obtained from an area of 2 X 2 pixels located on its center of mass and an image average was generated. Because intensity values of the controls can vary from session to session, a normalization value was

46 determined from the control experiments of each session and this value was used to normalize all images acquired during that session. The normalized values were then either averaged or used to compute cumulative distributions. Puncta that were positive for GAD65 were deemed GABAergic, since GAD (gamma-amino decarboxylase) is a vesicular enzyme in charge of converting glutamate to GABA (Kanaani et al. 1999); otherwise, they were considered glutamatergic (the majority (>80%) of the neurons in the hippocampus). Experiments were performed on at least three independent cultures.

17 Synaptic 'Clustering Index'

The clustering index (CI) of synapsin was assessed from neurons expressing EGFP- tagged synapsin, which where immunostained for Syb2 and GAD65. Glutamatergic and GABAergic synapses were analyzed separately. After correcting for background, the Syb2 fluorescence in each synapse was plotted as a function of the EGFP fluorescence, and the resulting graph was fit with a linear function. The slope of the function was defined as the per-image clustering index. Clustering indices per condition were compared across images.

18 Synapse width analysis

Synapse width was measured by drawing a line starting in the axon and through the synapse punctum, and fitting it using a Gaussian function as follows (Orenbuch et al. 2012a):

2 (푥−푥푐) − 2 (1) 푦 = 푦0 + 퐴푒 2푤

Where xc is the center of the punctum maximum, y0 is the fluorescence of the axon, w2 is the variance of the Gaussian, and A is its amplitude. The full-width at half- maximum (FWHM) fluorescence intensity is calculated as follows:

(2) 퐹푊퐻푀 = 2푤√ln⁡(4)

Fitting was performed by the least-squares error method using Origin (OriginLab). Independent images were acquired from cells grown on various coverslips obtained

47 from a minimum of three different cultures. The mean FWHM was determined for each image, and then a grand average was calculated for each condition.

Figure 3 - Measuring synaptic width. Top - illustration of a synapse segment with a line drawn starting in the axon, through the synapse punctum and onto an empty background area outside the neuron. Bottom - fluorescence line-profile of the synapse (black symbols) and Gaussian fit (blue line).

19 Calculation of synaptic targeting factor

The targeting index of synapsin (Gitler et al. 2004a) was measured per-synapse in neurons expressing an EGFP-tagged synaptic proteins and soluble TagBFP. Fluorescence was measured in the synaptic puncta (Syn) and in the axon. The ratio between synaptic and axonal value in the green channel (g) describes the enrichment of synapsin in the synapse, while the ratio in the blue channel (b) corrects for volume differences in these two locales. Lack of targeting is expressed as 0 by subtracting 1 from the result. Values larger than 0 - indicate that a protein is enriched in the synapse.

푆푦푛 /퐴푥표푛 (3) 푇퐹 = 푔 푔 − 1 푆푦푛푏/퐴푥표푛푏

The fluorescence values were measured by stretching a line profile through each synapse and fitting it with a Gaussian function centered on the synapse:

2 (푋−푋푐) − 2 (4) 퐺(푥) = ⁡ 푦0 + 퐴푒 2푤

2 Where y0 is the fluorescence of the axon, xc is the center of the punctum, w is the standard deviation (width) of the Gaussian, and A is the fluorescence attributed to the

48 vesicle cluster within the synapse (corrected for the imaging background). Fitting was performed by the least-squares error method using Origin (OriginLab). The synaptic fluorescence was calculated as the mean signal within the line segment Syn=Xc±w, while the fluorescence of the axon was the mean signal of the line segment Ax

20 Mitochondrial membrane potential measurements

TMRM was used in non-quenching mode to measure pre-existing ΔΨm. Cells were loaded with 25nM TMRM for 15 min at RT while gently shaking. For imaging of basal TMRM fluorescence, cells were transferred into standard extracellular solution containing TMRM, as well as APV and DNQX. After imaging baseline levels in each experiment, 5 μM FCCP was added to trigger full depolarization of the mitochondria. Basal fluorescence was normalized by the fluorescence measured after FCCP application (FFCCP) in each experiment. All the results were normalized by the F/FFCCP value of WT neurons. To convert changes in fluorescence to mV, the previously defined formula was used (Plášek et al. 2005; Valero et al. 2008):

(퐹퐾푂−퐹[퐾푂−퐹퐶퐶푃]) (5) ∆훹푥 = 58.7⁡푙표푔 (퐹푊푇−퐹[푊푇−퐹퐶퐶푃])

Where: ΔΨx- change in Ψm (in millivolts) from the control to the KO condition; FKO – basal TMRM fluorescence in NCLX-KO neurons; F[KO-FCCP] - TMRM fluorescence in NCLX- KO neurons after FCCP application (same for WT).

21 SypHy: measuring vesicle cycling

SypHy is a probe based on the internal fusion of a pH-sensitive GFP called pHluorin (pKa=7.6, Tanida et al. 2014), with the vesicular protein Synaptophysin I (SypI), so that pHluorin is located in the lumen of the SVs, and allows the reportage of presynaptic activity (Burrone et al. 2007; Sankaranarayanan et al. 2000). Briefly, pHluorin fluorescence is quenched by the acidic pH of the intact vesicle lumen (pH~5.5). Upon stimulation, SVs fuse with the plasma membrane, thus exposing their lumen to the neutral pH of the extracellular environment (pH~7.3), causing an increase in the

49 fluorescence of pHluorin. After endocytosis of the vesicle, pHluorin is requenched due to reacidification of the lumen by the vesicular proton pump (vATPase). Therefore, an increase in fluorescence reflects the exocytosis phase, and the subsequent decrease in fluorescence indicates the endocytosis phase of the vesicle cycle. The kinetics of exocytosis on its own can be determined by examining the time course of the rise in fluorescence upon stimulation in the presence of bafilomycin A, an inhibitor of the vATPase (Sankaranarayanan and Ryan 2000a). Importantly, blocking the vATPase does not stop vesicle recycling, only reacidification and neurotransmitter loading (Cousin et al. 1997; Zhou et al. 2000); this means that SV recycling by bafilomycin-treated neurons still reflects physiological rates. In our experiments, 12-14 DIV neurons were field-stimulated in the presence of APV and DNQX, with or without bafilomycin. After stimulation, the bath was perfused with saline in which 50mM NaCl was replaced with

NH4Cl. Ammonium ions are in equilibrium with aqueous ammonia, which is membrane permeable, and can thus diffuse into SVs to neutralize their lumen. Thus, the combination of SypHy and NH4Cl reveals intact SVs within the terminals, and the size of the total pool of SVs is measurable. The recycling pool size is estimated by measuring the plateau of fluorescence intensity obtained during exhaustive stimulation in the presence of bafilomycin, while the resting pool is that which completes the RcP to the total pool (Burrone et al. 2007; Kim and Ryan 2010).

The baseline fluorescence intensity of SypHy (F0) in each synapse of interest was the average value measured in 5 successive images acquired before stimulation. The change in fluorescence (F) at time t was calculated as F(t)-F0. Values for each synapse were either normalized by the maximal fluorescence intensity (Fmax) measured after the treatment with NH4Cl (F/ Fmax) to assess rate or the RcP fraction, or normalized by baseline fluorescence F0 to assess relative change in release.

22 Calcium imaging

22.1 Mitochondrial calcium

Neurons infected with hSyn:2MT-GCaMP6m (MitoGCaMP6m) were imaged during field-stimulation in the presence of APV (50µM) and DNQX (10µM) to block recurrent

50 network activity. Images were acquired at an acquisition rate of ~17Hz using the fast time lapse acquisition mode of our camera with 3 X 3 binning.

The baseline fluorescence intensity of MitoGcaMP6m (F0) in each mitochondria of interest was the calculated average value measured in 20 successive images acquired before stimulation. The change in fluorescence (F) at time t was calculated as F(t)-F0. A mean trace was determined for each experiment and then a grand average was calculated for each condition.

22.2 Presynaptic cytoplasmic calcium measurements

Neurons infected with hSyn:SypI-mCherry-GCaMP6f were imaged during field- stimulation in the presence of APV and DNQX to block recurrent network activity. An image of the red channel (R; mCherry) was taken before the stimulation. We verified in separate control experiments that the mCherry value is not affected by stimulation and does not noticeably change throughout the experiment. mCherry fluorescence values for each synapse were obtained from an area of 3 X 3 pixels located on its center of mass. During the stimulation phase, the green channel was imaged (G; GCaMP6f), and the GCaMP6f fluorescence was measured for each synapse in the same area as mCherry. Images of GCaMP6f were acquired at an acquisition rate of ~37Hzusing the fast time lapse acquisition mode of our camera at 3 X 3 binning. The G/R ratio was calculated for each synapse in each image in the time-lapse as the background-subtracted fluorescence of GCaMP6f divided by the background- subtracted fluorescence of mCherry.

23 FM dye loading and unloading

Neurons were loaded with FM1-43 at a final concentration of 10M by depolarizing them for 2 minutes using hyperkalemic saline in the presence of the dye. SVs of the RcP were exocytosed, and during endocytosis, their equivalent was labeled with the dye. The neurons were exposed to the dye in normal saline for an additional 5 minutes after depolarization, to allow labeling related to endocytosis to be completed (Ryan et al. 1993). Afterwards, neurons were washed with normal saline for 5 minutes, followed by 5 minutes washing with 1 mM ADVASEP-7 in normal saline. ADVASEP is used as a carrier to remove dye that is nonspecifically bound to the outer leaflet of

51 the plasma membrane or to other extracellular surfaces, thus reducing background staining (Kay et al. 1999). Finally, the cells are washed in normal saline and imaged to assess loading. For the unloading, cells were stimulated and images were acquired before and during stimulation. The degree of unloading was calculated as the change in fluorescence

(F) normalized by baseline background-corrected fluorescence F0. Experiments were performed at room temperature in the presence of APV (50M) and DNQX (10M) to reduce destaining due to spontaneous network activity.

Figure 4 – FM1-43 dye experiment. (a) Synaptic vesicles near the plasma membrane. (b) FM dye is added, binds to the outer membrane, and becomes fluorescent. (c) The preparation is stimulated and a vesicle fuses with the plasma membrane exposing the luminal membrane to the FM dye. (d) The vesicle is endocytosed with FM dye inside. (e) The FM dye is washed out of the bath and a labeled vesicle is imaged. (f) The preparation is stimulated again in a dye-free medium and vesicles exocytosis is measured as dye leaves the fusing vesicle. Adapted from Gaffield and Betz, 2006.

24 Fluorescence Recovery After Photobleaching (FRAP)

Photobleaching of SypI-EGFP was performed using a Zoom-FRAP protocol using the EZC1 software package on the confocal microscope. The synapses were exposed three times with a 488nm laser beam set to 8% power. The laser beam was focused on three different synapses in each experiment, without significantly affecting neighbouring synapses. Each bleaching session consists of capturing 5 pre-bleaching images, followed by zooming-in by a factor of 16 on the three selected synapses, each scanned three times with the laser set at 35% power, resulting in over a 1000 fold increase in power. Fluorescence recovery was monitored in time-lapse mode (18 images at 0.2 Hz and then 26 images at 0.06 Hz). Focus was adjusted manually to correct for focus drift,

52 when necessary. Experiments were performed in the presence of APV and DNQX in order to prevent spontaneous neural network activity. Fluorescence intensity values were measured from the bleached synapses and from adjacent empty areas (to obtain background values), as well as from three additional unbleached synapses. The florescence of each bleached synapse was normalized by the mean intensity of the unbleached synapses, in order to correct for whole-imaging bleaching and for intensity variations caused by focal drift. For each synapse, the average fluorescence value during the prebleached imaging was scaled to 1 and the averaged value immediately after the photobleaching was scaled to 0, since there was no correlation between the FRAP rate and either the depth of bleach or the initial fluorescence value of each bleached synapse (Orenbuch et al., 2012a). For each experiment, recovery-fraction traces were calculated as the average of the three bleached synapses. A grand average was calculated across experiments. Images were acquired using a Nikon C1Si spectral confocal microscope mounted on a Nikon FN1 microscope. A 1.0NA 60X water dipping objective (Nikon, Japan) was used.

Figure 5 – Schematic illustration of a FRAP of whole presynaptic terminals. Labeled vesicles (green circles) are transported in and out of presynaptic terminals through the axon. Fluorescence is bleached in a single presynaptic terminal (large blue circle). Thereafter, transport continues, so that intact vesicles enter the terminal and bleached vesicles (black circles) leave it. Adapted from Staras et al., 2013.

25 Fluorescence resonance energy transfer (FRET) imaging and analysis

25.1 The sensitized emission method (Ben-Gurion University)

Imaging of primary hippocampal neuronal cultures expressing SypI-ATeam1.03 or SypI-ATeam1.03NL was performed on a C1si spectral confocal microscope with the

53 same microscope and analysis parameters as previously described in the FRAP experiments.405 nm and 442 nm solid-state lasers were used as excitation light sources for ATeam1.03. Emission spectrums were measured at a spectral resolution of 5nm, in the 430-590 nm range (32 spectral bins) and the 450-610 nm range (32 spectral bins) for 405 nm and 442 nm excitation, respectively. 3 images of 512 X 512 pixels at 8 µs/pixel were averaged to reduce noise (Levy et al. 2011). Spectra (acquired at 12 bit resolution) were analysed by spectral unmixing against fluorescence references measured separately in neuronal cells expressing exclusively the donor

(mseCFP), the acceptor (cpmVenus) or mitoDsRed2. Emission quantification was established as the integral of the relevant spectrum. Unmixing was performed using an in-house analysis package written in Origin (OriginLab). Briefly, the emission spectra were fit by the least-squares method with the premeasured fluorescence spectra of the donor and acceptor, of mitoDsRed2, and that of the autofluorescence of the cells and the background fluorescence of the media. The FRET ratio (FR) was calculated as Acceptor/(Acceptor+Donor).

25.2 Acceptor-photobleaching method (Tel-Aviv University)

FRET determination by acceptor photobleaching was performed as described previously (Laviv et al. 2010). Briefly, the donor was excited at 440 nm and its emission was measured at 460–500nm before (IDA) and after (ID) acceptor photobleaching. Excitation was delivered to the acceptor at 514 nm, and emission was measured at 530–600 nm. Photobleaching of cpmVen was performed with the 514 nm laser line, by a single point activation module for rapid and efficient multi-region bleaching. The

FRET efficiency, Em, was calculated thus:

퐼퐷퐴 (6) 퐸푚 = 1 − 퐼퐷

25.3 FRET ratio calculation (Tel-Aviv University)

The ATeam1.03 FRET ratio (FR) was measured for each region of interest (ROI) at four consecutive times: before stimulation, at the completion of stimulation (10 Hz, 90 sec), and 3 and 10 min after its cessation. Intensity of the fluorescence emission excited with the 440 nm laser was measured simultaneously for mseCFP and cpmVenus at 460–500 nm (ID) and 530–600 nm (IA), respectively. Each value was

54 corrected for the background intensity measured from an ROI placed in close proximity to the bouton. In addition, images of the acceptor emission were obtained with the 514 nm laser excitation before and after stimulation. Only ROIs that exhibited <15% change in the intensity of directly excited cpmVen were included in the analysis.

Because 40% of IA originates from mseCFP, FR was calculated using a correction for the tail of mseCFP’s emission as follows: 퐼 −0.4∗퐼 (7) 퐹푅 = 퐴 퐷 퐼퐷

Analysis was performed using custom-written scripts in MATLAB as described previously (Laviv et al. 2010).

25.4 ATeam1.03-SypI calibration

Primary hippocampal cultures from WT mice were infected with SypI-ATeam1.03 or SypI-ATeam1.03NL. The cells were submerged into ATP-free sodium-based intracellular solution and permeabilized using 0.001% of saponin. Cells then were washed with the same solution and specific ATP concentrations were added. 405nm laser was used to measure FR.

26 Electrophysiology

26.1 Electrophysiological recordings from autaptic cultures

Electrophysiological measurements where performed in the whole-cell voltage-clamp configuration at a holding potential of −70 mV using the EPC-10 amplifier and PatchMaster software (HEKA). Data were sampled at 10 kHz and filtered at 3 kHz. Borosilicate glass pipettes were pulled using a P-97 horizontal puller (Sutter Instruments) and were fire polished to a resistance of 3–5 MΩ. Recordings were used if the access resistance was <25 MΩ. Synaptic activity was evoked by inducing action potentials by applying a step depolarization command of 0.5 ms to +60 mV. Stimulation trains (90sec) were delivered at 10 Hz. Miniature synaptic currents were analyzed using Clampfit (Molecular Devices). Glutamatergic and GABAergic neurons were differentiated by the kinetics of the synaptic currents (Gitler et al., 2004b), which was initially verified by the use of DNQX and picrotoxin, respectively.

26.2 Field-potential recordings

Extracellular stimulation was performed with a stimulus isolation unit ISO-Flex (A.M.P.I) using glass monopolar electrodes (0.5–1 MΩ) filled with ACSF. fEPSPs were recorded in current-clamp mode using MultiClamp 700B (Molecular Devices) with ACSF-filled patch pipettes (0.5–1 MΩ). The stimulating electrode was placed in the CA3 area and the recording electrode in the dendritic CA1 area. 5 consecutive stimulations were delivered at 5, 10, 20 and 50 Hz. 5-10 sweeps were conducted for each stimulus frequency and recordings were averaged over trials. Data was sampled at 10 kHz, amplified (gain 5), filtered at 3 kHz then digitized and analyzed using Clampfit. Recordings were acquired using MultiClamp 700B and Clampex software (Molecular Devices). Experiments were performed at 29 ± 1 °C with flow rates of 2 ml per minute. Data was included if the axon volley to fEPSP ratio was within the linear range and if there was an increase in the second pulse.

26.3 Probability of release

To record NMDAR-fEPSPs, magnesium was excluded from ACSF to relieve the block of NMDA receptors by extracellular magnesium ions. Slices were incubated for at least 30 min in magnesium-free ACSF containing Bicuculline (100 μM) and DNQX (5 μM) to block inhibitory inputs and AMPA receptor-mediated fEPSPs, respectively. Stimulation was conducted every 15 seconds for 5 min to obtain a baseline response. Stimulation was halted for 10 min while 40 μM (+)-MK801 was added and allowed to equilibrate. Recordings then were conducted for 6 additional minutes until 90% of the response was inhibited. Data was normalized so that the baseline amplitude was set as 1 and the final fEPSP amplitude recorded after MK801 application was set as 0. The rate of fEPSP(NMDA) decline was quantified by calculating the t½ of the decay traces. Recordings that showed epileptiform activity were excluded from the analysis.

26.4 Long-term plasticity induction via high-frequency stimulation

A 15 minute baseline period preceded long-term potentiation (LTP) induction using stimulation intervals of 15sec. Recordings were continued for at least 30 min following high-frequency stimulation (HFS) induction protocol: 5 pulses at 100 Hz, repeated

56 again after 20sec. LTP was quantified by comparing the mean fEPSP amplitude recorded 20-25 minutes after delivery of the HFS. In every recording, the fEPSP amplitude was normalized to the baseline period. Traces were excluded from the analysis if the baseline was not stable.

27 Statistics

Values are reported throughout as mean ± SEM. Comparisons of two datasets were performed by the Student’s t-test, after confirming a normal distribution by the Shapiro-Wilk normality test, or with the Mann-Whitney’s non-parametric test otherwise. Multiple comparisons of normally distributed datasets were performed using one-way ANOVA followed by Tukey's post-hoc analysis. Comparison of time dependent processes (FRAP, cumulative charge transfer, frequency facilitation) were performed using repeated-measures ANOVA. Comparison of multiple datasets whose distribution was not deemed normal was performed with the Kruskal-Wallis test, followed by Dunn's post-hoc test. Significance was set at a confidence level of 0.05. In all figures ns denotes p>=0.05, * denotes p<0.05, ** p<0.01, and *** p<0.001. Statistical analysis was performed with Origin, Graphpad Prism (La Jolla, CA), or SPSS 18 (IBM, Somers, NY).

Results

1 ATP binding to Synaspsin IIa regulates usage and clustering of vesicles in terminals of hippocampal neurons

Most results in this chapter have been published (Shulman*, Stavsky* et al., 2015).

1.1 Mutating an ATP-binding site in synapsin IIa does not disrupt its structure

Synapsins are ATP-binding proteins (Esser et al. 1998; Hosaka and Südhof 1998b). However, the physiological consequences of ATP binding had not been explored prior to our investigation. In the present study, we focused on ATP binding to synapsin IIa because in hippocampal glutamatergic neurons this synapsin isoform has profound effects on synaptic depression (Gitler et al. 2008) and on vesicle clustering (Orenbuch et al. 2012b). A complementary study on the role of ATP binding by synapsin I was published concurrently (Orlando et al. 2014). Our research strategy was to compare WT synapsin IIa with a synapsin IIa mutant in which ATP binding is abolished. Previous studies have established that a single point mutation in the related synapsin Ia (K269Q) completely abolishes ATP binding without destabilizing the protein structure (Hosaka and Sudhof, 1998b). Based on the high homology of the ATP binding residues in synapsins I and II (Hosaka and Südhof, 1998b; Figure 6A,B) and the phylogenetic conservation of the synapsin C-domain (Kao et al. 1999) we introduced the parallel K270Q mutation into synapsin IIa. Because point mutations can potentially destabilize proteins, the effect of the mutation on the overall structure of the protein was assessed in the lab in two complementary manners. First, SWISS-MODEL was used to predict the structure of the mutant using the reported crystal structure of the C- domain of synapsin IIa (PDB-ID: 1I7N) as a template. No obvious structural differences were observed between the structure of WT synapsin IIa and the predicted structure of the K270Q mutant (Figure XC). Second, we expressed EGFP-tagged WT synapsin IIa (EGFP-SynIIa) and the K270Q mutant (EGFP-SynIIa-K270Q) in HEK cells and detected them by Western blot analysis using synapsin II-specific antisera. The products were indistinguishable (Fig. 6D) and their molecular weight was consistent with the fusion

58 of synapsin IIa (Thiel et al. 1990) with EGFP. A weak band, which may represent a minor degradation product, appeared to the same extent in both preparations. We concluded that the mutation does not alter the structure of the protein.

Figure 6 - Mutating the ATP-binding site in synapsin IIa does not overtly alter its structure. (A) Domain structure of synapsins IIa and Ia. K marks the location within the C domain of K270/K269 in synapsins IIa/Ia, respectively, both of which participate in the coordination of ATP binding. (B) Alignment of a 31 amino acid-long segment of the well-conserved C-domain of synapsin II and synapsin I illustrates the preservation of Lys270/Lys269 (SynII/SynI; underlined and highlighted in yellow) and of its neighboring residues: red, identical amino acid; green, similar polarity. (C) Superposition of the crystal structure (PDB-ID: 1I7N) of the C domain of synapsin II (white ribbon, residues 113–420) and the predicted effect of the K270Q mutation (green ribbon, modelled using SWISS-MODEL). Lys(K) – cyan, Gln(Q) – red. (D) Western blot (anti synapsin II) of untransfected HEK cells or HEK cells expressing EGFP- SynIIa or EGFP-SynIIa-K270Q. Both constructs produce a fusion protein of the expected combined molecular weight of synapsin IIa and EGFP (75 + 28 KDa, respectively). A weak lighter band, presumably a degradation product, was present to a similar extent for both constructs. From Shulman et al., 2015.

1.2 Mutating the ATP-binding site affects the association of synapsin IIa with the vesicle cluster

We investigated whether the K270Q mutation affected synaptic targeting of synapsin IIa. Synapsins homodimerize and heterodimerize in vitro and in vivo, a fact that affects their synaptic localization (Gitler et al. 2004b). To avoid interference from endogenous synapsins in measuring synaptic localization and other synaptic properties, all experiments were performed using neurons devoid of synapsins, obtained from synapsin TKO mice (Gitler et al. 2004a). Both EGFP-SynIIa and EGFP-SynIIa-K270Q

59 formed a punctate pattern in TKO neurons consistent with synaptic localization (Figure 7A). To quantify their propensity to localize in synapses, we measured their targeting index (see Materials and Methods for details; a targeting index of 0 indicates lack of targeting). Our results confirm that both synapsin IIa and its K270Q mutant target to presynaptic terminals, but indicate that the disruption of ATP binding by synapsin IIa reduces its enrichment within this compartment (Figure 7B). To further examine this finding, we quantified the distribution of synapsin IIa around the presynaptic terminal (Orenbuch et al. 2012a) by measuring the width of the synapsin puncta both at rest and in active terminals (Figure 7C,D). We found that at rest the mutant was distributed more widely around the presynaptic terminals (Figure 7D). Stimulation has been shown to induce a reversible detachment of synapsin from the vesicles and its dispersion into adjacent axonal segments (Chi et al. 2001; Orenbuch et al. 2012c). In agreement, incubation of neurons expressing EGFP-SynIIa in hyperkalemic saline (90 mM K+) resulted in the dispersion of synapsin IIa from the synaptic vesicle cluster into the axon (Figure 7D). The width of the redistributed synapsin was similar for both constructs even though the mutant was already distended at rest. After repolarization, both constructs regained their original distribution (Figure 7D). We conclude that ATP binding to synapsin IIa participates in defining its location within the resting presynaptic terminal, but not in its activity-dependent redistribution.

Figure 7 - Mutating the ATP-binding site affects the association of synapsin IIa with the vesicle cluster. (A) EGFP-SynIIa (left, green) or EGFP-SynIIa-K270Q (right) were coexpressed with TagBFP (blue) to assess their synaptic targeting capability. Merged images (bottom) illustrate that both proteins form a punctate pattern along the axons, consistent with synaptic localization. (B) Synaptic targeting of both proteins was quantified by calculating their “targeting factor”. Both proteins are enriched in synaptic puncta but targeting of the K270Q mutant is weaker. Shown are mean ± SEM values, n = 19 and 20 images for EGFP-SynIIa and EGFP-SynIIa-K270Q, respectively, ∼20 synapses analyzed in each. (C) EGFP- SynIIa and EGFP-SynIIa-K270Q both form synaptic puncta at rest (Ctl, left), which disperse into the adjacent axonal segments following 1 min of depolarization using hyperkalemic saline (middle). Recovery to the original distribution is achieved after 10 min in normal saline (right). (D) Quantification of C. The initial width of EGFP-SynIIa-K270Q is larger, but during stimulation, the final width of both constructs is equal. The original width of each fully recovers after repolarization and rest. n = 24 and 20 image sets (control, stimulated, recovered) for EGFP-SynIIa and EGFP-SynIIa-K270Q, respectively, ∼20 matching synapses analyzed per image. From Shulman et al., 2015.

1.3 Mutating the ATP binding site in synapsin IIa enhances vesicle clustering at the terminal

In a previous study from our lab, we showed that synapsin IIa increases synaptic vesicle density within the presynaptic terminal by enhancing vesicle clustering (Orenbuch et al. 2012b). To observe whether this property is preserved in the synapsin IIa K270Q mutant, we quantified synaptobrevin 2 (Syb2) immunofluorescence

61 intensity in the synaptic puncta of cultured glutamatergic TKO neurons. We expressed EGFP fusion constructs of synapsin IIa or its mutant; we separately expressed Venus as a control for possible effects related to viral infection (Figure 8A). The term “clustering” is used here to describe the collection of vesicles within a restricted segment of the axon, the synaptic puncta. In agreement with our previous report, Venus had no effect on syb2 fluorescence intensity, whereas EGFP-SynIIa increased the Syb2 signal (Figure 8B). Surprisingly, not only did the K270Q mutant increase syb2 fluorescence in synaptic puncta, its effect was significantly greater than that of WT synapsin IIa (Figure 8B). The stronger effect of the K270Q mutant was not due to higher expression levels, because its peak fluorescence intensity in synaptic puncta was actually lower, in agreement with its wider distribution around the synaptic puncta. This observation implies that preventing ATP binding to synapsin IIa promotes vesicle clustering. To explore this finding in greater detail, we plotted the fluorescence of Syb2 as a function of EGFP-SynIIa for all individual synapses of each image (Figure 8C). We found a mostly linear correlation between these two values and, on this basis, we termed the slope of each plot its “clustering index" (CI). The clustering index measured for the K270Q mutant was significantly higher than that of WT synapsin IIa (Figure 8D, left), in agreement with the suggestion that the K270Q mutant has a higher capacity to cluster vesicles within the presynaptic terminal. We repeated this experiment by analyzing the effect of synapsin IIa on endogenous synaptophysin I (SypI), to eliminate the possibility that synapsin IIa and its mutant affected the CI by altering the expression of Syb2 or its enrichment in individual synaptic vesicles. The K270Q mutant increased the clustering index measured using SypI antisera in a similar manner to that observed for Syb2 (Figure 8D, right). Although not conclusive, the correspondence of the results obtained using these two proteins implies that the effect of mutating the ATP-binding site in synapsin IIa is indeed mainly related to the density of vesicles within the presynaptic terminal.

Figure 8 - The K270Q mutation increases the capability of synapsin IIa to cluster vesicles at the presynaptic terminal. (A) TKO neurons were induced to express EGFP-SynIIa or EGFP-SynIIa-K270Q (green) and were subsequently processed for Syb2 (red) and GAD65 (not shown) immunofluorescence. (B) Density of synaptic vesicles within presynaptic terminals was assessed by measuring the Syb2 immunofluorescence intensity at the center of mass of the synaptic puncta. The values measured in each imaging session were normalized by the mean values recorded in TKO neurons. Only glutamatergic synapses that are negative for GAD65 staining were included in the analysis. Venus, as a control, had no effect on the vesicle density, SynIIa increased it, and the K270Q mutant had a significantly larger effect. n = 39, 10, 30, and 32 independent images (respectively), each including >30 synapses. (C) Relationship between the synapsin IIa content in each synapse and the density of vesicles therein was termed the clustering index (CI). This value was calculated per each image by plotting the Syb2 versus EGFP fluorescence intensities and extracting the slope. Shown are representative analyses of single images depicting neurons expressing EGFP-SynIIa (red) or EGFP-SynIIa-K270Q (blue). (D) Left, distribution of clustering index values calculated for endogenous Syb2 in neurons expressing EGFP-SynIIa and EGFP-SynIIa-K270Q (n = 25, 30 independent images, respectively). Each symbol represents the clustering index calculated from a single image; bars, 10–90% percentiles; box, 25–75% percentiles; middle bar, median. The clustering index of the K270Q mutant is significantly larger than that of EGFP-synapsin IIa (p<0.001, Mann-Whitney non-parametric test). Right, the clustering index values were also calculated for endogenous synaptophysin I. Similarly to syb2, the clustering index for the K270Q mutant was larger (p<0.05). n = 11 and 16 images. From Shulman et al., 2015.

1.4 ATP binding by synapsin IIa does not affect vesicle redistribution or inter-synaptic sharing

Limited redistribution of SypI has been reported to occur during stimulation of neurons in a manner that could indicate intracellular redistribution of vesicles (Li and Murthy 2001). To examine directly whether ATP binding to synapsin affects activity- dependent redistribution of vesicles, we imaged SypI-EGFP in neurons that were co- infected with either TagBFP-SynIIa or TagBFP-SynIIa-K270Q. We then measured the width of the EGFP-SypI puncta in resting and depolarized neurons. No effect of the ATP-binding site mutation was observed under either condition (Figure 9A), consistent with similar axonal distribution and similar localization of vesicles within the terminals. We also investigated whether acute direct depletion of ATP affects the confinement of vesicles within the terminal. ATP depletion was achieved by incubating cultured neurons with deoxy-glucose and oligomycin (ATP-depleting saline), which induce a substantial drop in [ATP]syn (Rangaraju et al. 2014). The expected decrease in [ATP]syn was confirmed by imaging SypI-ATeam1.03 (see Figure 29 in chapter 4). No effect was observed on the width of SypI-EGFP puncta even after 30 min of incubation (Figure 9B), inconsistent with macroscopic intracellular redistribution of vesicles. Another approach to addressing effects of ATP binding by synapsin IIa on vesicle distribution is to probe how the K270Q mutation affects inter-synaptic vesicle mobility (Orenbuch et al. 2012b). The idea is that the K270Q mutation might strengthen the attachment of the protein to the vesicles, which consequently could increase their tendency to cluster. We therefore measured inter-synaptic vesicle mobility using fluorescent recovery after photobleaching (FRAP) of EGFP-tagged synaptophysin I (Staras et al. 2010, 2013). Using this technique, we previously reported that synapsin IIa restricts vesicle mobility (Orenbuch et al. 2012b). Here, we compared the effect of TagBFP-labeled WT synapsin IIa and of its K270Q mutant on the recovery of photobleached SypI EGFP (Figure 9D). Both WT synapsin IIa and synapsin IIa- K270Q retarded FRAP of SypI-EGFP compared with TKO neurons, but did not differ between them (Figure 9C), suggesting that both proteins restrict vesicle mobility in a similar manner. We conclude that although ATP binding by synapsin IIa plays an important role in determining the size of the vesicle cluster, it does not regulate the strength of

64 anchoring of the individual vesicles. A different possibility to explain the enhancement of vesicle clustering is that the mutation may increase the number of vesicle-binding opportunities on the cluster without altering the strength of the association itself.

Figure 9 - Mutating the ATP-binding site does not affect vesicle redistribution at rest, during synaptic activity, or during depletion of cellular ATP. (A) Hyperkalemic depolarization induced widening of SypI-EGFP puncta in neurons also co-expressing either TagBFP-SynIIa or TagBFP-SynIIa-K270Q. The extent of widening and its recovery were similar in both cases. n=19 and 20 image sets (control, stimulated, recovered), respectively, approximately 20 matching synapses were analyzed per image. Two-way repeated-measures ANOVA, p=0.76. (B) The width of SypI-TagBFP puncta is unchanged after 30 min of incubation in ATP-depleting saline (1µM Oligomycin, 20 mM 2-Deoxy-Glucose). n=15, 16, and 17 image pairs for neurons co-expressing Venus, EGFP-SynIIa, or EGFP-SynIIa-K270Q, respectively, approximately 20 synapses analyzed per pair. Two- way repeated-measures ANOVA followed by Tukey’s post hoc analysis, p=0.12, 0.62, and 0.17, respectively. (C) Vesicle sharing in TKO neurons expressing SypI-EGFP or those coexpressing SypI-EGFP and either TagBFP-SynIIa or TagBFP-SynIIa-K270Q, was assessed by measuring the recovery of SypI- EGFP after photobleaching single synapses. Shown is a representative experiment in a neuron expressing TagBFP-SynIIa-K270Q. Arrowhead, bleached synapse. Elapsed time is indicated in minutes, starting with bleaching (t=0). SypI-EGFP fluorescence recovered gradually and partially within 35 min. (D) Quantification of FRAP experiments illustrating that both TagBFP-SynIIa and TagBFP-SynIIa-K270Q restricted vesicle sharing in TKO neurons to a similar extent. Recovery was compared by repeated- measures ANOVA, n=18, 33, and 24 experiments, respectively. TKO versus SynIIa, ***p =0.001; TKO versus SynIIa-K270Q, *p=0.048; SynIIa versus SynIIa-K270Q, p=0.37.

1.5 Mutating the ATP-binding site enhances basal evoked synaptic release but does not affect spontaneous release

It is reasonable to think that clustering additional vesicles within the presynaptic terminals would enhance synaptic transmission. However, it has been shown that basal neurotransmitter release does not necessarily correlate with the total amount of vesicles in a terminal (Branco et al. 2010); rather, it is the size of the readily releasable pool (RRP) that is a better predictor of the amplitude of basal release. To examine this point directly, we measured the properties of evoked synaptic release in autaptic neurons (the electrophysiological measurements were conducted by Dr. Yoav Shulman and are presented here for completeness of the presentation). Autaptic neurons were voltage-clamped at -70 mV and synaptic release was evoked by inducing an action potential in the axonal compartment by a brief step depolarization command (Figure 10A; Gitler et al., 2004b). We noticed that the amplitude of excitatory postsynaptic currents (EPSCs) recorded in TKO autaptic neurons expressing the K270Q mutant were slightly but significantly larger than those recorded in neurons expressing synapsin IIa (Figure 10B). An increase in evoked EPSC amplitudes can arise due to an increase in the size of the RRP (N), in the vesicular release probability (Pr), the quantal size (q, i.e. the response to the fusion of a single vesicle), or a combination of the aforementioned parameters. To explore which change may have occurred, we examined Pr and q. Possible effects on Pr were assessed by measuring frequency facilitation in the three groups of neurons, by recording the responses to 5 consecutive stimuli delivered at 10 Hz (Regehr 2012). Responses in neurons expressing EGFP-SynIIa exhibited facilitation, while neurons expressing the K270Q mutant or those expressing soluble Venus exhibited similar depression (Figure 10C and D). This result implies (but does not prove) that synapsin IIa reduces the vesicular Pr, and that the K270Q mutation prevents this effect. Further support for this conclusion is supplied by the measurement of synaptic depression during longer trains of stimulation (see below). Measurements of miniature spontaneous EPSCs revealed no difference in the quantal size of the three groups (20.29±3.8 pA, 20.29±2.1 pA, 19.87±2.4 pA, respectively, n=5,9,5 recordings, 1-was ANOVA, p=0.99). Based on the combination of the results presented, we propose that synapsin IIa, and more so its mutant, increase the size of

66 the total vesicle population and consequently the size of the RRP, but that In parallel, synapsin IIa decreases the per-vesicle Pr, an effect that is not supported by the K270Q mutant. As an end result, basal release in neurons expressing synapsin IIa remains unchanged compared to control neurons, but is lower than that observed in neurons expressing the K270Q mutant.

Figure 10 - Effect of synapsin IIa and its mutant on basal synaptic properties. (A) Average EPSCs elicited by brief depolarization commands in autaptic synapsin TKO neurons. Shown are control TKO neurons expressing Venus (black, n = 14), EGFP-SynIIa (red, n = 15), and EGFP-SynIIa- K270Q (blue, n = 22). The action potential current preceding the EPSC was blanked for the purpose of clarity. Superposition of the average normalized EPSCs (right traces) illustrates that they fully overlap. (B) Quantification of mean EPSC amplitudes, n = 14, 18, and 22, respectively. Notice that both SynIIa and its K270Q mutant do not differ from control, although they do differ from each other. (C) The K270Q mutation annuls the effect of synapsin IIa on frequency modulation of release. Neurons were stimulated at 10 Hz and the responses to five subsequent stimuli were monitored. Shown are representative recordings, as indicated. Control neurons (Venus) and neurons expressing EGFP-SynIIa-K270Q exhibited depression, whereas neurons expressing EGFP-SynIIa exhibited facilitation. (D) Quantification of frequency-modulation. n = 11, 8, and 11, respectively. From Shulman et al., 2015.

1.6 ATP binding by synapsin IIa is essential for its ability to sustain synaptic release during periods of intense activity

Synaptic depression is a progressive and reversible activity-dependent decrease in synaptic transmission that is typically observed during sustained high-frequency stimulation. One of its suggested mechanisms is the gradual depletion of vesicles in

67 the RRP (Regehr 2012; Zucker and Regehr 2002). Therefore, the rate of depression is influenced both by changes in the rate of exocytosis and by the recruitment of releasable vesicles to the RRP (Gabriel et al. 2011). Due to the key role that the synapsins play in managing vesicle reserves, most manipulations affecting the expression of the synapsins (either deletion or overexpression) typically result in an alteration of the kinetics of synaptic depression (acceleration or deceleration, respectively). In glutamatergic hippocampal neurons, synapsin IIa plays a key role in defining depression kinetics. Moreover, it is the only synapsin isoform to rescue the synaptic depression phenotype of the synapsin TKO mouse (Gitler et al. 2008). To explore the role that ATP binding to synapsin IIa plays in sustaining synaptic transmission during periods of intense activity, we induced short-term depression in autaptic TKO hippocampal glutamatergic neurons expressing Venus, synapsin IIa, or its K270Q mutant. As expected, expression of EGFP-SynIIa significantly slowed synaptic depression as compared to the control (Figure 11A and B). In fact, synaptic transmission initially facilitated for ~100 stimuli, and only then depression occurred. On the other hand, EGFP-SynIIa-K270Q did not affect the kinetics of synaptic depression (Figure 11A and B). We conclude that ATP binding plays a key role in controlling the kinetics of synaptic depression. An approach to assessing total release during a train is to integrate the EPSC traces (also termed the charge transfer). We observed that the charge transfer was significantly enhanced by synapsin IIa, but not by the K270Q mutant (Figure 11C), in agreement with the specific slowing of synaptic depression by synapsin IIa. Moreover, expression of synapsin IIa, but not its mutant, appeared to desynchronize release in the EPSCs, similarly to previous reports concerning GABAergic neurons (Feliciano et al. 2017; Medrihan et al. 2013, 2015). Altogether, our results indicate that mutating the ATP binding site in synapsin IIa results in two contrasting outcomes: although more vesicles are contained within the presynaptic terminal expressing the mutant compared with WT synapsin IIa, these terminals are less capable of supporting sustained synaptic release.

Figure 11 - Mutating the ATP-binding site annuls the effect of synapsin IIa on synaptic depression. (A) Neurons were stimulated 900 times at 10 Hz and their synaptic responses were recorded. Shown are representative responses for control neurons expressing Venus (control, black) and for TKO neurons expressing EGFP-SynIIa (red) or its K270Q mutant (blue). Every 15th response in the stimulation train is shown. (B) Normalized end-point subtracted semilog plots of mean ± SEM EPSC responses to stimulation trains as in A (n = 11, 7, and 11, respectively). Depression is deferred by the expression of EGFP-SynIIa, but not of EGFP-SynIIa-K270Q, compared with the Venus control. Furthermore, the rate of depression, once initiated, is slower for EGFP-SynIIa than for the K270Q mutant. For clarity, symbols and error bars are shown for every 10th stimulus, and down to ∼5% of the initial amplitude. Solid lines denote single- exponent fits applied starting 5, 10, and 9 s after the first EPSC, respectively. (C) Average time constant of synaptic depression, calculated by fitting the EPSC amplitudes with a single-exponent decay function as shown in B. One-way ANOVA followed by Tukey's post hoc analysis. Control versus EGFP-SynIIa, ***p = 0.0008; control versus EGFP-SynIIa-K270Q, p = 0.39; EGFP-SynIIa versus EGFP-SynIIa-K270Q, *p = 0.014. (D) Because synaptic depression follows complex kinetics, the t1/2 of depression was calculated. EGFP-SynIIa delays depression, whereas the K270Q mutant does not. One-way ANOVA followed by Tukey's post hoc analysis. Control versus EGFP-SynIIa, ***p = 0.0005; control versus EGFP-SynIIa-K270Q, p = 0.32; EGFP-SynIIa versus EGFP-SynIIa-K270Q, *p = 0.013. (E) Cumulative charge transfer during EPSC trains. Although EGFP-SynIIa significantly increased the charge transfer throughout the train, the K270Q mutant was ineffective in this respect. n = 6, 6, and 8, respectively. Repeated-measures ANOVA using every fifth EPSC value within the 200th-500th stimuli range, followed by Tukey's post hoc analysis. Control versus EGFP-SynIIa, *p < 0.049; control versus EGFP-SynIIa-K270Q, p = 0.99; EGFP-SynIIa versus EGFP-SynIIa-K270Q, *p = 0.033. (F) Synapsin IIa progressively increases the ratio between the charge and amplitude of each EPSC during stimulation trains, consistent with a gradual desynchronization of synaptic release. The K270Q mutation annuls this effect. Repeated-measures ANOVA, followed by Tukey's post hoc analysis. Control versus EGFP-SynIIa, **p = 0.002; control versus EGFP-SynIIa-K270Q, p = 0.92; EGFP-SynIIa versus EGFP-SynIIa-K270Q, **p = 0.006.

1.7 Interaction between ATP binding by synapsin IIa and its phosphorylation state

One of the best documented mechanisms for the regulation of the synapsins is their phosphorylation. Synapsin IIa has in its A domain a well conserved serine (S10; Figure 12A) that serves as a phosphorylation site for both protein kinase A (PKA) and calcium/ calmodulin-dependent kinase I (CaMKI) (Cesca et al. 2010). This site is homologous to S9 in synapsin I, which has been identified as a key activity-dependent regulator of the association of vesicles with synapsin (Hosaka and Südhof 1999). To determine whether interfering with ATP binding to synapsin IIa affects phosphorylation at S10, we used an antibody that specifically identifies phospho-S10. Although this antibody was originally raised against phospho-S9 in synapsin I, we found that it also identifies phospho-S10 in synapsin IIa, probably due to the high homology of the protein sequence next to this site (Figure 12A). To demonstrate the utility of this antibody, we immunostained TKO neurons expressing exogenous TagBFP-SynIIa under conditions expected to induce or repress phosphorylation at S10. Indeed, synaptic terminals in neurons depolarized by hyperkalemic saline were extensively stained, whereas those in resting neurons treated with H89, an inhibitor of PKA, exhibited much weaker staining (Figure 12B). Because the absolute peak intensity of the synapsin IIa signal in synapses is affected by the K270Q mutation (see above), we normalized the phospho- S10 immunofluorescence signal in each synapse by the TagBFP signal, producing a phosphorylation ratio that is independent of the synapsin IIa quantity. The phosphorylation ratio measured in the presence of H89 was similar for neurons expressing synapsin IIa and its K270Q mutant (Figure 12C). In the case of TagBFP- SynIIa, the phosphorylation ratio was higher in resting cultures that were not treated with H89, suggesting the existence of basal phosphorylation activity in resting cultures. This was not observed for the K270Q mutant. Finally, depolarization of the neurons by hyperkalemic saline induced a significant increase in the previously- mentioned ratio for both conditions, but more so for TagBFP-SynIIa (Figure 12C). The fact that the K270Q mutation reduced the magnitude of phosphorylation of synapsin IIa at the S10 residue suggests an interaction between these two distant sites.

To separate between the effects of ATP binding and S10 phosphorylation on the properties of synapsin IIa, we produced a S10A mutant of synapsin IIa that is unphosphorylatable at S10. We then measured its effect on the two main parameters that were altered by the K270Q mutation (Figure 12D and E). We found that the S10A mutant was less efficient than the K270Q mutant in clustering vesicles, but it was similar to wild-type synapsin IIa (Figure 12D). We therefore suggest that, at rest, the K270Q mutant affects vesicle distribution in terminals independently of its dampening effect on the phosphorylation of S10. Conversely, the S10A mutant was as ineffective as the K270Q mutant in rescuing synaptic depression in active neurons (Figure 12E). Although it is possible that the two mutations disrupt independently the capability of synapsin IIa to rescue synaptic depression, it is also just as likely that the effect of the K270Q mutation on synaptic depression is mediated by its inhibitory effect on the phosphorylation of S10.

Figure 12 - Mutating the ATP-binding site in the C domain of synapsin IIa attenuates phosphorylation in site 1 within the A domain. (A) Sequence homology of the A domains of synapsin II and synapsin I. Red, Identities; yellow background, phosphorylation site 1. (B) An antibody for phosphorylation site 1 (S9) in synapsin I interacts specifically with the homologous site in synapsin IIa (S10). Neurons from synapsin TKO mice,

71 which lack all synapsins, were induced to express TagBFP-SynIIa (left column) and were immunostained for phospho-S10 (IF P-S10, right column). Neurons treated with H89 (top row) exhibited background immunofluorescence, whereas depolarized neurons (bottom row) exhibited robust P-S10 immunofluorescence. (C) Comparison of the extent of phosphorylation of TagBFP-SynIIa and its K270Q mutant. Although the phosphorylation ratio in H89-treated samples was similar, it was significantly lower for the K270Q mutant both in resting and in stimulated neurons. n = 21, 28, 26, 29, 30, and 32 independent images per condition, respectively, per column from left to right, one-way ANOVA followed by Tukey's post hoc analysis comparing TagBFP-SynIIa versus TagBFP-SynIIa-K270Q in H89, p = 0.11; the same in resting neurons, ***p≪0.001; and the same in stimulated neurons, **p = 0.004. (D) Clustering index measured using immunolabeling of endogenous syb2 is significantly lower for EGFP-SynIIa-S10A than it is for EGFP-SynIIa-K270Q. n = 29 and 23 images for K270Q and S10A, respectively, p = 0.0001, Mann–Whitney nonparametric test. (E) Semilogarithmic end-point subtracted plots of the mean normalized EPSC amplitudes elicited by 10 Hz stimulation in glutamatergic autaptic neurons expressing EGFP-SynIIa-K270Q (n = 11) or EGFP-SynIIa-S10A (n = 5). For clarity, symbols and error bars are shown for every 10th stimulus and down to ∼5% of the initial amplitude. Solid lines denote single-exponent fits applied in the indicated range. Synaptic responses depressed in a similar manner in both groups. (F) Average time constant of synaptic depression for data shown in E. p = 0.44, Student's t test.

1.8 Synapses containing mitochondria contain a higher density of synaptic vesicles in WT but not in synapsin TKO neurons

ATP production participates in the maintenance of large SV clusters in hippocampal neurons and allows sustained neurotransmitter release (Li et al. 2008). We were interested in investigating whether binding of ATP by the synapsins is related to this observation. To address this question, we examined if the presence of mitochondria, which are the main producers of ATP in the cell (Attwell and Laughlin 2001), and which support many steps in the SV cycle (Murthy and Camilli 2003), affects the size of the SV cluster in the presynaptic terminals. Mitochondria are present in approximately half the synapses of cortical and hippocampal neurons (MacAskill et al. 2010). We therefore compared SV clusters in synapses of WT neurons which either contained mitochondria (defined as overlapping or up to 1μm distant), to those devoid of mitochondria. We postulated that the inhomogeneous localization of the mitochondria could form [ATP] gradients within the axons. Although fast equilibration of ATP within the cytoplasm could conceivably neutralize [ATP] gradients (Pathak et al., 2015), [ATP] gradients have been reported (Hubley et al., 1996). In chapter 4 we directly ascertained that under our experimental conditions [ATP] was higher in synapses that contain mitochondria than in those that do not (Figure 31). We subsequently measured SV clustering also in synapsin TKO neurons, reasoning that if

72 binding of ATP to synapsin is a major determinant in defining the organization of the SV cluster at rest, then the effect of mitochondrial proximity should be neutralized in these neurons. To visualize the mitochondria, we infected hippocampal cultures with the mitochondrial marker hSyn:MitoDsRed2. Neurons expressing MitoDsRed2 were then immunostained for Syb2 to visualize the SVs (Figure 13A), and the intensity of syb2 labeling was determined separately in synapses that harbor mitochondria (as defined above) and those devoid of them. In WT neurons, Syb2 intensity was higher in presynaptic terminals containing mitochondria as compared to those lacking them (Figure 13B), implying that the presence of the mitochondria affects SV clustering. This effect was not observed in synapsin TKO neurons (Figure 13B). We conclude that mitochondrial proximity affects SV clustering. Whether this observation is due to the effect of [ATP] gradients on synapsin-dependent SV clustering will need to be explored in more detail in the future. Besides controlling [ATP] (addressed specifically in chapter 4), mitochondria affect other key parameters of cellular metabolism; an important one is the control of the spatiotemporal distribution of calcium in the cytoplasm (Billups and Forsythe 2002; Kang et al. 2008; Lee et al. 2000). Consequently, in chapters 2 and 3 we examined how dysregulation of calcium handling by mitochondria affects SV clustering and SV usage.

Figure 13 – Higher density of synaptic vesicles occurs in the presence of mitochondria in the presynaptic terminals. Primary hippocampal cultures from WT and TKO mice were infected with the mitochondrial marker

MitoDsRed2 (red) and were immunolabeled for Syb2 (green). (A) Representative image of cultured WT neurons. Bottom: magnification of dashed white box area in top image. Notice the localization of the mitochondria in respect to the SV cluster. Marked with white arrowheads are examples of SV clusters

73 that overlap with mitochondria (+), those that are adjacent to them (<1μm, adj.) and those that are devoid of them (-). (B) Density of the synaptic vesicles within the presynaptic terminals was assessed by measuring the Syb2 immunofluorescence at the center of mass of the synaptic puncta (Mean±SEM). The values measured in each picture were normalized by the mean values measured in WT neurons. Only synapses negative for GAD6 were included in this analysis. n=16 images acquired from at least 3 different WT and TKO cultures, >300 synapses analyzed in each image, *p=0.04 ***p<0.001, p>0.05 for not-significant, One-way ANOVA followed by Tukey's post hoc analysis. Values are shown for all synapses combined in each genotype, or separately for those containing and not-containing mitochondria.

2 The effect of NCLX on the properties of synaptic vesicle clusters in the presynaptic terminals

To study the effect of the regulation of synaptic calcium by mitochondria on synaptic properties, we examined how deleting the sodium/lithium/calcium exchanger NCLX affects SV clustering and the division of vesicles into functional pools. We reasoned that deletion of NCLX will result in alterations in calcium handling by mitochondria, and that this could shed additional light on our results regarding the effect of mitochondrial proximity on SV clustering.

2.1 Deletion of NCLX enhances the size of the presynaptic vesicle cluster

To study the effect of NCLX deletion on the presynaptic vesicle cluster, we assessed the size of the total SV pool by quantifying the Syb2 immunofluorescence (IF) intensity in synaptic puncta of neurons cultured from WT and NCLX-KO mice. NCLX-KO mice were obtained from Jackson and a colony was established in the BGU animal facility (for more details see Materials and Methods section). The IF intensity of the synaptic cluster NCLX-KO neurons was greater (Figure 14A), suggesting a higher density of SVs. To further investigate this finding, we quantified the distribution of the fluorescence in the synaptic puncta at rest by measuring the Full Width at Half Maximum (FWHM) of their distribution along the axonal axis. In the NCLX-KO neurons the clusters are wider (Figure 14B) illustrating that the SV clusters are also larger in size. These results imply that NCLX-KO neurons have a higher capacity to cluster vesicles within the presynaptic terminals. An alternative explanation to our results is that the deletion of NCLX alters the expression of Syb2 or its enrichment in individual synaptic vesicle. To address these possibilities, in addition to Syb2, we analyzed also the distribution of the vesicular glutamate transporter 1 (vGlut1). The result for vGlut1 was essentially the same (Figures 14C and D). These finding strengthen the conclusion that the presynaptic terminal contains a larger and denser synaptic vesicle cluster in the absence of NCLX.

Figure 14 – Deletion of NCLX increases the synaptic vesicle cluster size in presynaptic terminals. (A) Left - neurons from WT and NCLX-KO mice were immunostained for Syb2 (red). Right –the density of the synaptic vesicles within the presynaptic terminals was assessed by measuring the Syb2 immunofluoresnce at the center of mass of the synaptic puncta (mean±SEM). The values measured in each picture were normalized by the mean values recorded in WT neurons. Only synapses negative for GAD6 were included in this analysis (n=9 and 10 images acquired from at least 3 independent WT and NCLX-KO cultures, respectively, >500 boutons analyzed in each image, Mann-Whitney non-parametric test, ***p<0.001). (B) FWHM quantification of (A). n=14 and 19 images acquired from at least 3 independent WT and NCLX-KO cultures, respectively, >30 boutons analyzed in each image, Student’s t- test, ***p<0.001. (C) Same as in (A), but immunostained for vGlutI1 (green). n=13 images acquired from at least 3 independent WT and NCLX-KO cultures, >500 boutons analyzed in each image, Student’s t- test, *p=0.019. (D) FWHM quantification of (C). n=9 images acquired from at least 3 independent WT and NCLX-KO cultures, >30 boutons analyzed in each image, Student’s t-test, **p=0.003.

2.2 NCLX deletion increases the stability of the presynaptic total vesicle pool

SVs are transported from the cell body and in between synapses through the axon. When SVs enter the presynaptic terminal, they may join an existing SV cluster. The existence of larger SV clusters in NCLX-KO neurons could be caused by a stronger association between transported SVs and the SV clusters. To assess inter-synaptic vesicle mobility we measured Fluorescence Recovery After Photobleaching (FRAP) of EGFP-tagged synaptophysin I in live neurons (Staras et al., 2010, 2013). We indeed found lesser recovery in NCLX-KO neurons (Figure 15B), supporting the conclusion that

76 the synaptic vesicles are more strongly attached to presynaptic terminals in NCLX-KO neurons.

Figure 15 – FRAP experiments reveal higher synaptic vesicle stability in NCLX-KO neurons. (A) Representative FRAP experiment in WT neurons infected with hSyn:SypI-EGFP to measure inter- synaptic vesicle mobility. Two synaptic puncta are shown. An arrow indicates a single chosen punctum just before (left), immediately after (middle) and ~8 minutes after bleaching (right). The fluorescence of the chosen synapse diminished to ~70% from the baseline fluorescence. Notice that the neighboring synapse was not affected by the bleaching. (B) mean±SEM time course of FRAP experiments in WT and NCLX-KO neurons. The intensity of fluorescence was normalized to the average pre-bleaching fluorescence. Fluorescence recovery was monitored in time lapse mode (18 images at 0.2 Hz and then 26 images at 0.06 Hz). The last five images in each experiment, were taken to represent the extent of recovery at 8 min after bleaching (n=23 and 22 experiments, respectively, 3 boutons analyzed in each experiment, Student’s t-test, *p=0.03).

2.3 The RcP Fraction is higher in NCLX-KO neurons

The SV cluster in the synapse is not homogenous in function. Vesicles within the recycling pool (RcP) are those that maintain synaptic release during physiological stimulation, while resting pool vesicles may be recruited during more intense activity (Rizzoli and Betz 2005). Neurons can alter the relative division of SVs between these pools (Kim and Ryan 2010, 2013), and thus modulate their capability to maintain synaptic release. In chapter 3 we present evidence that deletion of NCLX leads to a decrease in the release probability, resulting in weaker synaptic transmission (chapter 3, Figure 21). We hypothesized that homeostatic plasticity processes could lead to a compensatory increase in the fraction of the RcP. To assess whether this was the case, we measured the RcP size in two ways. First, we performed hyperkalemic loading of FM1-43, a styryl dye which is loaded into recycling vesicles during the endocytosis of SVs after they fuse with the plasma membrane (Ryan et al. 1993). FM1-43 staining was significantly higher in NCLX-KO neurons (Figure 16A), consistent with the observed higher density of the presynaptic vesicle cluster (Figure 15). However, the fraction of

FM1-43 destaining during field stimulation was greater in NCLX-KO neurons (Figure 16B), suggesting that in the NCLX-KO neurons the relative size of the RcP is higher. Activity-dependent changes in presynaptic calcium levels can alter endocytosis rates (Wu and Wu 2014). Because we found that intraterminal calcium is lower in NCLX-KO neurons during stimulation (chapter 3, Figure 20), we decided to probe the RcP size in an endocytosis-independent manner. For this purpose we used the alkaline trapping method (Sankaranarayanan and Ryan 2000b) in neurons expressing sypHy (Burrone et al. 2007; Granseth et al. 2006), a genetically encoded indicator of synaptic vesicle recycling. In this protocol, stimulation is performed in the presence of bafilomycin A, a selective inhibitor of the SV proton pump. Bafilomycin blocks synaptic vesicle reacidification, thus masking the effect of ongoing endocytosis on the sypHy fluorescence signal (Sankaranarayanan and Ryan 2000b). Therefore, prolonged stimulation in its presence results in the progressive accumulation of unquenched reporter in RcP SVs, eventually revealing the entire RcP. At the completion of each experiment we applied saline containing ammonium chloride which neutralizes all acidic compartments and reveals the total vesicle pool (Kim and Ryan 2010). In this way we found that the relative RcP fraction in NCLX-KO neurons is larger than in WT neurons (Figure 16C). This result suggests the activation of a compensatory mechanism to enhance release.

Figure 16 – NCLX-KO neurons exhibit a higher RcP fraction. (A) Intensity of FM1-43 loaded into synaptic terminals by 2 minute-long hyperkalemic stimulation (mean±SEM, n=11 and 10 experiments, respectively, >30 boutons analyzed in each experiment, Student’s t-test, *p=0.02). (B) Progressive unloading of FM1-43 by electrical stimulation delivered at 20Hz for 150 S (Student’s t-test, **p=0.009). (C) Left - sypHy traces in WT and NCLX-KO hippocampal neurons stimulated at 20Hz for 120 S in the presence of bafilomycin A (mean±SEM). The graph shows the cumulative release of SVs, normalized by the total SV pool size (measured after the addition of NH4Cl in saline). Right – quantification of the mean fractional fluorescence in the last 5 images taken before

NH4Cl application, meaning the plateau at the end of the stimulus, which indicates the RcP relative size (n=13 and 10 experiments, respectively, >50 boutons analyzed in each experiment, Student’s t-test, **p=0.0016).

2.4 Endogenous synapsin I, but not synapsin II, is overexpressed in NCLX-KO presynaptic terminals

Synapsins are the most abundant phosphoproteins that are associated with the membrane of the SVs. They cluster SVs based on their dual interaction with SVs and cytoskeletal elements (Cesca et al. 2010). Because deletion of NCLX enhances SV clustering (Figure 14), we were interested to see if any synapsin isoform is upregulated in NCLX-KO neurons. We immunostained and quantified synapsin I and synapsin II in cultured hippocampal neurons of WT and NCLX-KO mice. Synapsin I immunofluorescence was higher in NCLX-KO than in WT synapses (Figure 17B) while synapsin II was the same in both groups (Figure 17C), suggesting a possible upregulation of SynI. Because synapsins interact with high affinity with SVs (Benfenati et al. 1989), the increase in the immunofluoresence of synapsin I can be a direct outcome of the enhanced total SVs cluster in NCLX-KO neurons (Figure 14). This explanation is supported by the similar ratio of the increase in syb2 and synapsin I IF signal (an increase of approximatly 20% in both Syb2 and SynI intensity, Figures 14A and 17B). An alternative explanation is that unlike synapsin II, the affinity of synapsin I to vesicles and its propensity to dimerize is calcium dependent (Hosaka and Südhof 1998a; Menegon et al. 2006), consistent with the altered calcium homeostasis in NCLX-KO neurons (see results in Chapter 3). Future research will examine which of these two explanations, if any, is correct.

Figure 17 - Synapsin I but not synapsin II overexpress in NCLX-KO.

(A) Synapsin I (SynI, green) and synapsin II (SynII, red) were immunolabeled using the respective antisera in WT and NCLX-KO neurons. Expression levels were assessed by measuring immunofluorescence intensity at the center of mass of the synaptic puncta. The intensity of SynI (B) but not SynII (C) is greater in NCLX-KO synaptic puncta. n= 10 and 11 images for SynI and SynII immunolabeling in WT and NCLX- KO neurons, respectively, ~1000 boutons were analyzed per image in at least 3 independent experiments. Two-sample t-test, SynI, **p=0.008; SynII, ns: p>0.5.

3 Effect of NCLX deletion on synaptic transmission and plasticity

After determining that deletion of NCLX enhances SV clustering in synapses, we characterized its effect on calcium dynamics, to propose a mechanism that could reflect the possible different effects of mitochondrial function (ATP and calcium) on synapse structure and function.

3.1 Mitochondria in NCLX KO neurons sequester more calcium at rest and are partially depolarized

First we examined free and buffered calcium in axonal mitochondria in resting conditions. We compared the resting mitochondrial calcium levels in primary hippocampal neuronal cultures from WT and NCLX-KO mice by infecting the cells with the mitochondrially-targeted calcium sensor hSyn:2MT-GCaMP6m (MitoGCaMP, Figure 18A). To ascertain that MitoGCaMP is targeted to the mitochondria, we co- infected the cells with the mitochondrial marker hSyn:Mito-DsRed2. The pattern of both reporters overlapped, exposing the network-like mitochondrial distribution (Figure 18B), illustrating that MitoGCaMP is localized to the mitochondria. MitoGCaMP fluorescence is determined by the calcium concentration factored by the quantity of the indicator; the latter is influenced by the size and/or shape of the mitochondria. Therefore, we normalized the GCaMP fluorescence (G) by that of Mito-

DsRed2 (R) (see Materials and Methods). We found that the G/R ratio is higher in NCLX- KO mitochondria, consistent with higher resting calcium levels (Figure 18C). Most of the calcium in the mitochondria is bound to phosphate and is therefore termed “buffered mitochondrial calcium”. We reasoned that in addition to the free calcium measured by MitoGCaMP, the deletion of NCLX may affect the pool of buffered calcium. To address this hypothesis, we treated the cells with the uncoupling agent FCCP that disrupts the trans-mitochondrial pH gradient, thus releasing the buffered calcium (Heytler and Prichard 1962). The ratio between the initial fluorescence of the mitochondria and that measured after full depolarization in NCLX-KO neurons was higher (Figure 18D), implying that the amount of buffered calcium in resting mitochondria is higher in NCLX-KO neurons.

Mitochondrial membrane potential (ΔΨm) is strongly affected by the resting mitochondrial calcium concentration and the calcium load (Åkerman 1978).

Therefore, we monitored ΔΨm using the potentiometric dye TMRM (tetramethylrhodamine methyl ester) (Scaduto and Grotyohann 1999) before and after depolarization by FCCP . We found that during resting conditions, the mitochondrial membrane potential in NCLX KO neurons is depolarized compared to WT, (Figure 18E). The reduction in TMRM fluorescence intensity in NCLX-KO neurons corresponds to a calculated depolarization of ~23mV, using a previously reported equation (Kostic et al. 2018; Plášek et al. 2005). Altogether, we show that mitochondria lacking NCLX sequester higher amounts of calcium and are partially depolarized, implying that NCLX plays a key role in shaping the mitochondrial resting calcium load, similarly to observations in cardiac cells (Luongo et al. 2017).

Figure 18 - Mitochondrial calcium levels and rates are affected by the deletion of NCLX in resting conditions. (A) Schematic structure of MitoGCaMP6m (top) and targeting of the green and red fluorophores to the mitochondrial matrix (bottom). (B) Representative images of a primary hippocampal culture infected with MitoGCaMP6m and with the mitochondrial marker MitoDsRed2. Immunostaining for β3 tubulin was employed to identify neurons. (C) Quantification of the ratio between the green fluorescence of

MitoGCaMP6m (G) and the red fluorescence of MitoDsRed2 (R) to assess basal mitochondrial calcium levels in resting neurons. Mean±SEM, n=8 and 11 experiments, respectively, >40 mitochondria analyzed in each, ***p<0.001, Student's t-test. (D) Comparison of buffered mitochondrial calcium levels in WT and NCLX-KO neurons using bath application of FCCP. Left: mean±SEM fluorescence traces, normalized by FFCCP, the fluorescence measured during the FCCP-induced plateau. Right: mean±SEM F/FFCCP value measured before the application of FCCP (n=11 and 12 experiments, respectively, >75 mitochondria

82 analyzed in each, Student's t-test, *p=0.027). (E) Left - representative fluorescence images of NCLX-KO and WT cultured hippocampal neurons continuously superfused with the ∆Ψm indicator TMRM. FCCP was applied after baseline imaging to normalize the signal, as in D. Right: Initial F/FFCCP, normalized by average WT value (n=17 and 11 experiments, respectively, >25 mitochondria analyzed in each, Student's t-test, ***p<0.001).

3.2 Calcium shuttling through the mitochondria is altered in the absence of NCLX

A decrease in the mitochondrial membrane potential and calcium overload, as we found in NCLX-KO neurons, can reduce mitochondrial calcium influx, efflux or both (Duchen 1992; Rottenberg and Scarpa 1974). To examine the effect of deleting NCLX on mitochondrial calcium handling during neuronal activity, we applied field stimulation to hippocampal neurons expressing MitoGCaMP. Fluorescence measurements during stimulation trains (10Hz, 2sec) revealed that the change in mitochondrial calcium is smaller in NCLX-KO neurons compared to WT ones (Figure 19A), suggesting lower calcium influx into the mitochondria in the absence of NCLX. Importantly, although the calcium influx rate (in terms of its time constant) remains the same (Figure 19B), the capacity for calcium efflux is decreased in KO compared to WT neurons, demonstrated by the lower slope in the efflux phase (Figure 19C). These results suggest that deletion of NCLX affects both influx and efflux of calcium into/out of the mitochondria. Thus, the activity-dependent axonal mitochondrial calcium influx capacity is lower but this is not due to a difference in the rate of calcium uptake; on the contrary, calcium efflux capacity and rate are reduced in the absence of NCLX.

Figure 19 – NCLX deletion decreases calcium levels in the mitochondria during neuronal activity. Cultured hippocampal neurons were stimulated for 2 S at 10 Hz (black bar) in the presence of APV and DNQX to block recurrent activity and mitochondrial calcium levels were measured using MitoGCaMP6m.

(A) Left: mean±SEM ΔF traces; right: mean±SEM peak ΔF values (n=11 and 17 experiments, respectively, >45 mitochondria analyzed in each, Student's t-test, ***p<0.001. (B) Average half life time (t½) of the data shown in F. Student's T-test, p=0.15 not significant. (C) Quantification of the initial efflux rate (linear fit 1-1.5 seconds after the peak). Student's t-test, ***p<0.001.

3.3 NCLX deletion affects cytoplasmic calcium levels in presynaptic boutons during synaptic activity

Mitochondria are present at a high incidence in presynaptic terminals (Harris et al. 2012) and can be tethered to vesicle release sites (Rowland et al. 2000). Together with their ability to handle high calcium fluxes (Duchen 1999; Nicholls and Budd 2000), they can have a significant impact on the spatio-temporal distribution of calcium in the cytoplasm. We therefore sought to monitor the levels of cytoplasmic calcium within presynaptic terminals during induced synaptic activity. For that purpose, we constructed the hSyn:SypI-mCherry-GCaMP6f construct and infected neurons with it (Figure 20A). This construct is localized to the presynaptic terminal by synaptophysin I, a synaptic vesicle protein, as is illustrated by immunolabeling for the vesicular protein vGlut1 (Figure XB). mCherry, a calcium-insensitive fluorescent protein, is included in addition to GCaMP6f, allowing ratiometric calcium imaging (G/R), which factors out differences in the quantity of synaptic vesicles or expression levels. At rest, G/R values were comparable between the two genotyoes, indicating a similar resting calcium concentration. Upon electrical stimulation (20 Hz, 1 sec) we observed a smaller change in the G/R ratio in NCLX-KO neurons compared to WT neurons, illustrating that presynaptic calcium in the terminals increased to lower values (Figure 20C). Calculation of the initial efflux rate observed after cessation of stimulation revealed that it was slower in NCLX-KO neurons (Figure 20D), consistent with impaired mitochondrial calcium buffering and shuttling.

Figure 20 - Cytoplasmic calcium levels reached during stimulation are different in NCLX-KO and WT synaptic boutons. (A) Schematic structure of SypI-mCherry-GCaMP6f (top), and synaptic targeting of the red and green fluorophores to synaptic vesicles. (B) Representative image of WT primary neuronal cultures infected with SypI-mCherry-GCaMP6f and immunostained for vGlut1, a synaptic vesicle protein (blue). The calcium indicator is localized to the synaptic boutons. Immunostaining for β3 tubulin was employed to identify neurons. (C) Neurons expressing SypI-mCherry-GCaMP6f were stimulated at 20 Hz for 1 S. GCaMP6f fluorescence (G) was measured and normalized by mCherry fluorescence (R) measured prior to stimulation. Left: mean±SEM G/R traces. Right: mean±SEM peak G/R values (n=13 and 18 experiments, respectively, >55 boutons analyzed in each, Student's t-test, **p=0.003). (D) Time constant of the decay in the G/R value in C. Mann-Whitney non-parametric test, ***p<0.001.

3.4 Synaptic release is lower in NCLX-KO neurons due to a decrease in the probability of release

Synaptic release is triggered by an increase in the calcium concentration in the presynaptic terminal (Baker 1972; Katz and Miledi 1969). We therefore expected that synaptic transmission would be affected by the deletion of NCLX. To study the role of NCLX in controlling synaptic properties, we infected NCLX-KO and WT neurons with hSyn:sypHy, a synaptic release sensor (Burrone et al. 2007). NCLX-KO neurons exhibited lower peak ΔF/F0 values during 5 sec field stimulation at 20Hz compared to WT neurons (Figure 21A), suggesting weaker synaptic release in NCLX KO neurons. A possible explanation of this observation was that lower calcium levels led to a decrease in the probability of release (Pr; Schneggenburger and Neher, 2005); we therefore compared Pr in WT and NCLX KO neurons in acute hippocampal slices. For this purpose, we used MK-801, an activity-dependent irreversible blocker of NMDARs (N-methyl-d-aspartate receptors) (Song et al. 2018). In this protocol, synaptic NMDAR responses are unmasked by blocking AMPA receptors and omitting magnesium from the bathing medium. When applying low-frequency stimulation, NMDARs are activated by secreted glutamate; the membrane depolarizes, as evidenced by fEPSP (field excitatory postsynaptic potential) recordings, and concurrently, only the active NMDARs are irreversibly blocked by MK-801. During successive stimuli, more synaptic contacts are blocked. Because the fraction of synapses that are activated during each stimulus is determined by Pr, a lower Pr leads to blockade of a smaller fraction of the existent synapses, and a slower decay in the fEPSPs amplitude (Jackman et al. 2016). We found the blockage rate to be slower in the Schaffer collaterals connecting the

CA3 and CA1 hippocampal areas in NCLX-KO slices (Figure 21B,C), suggesting a lower

Pr in NCLX neurons.

When we enhanced Pr in cultured NCLX-KO neurons by increasing the extracellular calcium concentration to 5mM, their responses, as measured using sypHy, equalized to that of WT slices (Figure 21D), illustrating that the release capacity of NCLX-KO neurons is not in itself limiting. We therefore conclude that deletion of NCLX leads to a decrease in Pr, consistent with the lower peak levels of synaptic calcium during electrical stimulation (Figure 20).

Figure 21 - Deletion of NCLX decreases synaptic release and the initial probability of release. (A) Left: mean±SEM sypHy traces in WT and NCLX-KO neurons. Neurons were field-stimulated at 20Hz for 5 sec and the sypHy signal was monitored before, during and after stimulation. Right: Peak mean±SEM ∆F/F0 fluorescence obtained from the traces at left (n=19 and 17 experiments, respectively,

>55 boutons analyzed in each, Student's t-test, **p=0.009). (B) Pr was measured from Schaffer collateral synapses onto CA1 neurons using the MK-801 protocol. Stimulation was delivered every 15 S while recording NMDAR-fEPSPs in saline devoid of magnesium ions containing DNQX and bicuculline, to isolate NMDAR responses. Traces were recorded before and after bath application of MK-801. Top: traces averaged from 10 trials before (black traces) and 20-30 trials after (orange trace) MK-801 application. MK-801 partially blocked NMDAR responses. Bottom left: Average traces illustrating different rates of progressive blockage of NMDAR-fEPSPs in WT and NCLX-KO neurons during repetitive stimulation at 1/15 Hz. Bottom right - half-decay times of the NMDAR-fEPSP peak amplitudes (n=9 and 6 recordings from 4 WT and 3 NCLX-KO mice, respectively. Student’s t-test, *p=0.01). (C) As in (A) but performed in extracellular solution containing 5mM calcium. Left: sypHy peak responses (but not recovery) in WT neurons recorded with 2 mM external calcium are similar to those of NCLX-KO neurons measured with 5mM external calcium. Right: Quantification of peak responses (mean±SEM) of traces

86 on left (n=11 experiments, >55 boutons analyzed in each, Mann-Whitney non-parametric test, p=0.8 not significant).

3.5 NCLX-KO exhibit higher frequency facilitation in the hippocampus Schaffer Collateral pathway

A possible outcome of a change in Pr, such as that we identified in NCLX-KO slices (Figure 21), is an alteration in short term plasticity, and specifically in synaptic facilitation. Synaptic facilitation is a phenomenon in which postsynaptic potentials are augmented during short bursts of repetitive stimulation. This phenomenon is generally considered to result from the accumulation of residual calcium within the presynaptic terminal during the stimulation train; calcium accumulation leads to a progressive increase in Pr and an enhancement in neurotransmitter release (Charlton et al. 1982; RENGEL 1992). When the initial Pr is low, it can increase significantly during the successive stimuli, whereas, when it is initially high, the capacity to enhance it is lower (Pr is limited by 1). Furthermore, depletion of the readily releasable pool of vesicles by the initial stimuli leads to an overall decrease in the synaptic response during successive responses, termed depression. The rate of depletion is proportional to Pr, and therefore is stronger when the initial Pr is high. The end result is that when initial Pr is low, responses tend to first increase and then depress, whereas, when initial Pr is high, depression is predominant. We compared the change in fEPSPs recorded in the CA1 area of the hippocampus during repetitive stimulation (5 stimuli) of the Schaffer collateral pathway in NCLX-KO and WT acute slices (Figure 22A). In all stimulation frequencies we tested, NCLX-KO slices exhibited higher facilitation

(Figures 22B and C), consistent with a lower initial Pr (Figure 21). For single stimuli, fEPSP amplitude was linearly dependent on stimulation intensity (Figure 22D), indicating that recordings were not saturated.

Figure 22 - NCLX-KO exhibit higher frequency facilitation. (A) Representative fEPSPs recorded from the CA1 area delivering 5 stimuli to the Schaffer collaterals at 20 Hz in acute slices prepared from WT (black) and NCLX-KO (blue) mice. Note the higher degree of facilitation in the 2-5 responses in NCLX-KO slices. (B) fEPSP amplitude plotted against stimulation intensity. The stimulation intensity is normalized by an intensity producing a response of 0.3mV, illustrating that responses in these slices are not saturated, and that both WT and NCLX-KO slices respond similarly to alteration of the stimulation intensity. (C) fEPSP amplitudes normalized by first response in each train (mean±SEM), in WT and NCLX-KO slices at 5 Hz, 10 Hz, 20 Hz and 50 Hz, from left to right (n=18 and 21 recordings from 8 and 9 WT and NCLX-KO mice, respectively). The stimulation intensity in each train was set to produce an initial response of ~0.3 mV. Notice that P2 responses facilitated at all frequencies; at high frequencies (20 and 50 Hz), a subsequent depression of P3-5 vs. P2 was observed. (D) Quantification of the P2/P1 ratio as shown in (C). Stronger facilitation was observed in NCLX-KO slices at all frequencies. n=18 and 21 slice recordings acquired from 8 WT mice and 9 NCLX- KO mice, respectively. Student's t-test, 5Hz: **p=0.002; 10Hz: *p=0.015; 20Hz: **p=0.003; 50Hz: ***p<0.001.

3.6 Long-term potentiation is impaired in NCLX-KO

Long-term potentiation (LTP) is a phenomenon in which brief tetanic stimulation of afferent fibers results in a long-lasting increase in synaptic strength or efficacy (Bliss and Lømo 1973). Calcium influx into postsynaptic terminals is necessary for generating and sustaining Schaffer collateral LTP (SC-LTP; Lynch et al., 1983). Because our results indicate that NCLX affects calcium levels at the presynaptic terminal during stimulation trains and that it affects neurotransmitter release, we hypothesized that it could also decrease calcium transients within the postsynaptic terminals, thus compromising LTP. To investigate this hypothesis, we recorded Schaffer collateral LTP

88 in WT and NCLX-KO slices. We found that unlike WT slices, which exhibited LTP following high-frequency stimulation (HFS), this protocol failed to elicit LTP in NCLX- KO slices (Figure 23B,C). Thus, our results indicate that deletion of NCLX abolishes LTP.

Figure 23 - NCLX-KO slices fail to exhibit hippocampal Schaffer-collateral LTP. (A) Schematic for extracellular recordings at Schaffer collateral CA1 cells from acute hippocampal slices. The LTP schematic representation of induction protocol (5 pulses at 100 Hz, repeated once after 20 sec) is indicated below. (B) fEPSP in area CA1 evoked by stimulating the Schaffer collateral pathway. The CA3 area was stimulated to evoke fEPSP of approximately 0.3mV, and the fEPSPs were normalized by the mean of the baseline responses. Mean±SEM *100 (%) responses in LTP experiments in WT (n=17 slices from 7 mice) and NCLX-KO (n=9 from 4 mice). WT slices exhibited 30% LTP while NCLX-KO did not after HFS (arrow). (C) Mean±SEM responses 15 minutes after HFS (as indicated by bar in B). Mann-Whitney non-parametric test, *p=0.015.

4 The relationship between synaptic ATP levels, mitochondria and synapsins

In chapter 1 we investigated how mutating the synapsin IIa ATP binding site affects synaptic properties, and we raised the possibility that ATP gradients and activity- dependent changes in ATP may interact with ATP binding by synapsin. In this chapter we expand on this topic and also explore the possibility of an opposite effect: do synapsins have an impact on presynaptic ATP levels during resting conditions and during synaptic activity? First we addressed directly the question whether neuronal activity changes the synaptic ATP concentration ([ATP]syn), because such changes could affect synaptic properties. Besides the well-documented requirement of ATP for the function of many cellular mechanisms related to exo-endocytosis and vesicle transport (Harris et al. 2012; Rangaraju et al. 2014), the concentration of ATP was also shown to influence the biochemical properties of the synapsins (Hosaka and Südhof 1998b; Shulman et al. 2015). However, in order for ATP binding by synapsin to serve as an effective synaptic control mechanism, [ATP]syn should respond to physiologically relevant synaptic activity. Because synaptic transmission is a major consumer of cellular energy

(Harris et al. 2012; MacAskill et al. 2010), it is reasonable to assume that [ATP]syn decreases during intense synaptic activity, although there exist buffer mechanisms to prevent or delay this eventuality (Rosenmund and Westbrook 1993; Wyss and Kaddurah-Daouk 2000).

It was recently reported that [ATP]syn indeed drops after the cessation of stimulation

(Rangaraju et al. 2014); we confirmed that [ATP]syn is altered by intense neuronal activity under our experimental conditions, in support for a regulatory role for

[ATP]syn. Following, we examined how deleting the synapsins affects [ATPsyn], to understand whether the synapsins could serve as a buffer mechanism for ATP.

4.1 Construction and use of a synaptic ATP sensor

ATeam1.03 is an ATP-sensor based on fluorescence (Förster’s) resonance energy transfer (FRET) between a monomeric cyan fluorescent protein (mseCFP, the donor) and circularly permutated Venus (cpmVenus, the acceptor) linked by the ATP-binding

ε-subunit of the B. Subtilis F0F1 ATPase, which changes its conformation upon binding ATP (Imamura et al., 2009). To target ATeam1.03 to the presynaptic terminal, which is the relevant location for studying the environment of synapsin, we fused it to SypI, a transmembranal synaptic vesicle protein (Johnston et al. 1989). The fusion protein SypI-ATeam1.03 (Figure 24A) targeted efficiently to presynaptic terminals of hippocampal neurons (Figure 24B), as evidenced by its colocalization with the endogenous v-SNARE synaptobrevin 2 (Figure 24C).

Figure 24 - The synaptic FRET-based ATP sensor ATeam1.03-SypI. (A) Schematic structure of the synapticly targeted FRET-based ATP sensor SypI-ATeam1.03. ATeam1.03 was fused to the C-terminal of synaptophysin I, thus targeting it to synaptic vesicles. (B) SypI-ATeam1.03 expressed in WT hippocampal neurons. (C) SypI-ATeam1.03 expressing neurons were immunostained for synaptobrevin 2 as a marker of synaptic vesicles. Both the mseCFP (blue) and cpmVenus (green) components of SypI-ATeam1.03 colocalize with endogenous synaptobrevin 2 (red and see merged image at bottom), illustrating that SypI-ATeam1.03 is targeted to presynaptic terminals. From Shulman et al., 2015.

4.2 Acceptor-photobleaching and sensitized emission methods for determination of FRET during synaptic activity

ATeam1.03 FRET was measured using two alternative methods. First, the acceptor- photobleaching method (Laviv et al. 2010) (in collaboration with the Slutsky lab, TAU). Briefly, energy transfer from mseCFP (donor) to cpmVen (acceptor) quenches the donor fluorescence. Elimination of cpmVen absorption via selective photobleaching results in an increase in mseCFP fluorescence, which allows unambiguous determination of FRET efficiency (Em) without relying on the fluorescence of the acceptor (see Materials and Methods). For this purpose, cpmVen was selectively photobleached using the 488nm laser line, which does not affect the mseCFP donor

(Figure 25B), and mseCFP emission increased due to its dequenching (Figure 25B). FRET was measured under resting conditions and after electrical field stimulation at

10Hz for 90 sec. Resting terminals exhibited a robust Em of 0.23±0.02, which was reduced by 35% to 0.15±0.02% by stimulation (Figure 25C), indicating a drop in synaptic ATP following neuronal activity.

Figure 25 – ATeam1.03 FRET measurements using the acceptor photobleaching method. (A) Representative confocal images of boutons of a neuron infected with SypI-ATeam1.03. (B) Pseudocolor-coded fluorescent images of cpmVen (left) and mseCFP (right) in a resting bouton before (top) and after (bottom) cpmVen was bleached using a 514nm laser. Note the increase in mseCFP fluorescence after cpmVen photobleaching. (C) FRET efficiency (Em) of SypI-ATeam1.03 in boutons decreased from 0.23±0.02 to 0.15±0.02 by stimulation at 10Hz for 90 sec. n=1/41 and 1/35, respectively. Two-tailed unpaired Student’s t test. **p=0.007. From Shulman et al., 2015.

A major disadvantage of the acceptor-photobleaching method is that it involves irreversible bleaching of the acceptor, and is therefore less appropriate for measuring dynamic processes (it allows only single-point measurements). Therefore, we also measured FRET using SpRET, a spectral sensitized emission method developed in our lab which relies on unmixing the emission spectrum to calculate FRET (Levy et al. 2011). To calculate the FRET efficiency (E), three fluorescence components (besides autofluorescence) are measured: Donor emissions, acceptor emissions due to FRET, and acceptor emissions due to direct excitation by the light source (not FRET). The donor and total acceptor emission values are extracted by spectral unmixing of the emission spectrum, as is explained in the Materials and Methods section. The acceptor emission is then separated into its two components. The direct excitation component can be calculated by factoring the quantity of the acceptor, which is measured in a separate acquisition step using a lower energy laser line (488nm). Factoring is performed using an independently determined conversion factor based on measurements performed in cells expressing exclusively the acceptor. Although this

92 approach yields E, the requirement for an additional acquisition step is cumbersome. In a simpler scheme, this step is omitted and instead of E, the readout is FR (FRET ratio). FR is the ratio between acceptor emission and the sum of acceptor and donor emissions (without separating direct excitation). FR is not an absolute value, and it depends on the experimental configuration: If measurements are performed under conditions that minimize direct excitation of the acceptor, the baseline FR is minimal, and its dynamic range (ratio between maximal and minimal value) is large. This aim can be achieved using the 405nm laser line for excitation, because it excites cpmVen only weakly. However, two technical considerations convinced us to use the 442nm laser line in some experiments (as itemized below) even though its direct excitation of cpmVen is stronger, resulting in a higher baseline FR and a smaller dynamic range.

First, unlike the 405nm line, the 442nm line can excite mito-DsRed2, and thus mitochondria could be imaged in the same acquisition step used to measure FRET (see Figure 31), without requiring an additional acquisition step. Second, due to optical chromatic aberrations that could not be compensated (no water-immersion objectives corrected for 405nm light were commercially available for our microscope), the images formed using the 405nm laser were inferior in resolution. In the paragraphs below we state which laser lines were used in each experiment.

Figure 26 – Spectral FRET determinations. (A) Neurons from WT mice were infected with SypI-ATeam1.03. Images of the same neurons were taken while exciting at 405nm and 442nm. (B) Emission spectral analysis of the same synapse excited at 405nm (left) and 442nm (right). Graphs are normalized by the peak of the donor’s spectrum. Notice the emission of mseCFP (cyan), cpmVenus (yellow) and their sum (black). The symbols represent the

93 emission raw data acquired every 5nm. The basal FR is higher for the 442nm excitation line, resulting a lower dynamic range for this condition.

4.3 Calibration of SypI-ATeam1.03 and SypI-ATeam1.03NL

To determine the capability of ATeam1.03-SypI to measure changes in synaptic [ATP], we performed an in-situ calibration to determine their constant of dissociation (Kd) and their dynamic range. We measured both parameters for the original ATeam1.03, and for ATeam1.03NL, a newer variant that exhibits lower affinity to ATP (i.e., a higher

Kd value ,Tsuyama et al., 2013). Both sensors were fused to SypI in order to stably anchor them to a non-soluble cellular structure (synaptic vesicles). The fusion proteins were expressed in neurons, and the membrane of the neurons was permeabilized by perfusing the chamber with calcium-free saline mimicking the intracellular environment, but containing a low concentration of saponin (0.01%; see Materials and Methods for more details). Using this wash solution, the integrity of the plasma membrane was compromised, as evidenced by the fast leakage of soluble fluorescent proteins, while preserving the gross structure of the cells, as evidenced by the conservation of anchored fluorescent proteins (data not shown). After permeabilizing the plasma membrane, the [ATP] in the bath solution was increased stepwise and the FR at the synaptic puncta was measured (using the 405nm laser line). We found that at room temperature the Kd (the deflection point of the dose-response curves) for ATeam1.03 was [ATP]=1.4 mM and for ATeam1.03NL it was 2.6 mM, similarly to previously reported values (Imamura et al. 2009; Tsuyama et al. 2013). Compared to ATeam1.03, ATeam1.03NL exhibited a lower dynamic range, due to lower FRET efficiency between its components (its acceptor mVenus is not circularly permuted), indicating that it is expected to be less precise. When we compared the dose-response curves of both constructs to the FR measured in intact neurons at rest prior to permeabilization, we found that the measured FR was similar to the highest FR level of both constructs (Figure 27, horizontal dashed lines). Thus, both ATeam1.03-SypI and ATeam1.03NL-SypI appear to be either at saturation or close to it at resting [ATP] under our experimental conditions. However, because the calibration procedure involves permeabilization of the cells and consequent significant changes in the cellular environment, it is possible that the Kd values that we

94 measured are not identical to those relevant for work with intact cells, as is often the case in in-vitro calibrations(Roe et al. 1990). Furthermore, we note that the ATeam sensors are sensitive to the ambient temperature, i.e. their affinity to ATP is inversely related to temperature (Imamura et al. 2009). To conclude, the ATeam constructs are not expected to be sensitive to subtle changes in [ATP] at room temperature but can be useful for the determination of substantial decreases in [ATP]. Moreover, future experiments using ATeam1.03 or ATeam1.03NL should be performed at near-body temperature, at which the Kd of ATeam1.03 for ATP is reported to be close to 4 mM (Imamura et al. 2009), in which case the sensors are not projected to be saturated, allowing the measurement of subtler changes in [ATP], including increases.

Figure 27 - ATP affinity calibration of AT1.03 and AT1.03NL. Neurons were permeabilized using 0.01% saponin in calcium-free saline supplemented with increasing concentrations of Na3ATP as indicated. The dissociation constant (Kd) was found to differ between the

ATeam1.03 and ATeam1.03NL constructs: The Kd of the former (Black) was 1.4 mM, while that of the latter (Red) was 2.6 mM. Black and red dashed lines represent the mean basal FR acquired with AT1.03 (0.59 ± 0.002; mean ± SEM) and AT1.03NL (0.49 ± 0.02; mean ± SEM) respectively. The value of FR at saturation differs for ATeam1.03 and ATeam1.03NL, thus affecting the effective dynamic range obtained under the experimental conditions. n=3 independent experiments each, mean±SEM.

4.4 Intense synaptic activity consumes ATP in the presynaptic terminal

In collaboration with the Slutsky lab (TAU) we performed FR measurements during electrical stimulation (Tel-Aviv University, Figure 28). Stimulation at 10 Hz for 90

95 seconds induced a significant 13.7±2% drop in FR in the presynaptic terminals. After the stimulation was terminated, FR gradually recovered over 10 minutes to pre stimulation values. As a control we used SypI-ATeam1.03R122K/R126K, in which mutations to the ε-subunit of the F0F1 ATPase render it insensitive to ATP and in the open (no FRET) state (Imamura et al. 2009). No changes were observed in FR with this construct (Figure 28) strongly indicating that stimulation does not induce significant ATP- independent alterations in either the donor or acceptor fluorescence. Thus, we conclude that [ATP]syn drops transiently during stimulation and recovers over minutes at rest, consistent with the hypothesis that ATP can serve as a modulator of synapsin function.

Figure 28 - Monitoring activity-dependent changes in [ATP]syn using a synaptic FRET-based ATP sensor. (A) Schematic structure of the synapticaly targeted FRET-based ATP sensor SypI-ATeam1.03. (B) SypI- ATeam1.03 expressed in WT hippocampal neurons. Notice its punctate synaptic localization. (C) SypI- ATeam1.03 expressing neurons were immunostained for synaptobrevin 2 as a marker of synaptic vesicles. Both the mseCFP (blue) and cpmVenus (green) components of SypI-ATeam1.03 colocalize with endogenous Syb2 (red and see merged image at bottom). (D) The SypI-ATeam1.03 FRET ratio (FR,

FcpmVen/FmseCFP) was calculated in neurons at rest, after 90 s stimulation at 10 Hz, and finally 3 and 10 min after cessation of stimulation (black bars). FR was reduced to 82.7 ± 2% by stimulation and gradually recovered back to 100 ± 2% 10 min after cessation of stimulation. Shown are mean ± SEM values normalized by the initial FR. n = 5/84 coverslips/boutons. SypI-ATeam1.03R122K/R126K, which is insensitive to ATP, did not show a significant change in FR after stimulation or during the recovery period (gray bars). n = 4/96. From Shulman et al., 2015.

To strengthen the claim that the decrease in FRET within SypI-ATeam1.03 that was observed during synaptic stimulation was due to a decrease in [ATP]syn, we measured FRET while challenging the neurons with oligomycin and 2-deoxy-glucose, which compromise mitochondrial ATP production and glycolysis, respectively (Rangaraju et

96 al. 2014). This treatment reduced the SypI-ATeam1.03 signal progressively over time, down to the level measured with SypI-ATeam1.03R122A,R126A (Figure 29).

Figure 29 - Decline in [ATP]syn due to blockage of ATP synthesis. Neurons expressing SypI-ATeam1.03 were imaged at rest (black squares) or while being incubated with 1 μM oligomycin and 20 mM 2-deoxy-glucose (green circles). Neurons expressing an ATP insensitive mutant of SypI-ATeam1.03 R122K/R126K were imaged at rest (gray triangles). The fluorescence ratio (FR) is indicative of bound [ATP]. FR in the synapses of oligomycin+2-deoxy-glucose treated neurons declined towards minimal (RR) levels over more than 20 minutes. From Shulman et al., 2015.

4.5 [ATP]syn levels are lower in synapsin TKO neurons, but can be rescued by the reintroduction of synapsin IIa

It has been proposed that the ATP concentration is maintained in synapses not only by its local production by synaptic mitochondria, but also due to its association with abundant ATP-binding proteins (Chavan et al. 2015). One of the main groups of synaptic ATP-binding proteins is the synapsins (Esser et al. 1998), and we posited that neurons lacking all synapsins may possess lower ATP levels in their synapses. We infected neurons of WT and synapsin TKO mice with SypI-ATeam1.03 and measured FR on our confocal microscope. Our measurements revealed that in the absence of synapsins, synapses exhibited a lower FR (Figure 30A), suggesting lower basal [ATP]syn. To exclude the possibility that this effect is caused by secondary compensatory mechanisms unrelated to synapsins, we reintroduced SynIIa into TKO neurons. We found that expression of tagBFP-SynIIa in TKO neurons increased FR back to WT levels

(Figure 30A, red), thus supporting our proposal that synapsins affect [ATP]syn. Furthermore, we checked whether the presence of synapsins affects the magnitude of the decrease in [ATP]syn during neuronal activity. We measured the change in FR

97 when exposing the neurons to hyperkalemic saline, which induces cell depolarization and synaptic activity. In both WT and synapsin TKO neurons the depolarization induced a significant drop of ~11% in the FR measured in presynaptic terminals (Figure 30B). Therefore, while resting ATP was effected by the presence of the synapsins, synaptic activity reduced ATP in the synapses in a similar manner (see however comment above concerning the sensitivity of ATeam1.03 measurements). These results suggest that synapsins could participate in determining and maintaining the resting levels of ATP within presynaptic terminals.

Figure 30 – Basal FR is higher in WT than in TKO neurons. (A) SypI-ATeam1.03 was expressed in neurons from WT, TKO mice, and TKO neurons expressing SynIIa- tagBFP. We measured the basal FR (using 405nm excitation). Each symbol represents an average of 5 consecutive images taken at 1 minute intervals. n=19, 14 and 8 experiments, respectively, >30 synapses analyzed in each, TKO: **p=0.002; TKO-SynIIa: **p=0.0011, One-way ANOVA followed by Tukey's post hoc analysis. (B) Neurons from WT and TKO mice were infected with SypI-ATeam1.03 to measure basal FR (using 442nm excitation) before and after applying a hyperkalemic solution. Shown are mean±SEM values, normalized by the FR measured in WT neurons at rest. n=8 and 5 experiments, >30 synapses analyzed in each, Student's t-test,WT: *p=0.011; TKO: *p=0.03.

4.6 Correlation of [ATP]syn and mitochondria in resting neurons

The larger SV cluster in synapses containing mitochondria (chapter 1, Figure 13) could be attributed to the local production of ATP, although other mechanisms, such as modulation of calcium dynamics (see chapters 2-3) are also possible. We examined whether synapses containing mitochondria exhibit higher [ATP]syn, and whether this value was sensitive to the presence of synapsins. For that purpose we co-infected neuronal cultures with SypI-ATeam1.03 and the mitochondrial marker MitoDsRed2. The emission spectrum was unmixed into channels (see materials and methods; Figure 31A small panel, Figure 31B) allowing us to determine the distance of the

98 mitochondria from the SV clusters. We observed that synapses harboring mitochondria had higher FR values in WT but not in synapsin TKO neurons (Figure 31C). This observation could mean that the presence of mitochondria affects local ATP levels, but that this effect depends on the presence of the synapsins. Whether this is related to ATP binding by the synapsins was not explored, and will be examined in the future by expressing in synapsin TKO neurons synapsins that are mutated in their ATP binding sites.

Figure 31 - A correlation between resting [ATP]syn and mitochondria.

(A) Neurons from WT mice were infected with the SypI-ATeam1.03 and MitoDsRed2 vectors resulting in targeting of ATeam1.03 to synapses (Green) and visualization of mitochondria (Red). Notice that after unmixing of the spectra, mitochondria can be detected readily and synapses can be categorized into synapses that contain mitochondria (+ mit), synapses with adjacent mitochondria (adj. mit) and synapses without mitochondria (- mit). (B) Emission spectrum (excitation at 442 nm) from synapses devoid of mitochondria (left) or adjacent to a mitochondrion (right). Notice the emission of cerulean

(cyan), venus (yellow) and DsRed2 (red line). (C) Neurons from WT and synapsin TKO mice were infected with SypI-ATeam1.03 and MitoDsRed2. The FR was calculated in synapses with and without mitochondria. Each dot represents a mean of 5 images during resting conditions. n=18 and 11 experiments from each group, >35 synapses analyzed in each, Two-sample t-test, WT: **p=0.004; synapsin TKO, ns: p>0.5.

Discussion

1 ATP-binding to synapsin IIa

Although the fact that synapsins are ATP-binding proteins was reported more than a decade ago, the implications of this observation had not been explored prior to my project. Fortunately, the amino acids that participate in ATP binding to synapsin I were mapped (Esser et al. 1998; Hosaka and Südhof 1998b), allowing us to exploit the high homology of the C-domains of synapsin I and II to identify the phylogenetically conserved K270 residue in synapsin IIa (Figure 6). We mutated K270 to glutamine (as reported to block ATP binding, Hosaka and Südhof, 1998a; Orlando et al., 2014) and examined its effect on key presynaptic properties. Our results indicate that preventing ATP binding to synapsin IIa increases its capability to cluster vesicles, attestable by the higher Syb2 content in synaptic puncta of neurons expressing SynIIa-K270Q and by the larger vesicle clustering index measured using both Syb2 and SypI (Figure 8). The mutation had no measurable effect on intersynaptic vesicle mobility (Figure 9), indicating that it was not necessarily the strength of the association between synapsin IIa and vesicles that is enhanced, but rather the quantity of per-vesicle effective association sites. Nevertheless, the mutation lowered the capacity of synapsin IIa to localize to the synaptic terminals (Figure 7) or to sustain synaptic transmission during periods of intense demand (Figure 11). This probably reflects a disruption of the synapsin-dependent recruitment of reserve vesicles to the RRP that takes place during periods of intense activity (Gabriel et al. 2011; Garcia-Perez et al. 2008; Neher 2012; Walter et al. 2013), with only a minor effect on the resting size of the RRP (Figure 10 and Gitler et al., 2008). Therefore, the ATP-binding site mutation appears to produce a deficit in the coupling between activity (demand) and supply of the stored vesicles; phosphorylation of synapsin IIa could serve as the underlying mechanism. Synapsins are known to control vesicle availability by providing reserve vesicles during periods of intense synaptic activity (Cesca et al. 2010), but were also shown in some cases to affect basal release (Coleman and Bykhovskaia 2016; Hilfiker et al. 2005; Hilfïker et al. 1998; Llinas et al. 1985). The supply of vesicles involves the phosphorylation of synapsins, which profoundly lowers their affinity to vesicles and to

100 cytoskeletal elements (Bykhovskaia 2011; Cesca et al. 2010; Hosaka et al. 1999). We found that the K270Q mutation diminishes the phosphorylation of synapsin on its PKA/CaMKI phosphorylation site 1 (Figure 12). This key observation implies a functional interaction between the ATP-binding site and the regulation of synapsin IIa by phosphorylation, even though the two sites are quite distant within the protein structure (Figure 6A). Although we did not identify how ATP binding alters the phosphorylation of synapsin IIa, we note that synapsin has been proposed to act as a phosphate transfer enzyme (Esser et al. 1998; Hosaka and Südhof 1998a). Because its target was not identified, it is possible that synapsin IIa directly or indirectly enhances its own phosphorylation. To determine whether the effects of the K270Q mutation on vesicle clustering and on synaptic depression are mediated by its observed inhibition of phosphorylation at S10, we examined the effect of the S10A mutation directly on both properties (Figure 12D–F). We found that the effect of K270Q on vesicle clustering in resting neurons is independent of S10, because it is not recapitulated by the S10A mutation (Figure 12D). In contrast, both the K270Q and S10A mutations disrupted the capability of synapsin IIa to slow synaptic depression in active neurons (Figure 12E,F). Therefore, our results indicate that synapsin IIa is regulated both by

[ATP]syn and by phosphorylation. We postulate that ATP serves as an allosteric modulator of synapsin IIa, as was shown for other proteins (Lu et al. 2014), and predict that ATP-free synapsin IIa should be more affine to vesicles because it is the K270Q mutant, and not S10A, that clusters vesicles to a higher degree. This prediction remains to be corroborated directly by biochemical assays. Furthermore, we propose that the deficit of the SynIIa-K270Q mutant in rescuing synaptic depression is due to its lower propensity for phosphorylation at S10. Although this is not necessarily the only interpretation of our results, we suggest that, in this manner, synapsin IIa fine-tunes the processes of vesicle clustering and vesicle recruitment, depending on the level and duration of neuronal activity. We speculate that, at rest, dephosphorylated synapsin IIa tends to accumulate and cluster vesicles in the reserve pool. Mutating the ATP-binding site enhances this function due to the proposed higher affinity of synapsin IIa to vesicles when not bound to ATP. With the commencement of activity, phosphorylation of synapsin IIa facilitates the recruitment of the previously accumulated vesicles. Both

101

K270Q and S10A are deficient in this respect due to their lower/non-accessibility to phosphorylation at S10. Finally, we predict that a decrease in [ATP]syn associated with sustained activity eventually enhances the affinity of synapsin IIa to vesicles, impeding the severe depletion of remaining vesicles.

2 The role of synapsins and mitochondria in regulating presynaptic [ATP]

Mitochondria can modulate local changes in [ATP] in addition to their calcium regulation capabilities. We decided to investigate if the presence of mitochondria within presynaptic terminals affects SVs clustering in a synapsin-dependent manner. We found that synapses harbor larger SVs cluster when a mitochondrion is localized to the presynaptic terminals, in agreement with the literature (Smith et al. 2016), and that this effect is present only in WT neurons but not in synapsin TKO ones (Figure 13). While synapsins are long known to participate in SVs clustering (Cesca et al. 2010), it was recently proposed that they form a liquid phase within the cytoplasmic aqueous environment which captures within it small lipid vesicles (Milovanovic et al. 2018). Is this observation affected by [ATP] gradients? Synapsins' function is regulated by ATP- binding: although we showed that the mutated synapsin IIa enhances SV clusters (Figure 8), synapsin I on the contrary may change its conformation in response to ATP binding, creating a more stable tetrameric form (Brautigam et al. 2004), which could contribute to a larger SV cluster. It is also intriguing to note that ATP itself has been proposed to be an hydrotrope at physiological concentrations (Patel et al. 2017), meaning that it can itself affect the solubility, association (and phase separation, Kang et al., 2018) of cellular proteins. How these observations are related to our findings remains to be determined. Our results indicate that synapsins may be regulated by local changes in ATP, but also that synapsins may have an impact on resting [ATP]syn levels.

Various steps of neuronal activity have been shown to be major energy sinks (Harris et al., 2012), leading to the question of whether ATP fluctuates in an activity- dependent manner in synapses. The specific steps of neurotransmission that consume most ATP have been visualized using a synaptically-targeted modified luciferase

102 termed Syn-ATP (Rangaraju et al. 2014). We utilized the ATP sensor ATeam1.03 for the same purpose, targeting it to presynaptic terminals by fusing it to SypI. The advantage of the SypI-ATeam1.03 sensor over Syn-ATP is that it requires no additional accessory materials (such as luciferin), and that the readout is ratiometric by nature (i.e., the readout does not depend on the level of expression of the sensor or on the volume of the imaged cellular compartment). The study conducted by Rangargaru et al. concluded that [ATP]syn decreases minutes after completing a period of strong stimulation, during the time when vesicle recycling occurs. In contrast, we observed a decrease in [ATP]syn immediately after delivering stimulation at 10 Hz; this decrease recovered over several minutes (Figure 28 and 30). Because we did not measure

[ATP]syn during the stimulation train itself, we cannot define the precise timing of the decrease in [ATP]syn, but it appears to take place earlier than in the report by Rangaraju. A possible reason for this apparent discrepancy is that the underlying principle of action of Syn-ATP and SypI-ATeam1.03 are dissimilar; SypI-ATeam1.03 is based on FRET, whereas Syn-ATP relies on enzymatic activity. Therefore, key biochemical properties of the probes may differ in a manner that can affect their output. Notwithstanding the possible differences, it is clear that synaptic activity alters

[ATP]syn.

The changes in [ATP]syn were observed within a time frame that is relevant to synapsin function, lending credibility to the claim that [ATP]syn can serve as a modulator of synapsin, as we claim in chapter 1. Are the changes in [ATP]syn within a relevant concentration range to affect synapsin? The in vitro affinity of synapsin to the ATP analog ATPγS is in the submicromolar range (Brautigam et al. 2004; Hosaka and Südhof 1998a), which is outside of the physiologically relevant range. However, it was shown that Mg2+ shifts this affinity to the low millimolar range (Hosaka and Südhof 1998a), placing it within the range in which activity-related [ATP]syn changes were reported (Rangaraju et al. 2014) and potentially maximizing the projected sensitivity of synapsin to changes in [ATP]syn. We note that, unlike synapsin II, ATP binding by synapsin I is calcium dependent (Hosaka and Südhof 1998a). Because persistent synaptic activity induces temporary increases in cytosolic calcium, the putative regulation of synapsin I by ATP may be more strongly dependent on the activity level of the neuron than on resultant alterations in [ATP]syn. The study of the differential consequences of ATP

103 binding to synapsin I and II is therefore warranted. It is important to realize that the interplay between activity-dependent changes in calcium, magnesium, pH, and

[ATP]syn, as well as in the affinity of synapsins to ATP, are expected to be more complex than those alluded to here. Future quantitative models on the effects of [ATP]syn on synapsin-related vesicle availability will need to take this fact into account. Recent research has proposed that ATP is concentrated in synapses not only due to its local production there, but also due to its association with abundant ATP-binding proteins (Chavan et al. 2015). Because one of the main groups of synaptic ATP-binding proteins is the synapsins, we were in an excellent position to investigate whether neurons lacking all synapsins maintain lower ATP levels in their synapses. Indeed, in resting neurons we observed lower [ATP]syn levels in synapsin TKO neurons compared to WT (Figure 30A). Reintroducing synapsin IIa into the neurons restored the ATP levels (Figure 30A, red). During neuronal activity induced by hyperkalemic solution, we observed no difference in the decrease in [ATP]syn in TKO neurons (Figure 30B), suggesting that synapsins could play a significant role in maintaining resting ATP levels within the presynaptic terminals.

The contribution of mitochondrially-derived ATP production to presynaptic ATP levels is still a matter of debate in the literature. Although ATP diffusion in muscle cytoplasm was reported to be similar to that observed in solution (Hubley et al. 1995), it was later reported to be three orders of magnitude slower (Selivanov et al. 2007), explaining how local consumption of ATP can form ATP gradients. Neurons exhibit intricate forms in which discrete presynaptic terminals are separated by thin axons, promoting slow diffusional exchange (Santamaria et al. 2006). Because ATeam1.03 can be targeted to synapses, and because not all presynaptic terminals contain mitochondria, we were able to examine directly whether [ATP] equilibrates efficiently across synapses. Our results indicate that [ATP]syn in resting neurons is correlated with the presence of mitochondria (Figure 31). Recently published papers either agree (Sun et al. 2013) or disagree (Chavan et al. 2015; Pathak et al. 2015) with this idea. One study demonstrated that ATP produced by mitochondria diffuses rapidly between synapses so that the localization of the mitochondria to presynaptic terminals is not crucial for the supply of their energy requirements (Pathak et al., 2015). An alternative

104 explanation for this result could be that although they used the same sensor, ATeam1.03, theirs was in its soluble form. Soluble fluorescent proteins are highly diffusive and achieve equilibrium rapidly (Tevet and Gitler 2016; Zheng et al. 2011). This raises the possibility that the mobility of the sensor could have masked extant differences in [ATP]syn. A significant correlation between the presence of mitochondria and higher resting [ATP]syn exists only in the presence of the synapsins (Figure 31), which agrees with the idea that proteins could serve as ATP-buffers. However, it is still unclear whether the quantity of the synapsins and their function are affected by the proximity of mitochondria and consequently the abundance of ATP, or whether the synapsins themselves promote sequestering of ATP molecules. Could synapsins bind ATP and buffer it, or can synapsins determine in an undetermined way the metabolic state within the synapse? Further research is needed to investigate the interaction between synapsins and [ATP]syn.

3 The role of NCLX in SV clustering

Mitochondria regulate synaptic activity not only by providing ATP but also by regulating local calcium transients. In neurons, besides being the initiator of synaptic transmission, calcium is also a regulator of SV clustering, SV recycling and synaptic transmission. The contribution of mitochondrial calcium to these processes is less studied, although its significance has been reported by many (Billups and Forsythe 2002; Kang et al. 2008; Lee et al. 2000). One of the problems in researching the role of calcium shuttling via the mitochondria and its significance in synaptic properties is the lack of identified mitochondrial transporters, but recently NCLX was characterized (Palty et al. 2010; Sekler et al. 2009) and we were able to utilize NCLX-KO mice to address the main question of how the mitochondrial calcium efflux system, mediated via NCLX, affects basic synaptic properties as well as plasticity. By comparing WT to NCLX-KO neurons we suggest that the synaptic vesicle cluster is bigger in neurons lacking NCLX as indicated by stronger labeling of Syb2 and vGlut1 (Figure 14). The SVs were distributed over a wider area in NCLX-KO neurons (Figure 14B and D) implying that the presynaptic terminals in NCLX-KO possess a larger total SV pool. To further support our conclusion, we assessed intersynaptic SVs mobility using FRAP. Our results

105 suggested that SV clusters in NCLX-KO neurons are less mobile and thus more stable (Figure 15). We note that the resolution of wide-field microscopy limits the interpretation of the results presented above, and that in the future electron microscopy or super-resolution studies will need to be performed to corroborate our interpretation. The SV cluster in the presynaptic terminal is not homogenous in function and is divided into different functional pools. Vesicles in the RcP are the vesicles that can be recruited during prolonged and intense synaptic activity (Denker and Rizzoli 2010). We were interested in measuring the relative size of the RcP to see whether deletion of NCLX affects the division of SVs into different functional pools. Our results indicate that in NCLX-KO neurons the fraction of the RcP is larger, as seen by sypHy and FM1-43 experiments (Figure 16). We suggest that long-term compensatory mechanisms are activated in the NCLX-KO neurons to counterbalance weaker synaptic transmission (see chapter 3) by increasing the relative size of the RcP. It has been shown that the RcP size is highly variable and regulated by CDK5/calcineurin activity (Kim and Ryan 2013; Marra et al. 2012). Phosphorylation of synapsin I by CDK5, regulates the resting and recycling pools of synaptic vesicles in an activity-dependent manner (Verstegen et al. 2014). Because synapsins are known to be SV binding proteins (Bykhovskaia 2011), we looked at the expression of endogenous synapsins I and II. We found that the IF label for synapsin I but not synapsin II was increased in NCLX-KO synapses (Figure 17). This could arise due to the larger SV cluster in NCLX-KO neurons, but another explanation could be that synapsin I is regulated by calcium by synapsin II is not (Hosaka and Südhof 1998a). Moreover, calcium increases the ability of synapsin I to bind SVs and to dimerize (Orlando et al. 2014), consistent with a higher representation in the presynaptic terminal. Due to the fact that in this experiment the same synapses were analyzed for both synapsin isoforms, we speculate that synapsin I is indeed calcium-dependent and that mitochondrial calcium homeostasis partakes in local calcium changes that affect SV clustering.

106

4 The role of NCLX in modulating calcium homeostasis, neuronal transmission and plasticity

Due to the singular importance of calcium ions as second messengers, the regulation of calcium dynamics within the mitochondria and in their vicinity has substantial effects on key cellular processes. NCLX plays a central role in controlling the balance between mitochondrial and cytoplasmic calcium, especially following calcium influx through the plasma membrane; indeed, conditional late deletion of the NCLX gene Slc8b1 in mouse cardiomyocytes in-vivo caused mitochondrial dysfunction and eventual lethality (Luongo et al. 2017). Specifically, cardiac mitochondria were depolarized, overloaded with calcium, and consequently, calcium uptake from the cytoplasm was decreased. We decided to examine the role of NCLX in regulating calcium dynamics in neurons and how this may alter their synaptic properties, considering that calcium directly triggers synaptic vesicle fusion and that altering its spatiotemporal dynamics modifies synaptic release and/or plasticity. We found striking similarities between our results and those reported for the cardiac system. Specifically, at rest the mitochondria in NCLX-KO neurons exhibited higher levels of calcium and were partially depolarized. By measuring calcium while fully depolarizing the mitochondria using FCCP (to unbuffer occluded calcium), we concluded that the fraction of buffered calcium in NCLX-KO mitochondria was higher (Figure 18). Our results thus suggest that the mitochondria in NCLX-KO neurons are overloaded with calcium. Furthermore, calcium influx into the mitochondria during trains of action potentials was reduced, but not due to a lower rate of influx; rather, efflux was weaker (Figure 19). Multiple mechanisms, only some of which involve the mitochondria, determine cytoplasmic calcium levels at rest and during activity-dependent calcium influx. We observed that neuronal resting cytoplasmic calcium levels were unaltered in NCLX-KO neurons, suggesting that mechanisms other than mitochondria may dominate, a conclusion that will need to be examined in more detail in future studies. Surprisingly, the magnitude of presynaptic calcium transients during neuronal activity was lower in NCLX-KO neurons, even though the calcium-clearance rate was slower (Figure 20). Our findings are therefore congruent with previous reports that suggested that

107 mitochondrial calcium homeostasis regulates plasma-membrane voltage gated calcium channels (Herrington et al. 1996). This observation could also contribute to the observed lower calcium influx rate into the mitochondria (Figure 19). For our calcium imaging experiments we used the genetically-encoded calcium indicator GCaMP (Nakai et al. 2001) which combines a circularly permuted GFP with calmodulin at the C-terminus and the calmodulin binding region at the N-terminus. When the calmodulin segment binds calcium, it interacts with its binding region, leading to a conformation change that induces a substantial increase in fluorescence. In our experiments we utilized GCaMP6m and GCaMP6f, because they are highly sensitive to calcium and exhibit an improved signal to noise ratio (Chen et al. 2013; Horikawa 2015). Even though they exhibit improved sensitivity for calcium, there is a limitation in assessing the true rise or decay times of calcium fluxes because the sensor kinetics are still dominated by intrinsic properties of the indicator (Brockhaus et al. 2019) and not necessarily from the influx and efflux rates, especially the ultra-fast cytosolic fluxes. This fact may have masked some of the effects of the NCLX-KO on both the mitochondrial and cytosolic flux rates. Another difficulty we encountered was that if basal calcium levels differ, this would affect the measured F0, and therefore the ΔF/F0 values we calculated could be misleading. For that reason we fused the synaptic GCaMP6f with the calcium-insensitive fluorescent protein mCherry, thus forming a ratio-capable sensor. Using it (SypI-GCaMP6f-mCherry) we measured a G/R ratio, which excludes effects related to the quantity of the sensor or the volume of the imaged area. In this manner, only differences in calcium could affect the measured ratio, and we could ascertain that basal calcium was unaffected. Although still in progress, we are repeating the same approach for MitoGCaMP6m and our preliminary results are in agreement with the results presented in figure 19. A decrease in presynaptic calcium levels during stimulation is expected to weaken evoked synaptic transmission (Schneggenburger and Neher 2005). Indeed, vesicle release, as measured using sypHy, was decreased (Figure 21A). More definitively, we observed that the initial Pr in NCLX-KO neurons was reduced (Figure 21B), and that synaptic facilitation was enhanced in NCLX-KO neurons, consistent with a lower initial

Pr (Figure 22). Interestingly, the amplitude of synaptic release could be compensated by elevating extracellular calcium (Figure 21C), in agreement with our results

108 indicating that under standard conditions the activity-induced presynaptic calcium transient in the NCLX KO neurons was lower. Recently, a mutation in the human Slc8b1 gene (NCLXP367S), which is associated with severe mental retardation and cardiomyopathies, was identified in a Pakistani family, consistent with dysregulation of calcium dynamics in the brain and the heart (private communication, Israel Sekler, BGU). This mutation leads to loss of function in NCLX due to the close proximity of the mutation to the NCLX catalytic site. We were interested in determining whether deletion of NCLX produces a synaptic effect that could be consistent with the mental retardation observed in the siblings harboring the NCLXS367P mutation. To examine this reasoning, we tested the effect of deleting NCLX on one form of long-term plasticity, namely SC-LTP. We found that slices from NCLX- KO mice could not sustain SC-LTP, as compared to WT slices (Figure 23). We did not explore the mechanism by which NCLX deletion affects LTP. Nevertheless, we note that the release of calcium back into the cytosol of the postsynaptic terminal, after its accumulation in mitochondria during rounds of intense activity, has been suggested to play a part in LTP induction (Tang and Zucker 1997). According to this model, efflux from mitochondria contributes to a slight sustained elevation in cytoplasmic calcium levels which contribute to the formation of LTP. We postulate that in NCLX-KO, due to being overloaded, the mitochondria cannot regulate calcium levels efficiently and thus cannot contribute significantly to the formation of LTP. To conclude, NCLX function facilitates SC-LTP, illustrating the importance of mitochondrial control of calcium dynamics for higher brain functions. It is somewhat surprising that the NCLX-KO phenotype in the brain appears less severe than that observed in the cardiac study using conditional deletion in the heart (Luongo et al. 2017). This could be due to the different developmental stages during which NCLX was annulled in the two studies. In the conventional NCLX KO mice, because NCLX functionality is missing altogether, compensatory mechanisms may be activated during early developmental stages, allowing somewhat normal neuronal network activity. For example, we observed higher calcium buffering capacity in the NCLX-KO mitochondria in parallel to a decrease in their influx capacity. This could be explained by the partial depolarization of the mitochondria which reduces the electrical driving force for calcium entry but could also be the result of secondary inhibition of the MCU,

109 as was previous proposed in relation to an interaction of MCU with EMRE (Sancak et al. 2013). Likewise, when NCLX deletion was performed during the development of the cardiac system in the conditional KO mice, undefined compensatory adaptations were activated (Luongo et al. 2017), leading to less severe outcomes. It is therefore plausible that compensatory mechanisms are activated during development in the brain of NCLX-KO mice and that these promote neuronal viability, even if functionality remains somewhat impaired. To conclude, our results obtained in NCLX-KO neurons support a role for NCLX in basic synaptic mechanisms. We show synaptic deficits in NCLX-KO neurons consistent with mitochondrial involvement in synaptic regulation – both in basic regulation of release and in synaptic plasticity phenomena that are pivotal for learning and memory in the brain.

110

Conclusions

The regulation of synaptic transmission is a complex and fascinating subject. Here we add a new angle to this story by revealing that the properties of synapsin IIa, a protein that affects the availability of vesicles during intense synaptic activity, is tightly bound to the presynaptic dynamics of ATP, a subject that has been in the spotlight for many years in neuroscientific research. Furthermore, we provide evidence for the importance of presynaptic mitochondria for synaptic function in a synapsin- dependent manner. Moreover, we utilized NCLX-KO neurons to investigate how mitochondrial calcium efflux, which is mediated by NCLX, affects basic synaptic properties, synaptic transmission and plasticity. We postulate that some of the long- term effects we saw in NCLX-KO neurons are mediated by synapsin due to the well- documented regulation of synapsins by calcium. In addition, we introduced throughout this work new methods and approaches to study the topic of synaptic energetics. We conclude that synaptic activity is highly regulated by mitochondrial ATP and calcium signaling. Furthermore, we add new insight on the role of synapsins in modulating SVs clustering by regulating their activity by ATP and calcium. There is still much to investigate on this topic: What is the mechanism by which synapsins affect local ATP levels, or conversely, how are they themselves affected by changes in ATP? Does NCLX affect synapsin function indirectly, for example by altering calcium-dependent phosphorylation? Do NCLX-KO mice exhibit deficits in presynaptic LTP as well as in postsynaptic LTP? Some of these questions will be addressed in the future in our lab.

111

References

Agrawal A, Pekkurnaz G, Koslover EF. Spatial control of neuronal metabolism through glucose-mediated mitochondrial transport regulation. Elife 7: 1–27, 2018. Åkerman KEO. Changes in membrane potential during calcium ion influx and efflux across the mitochondrial membrane. BBA - Bioenerg 502: 359–366, 1978. Alabi ARA, Tsien RW. Synaptic vesicle pools and dynamics. Cold Spring Harb Perspect Biol 4: a013680–a013680, 2012. Alford S, Frenguellit BG, Schofieldt JG, Collingridge GL. Characterization of Calcium signals induced in hippocampal CA1 neurones by the synaptic activation of NMDA receptors. J Physiol 469: 693–716, 1993. Ashrafi G, de Juan-Sanz J, Farrell RJ, Ryan TA. Molecular Tuning of the Axonal Mitochondrial Ca2+ Uniporter Ensures Metabolic Flexibility of Neurotransmission. Neuron 105: 1–10, 2019. Ashrafi G, Ryan TA. Glucose metabolism in nerve terminals. Curr Opin Neurobiol 45: 156–161, 2017. Atluri PP, Regehr WG. Determinants of the time course of facilitation at the granule cell to Purkinje cell synapse. J Neurosci 16: 5661–5671, 1996. Attwell D, Laughlin SB. An energy budget for signaling in the grey matter of the brain. J Cereb Blood Flow Metab 21: 1133–1145, 2001. Augustine GJ, Santamaria F, Tanaka K. Local Calcium Signaling in Neurons Review. Neuron 40: 331–346, 2003. Baker PF. Transport and metabolism of calcium ions in nerve. Prog Biophys Mol Biol 24: 177– 223, 1972. Barbieri R, Contestabile A, Ciardo MG, Forte N, Marte A, Baldelli P, Benfenati F, Onofri F. Synapsin I and Synapsin II regulate neurogenesis in the dentate gyrus of adult mice. Oncotarget 9: 18760–18774, 2018. Barco A, Alarcon JM, Kandel ER. Expression of constitutively active CREB protein facilitates the late phase of long-term potentiation by enhancing synaptic capture. Cell 108: 689–703, 2002. Baughman JM, Perocchi F, Girgis HS, Plovanich M, Belcher-Timme CA, Sancak Y, Bao XR, Strittmatter L, Goldberger O, Bogorad RL, Koteliansky V, Mootha VK. Integrative genomics identifies MCU as an essential component of the mitochondrial calcium uniporter. Nature 476: 341–345, 2011. Baysal K, Jung DW, Gunter KK, Gunter TE, Brierley GP. Sodium-dependent Calcium efflux

112 mechanism of heart mitochondria is not a passive: Ca 2+ /2Na + exchanger. Am J Physiol - Cell Physiol 266, 1994. Benfenati F, Greengard P, Brunner J, Bahler M. Electrostatic and hydrophobic interactions of synapsin I and synapsin I fragments with phospholipid bilayers. J Cell Biol 108: 1851–1862, 1989. Benfenati F, Onofri F, Giovedi S. Protein-protein interactions and protein modules in the control of neurotransmitter release. Philos Trans R Soc B Biol Sci 354: 243–257, 1999. Bergsman JB, Krueger SR, Fitzsimonds RM. Automated criteria-based selection and analysis of fluorescent synaptic puncta. J Neurosci Methods 152: 32–39, 2006. Berridge MJ, Bootman MD, Roderick HL. Calcium signalling: Dynamics, homeostasis and remodelling. Nat Rev Mol Cell Biol 4: 517–529, 2003. Billups B, Forsythe ID. Presynaptic mitochondrial calcium sequestration influences transmission at mammalian central synapses. J Neurosci 22: 5840–5847, 2002. Birks R, MacIntosh FC. Acetylcholine metabolism of a sympathetic ganglion. Biochem Cell Biol 39: 787–827, 1961. Bliss TVP, Collingridge GL. A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361: 31–39, 1993. Bliss TVP, Lømo T. Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. J Physiol 232: 331–356, 1973. Böhme MA, Grasskamp AT, Walter AM. Regulation of synaptic release-site Ca2+ channel coupling as a mechanism to control release probability and short-term plasticity. FEBS Lett 592: 3516–3531, 2018. Boido D, Farisello P, Cesca F, Ferrea E, Valtorta F, Benfenati F, Baldelli P. Cortico-hippocampal hyperexcitability in synapsin I/II/III knockout mice: Age-dependency and response to the antiepileptic drug levetiracetam. Neuroscience 171: 268–283, 2010. Bragadin M, Pozzan T, Azzone GF. Kinetics of calcium(2+) ion carrier in rat liver mitochondria. Biochemistry 18: 5972–5978, 1979. Branco T, Marra V, Staras K. Examining size-strength relationships at hippocampal synapses using an ultrastructural measurement of synaptic release probability. J Struct Biol 172: 203– 210, 2010. Brautigam CA, Chelliah Y, Deisenhofer J. Tetramerization and ATP Binding by a Protein Comprising the A, B, and C Domains of Rat Synapsin I. J Biol Chem 279: 11948–11956, 2004. Brini M, Calì T, Ottolini D, Carafoli E. Neuronal calcium signaling: Function and dysfunction.

113

Cell Mol Life Sci 71: 2787–2814, 2014. Britti E, Delaspre F, Tamarit J, Ros J. Mitochondrial calcium signalling and neurodegenerative diseases. Neuronal Signal 2: NS20180061, 2018. Brockhaus J, Brüggen B, Missler M. Imaging and Analysis of Presynaptic Calcium Influx in Cultured Neurons Using synGCaMP6f. Front Synaptic Neurosci 11: 1–12, 2019. Browning MD, Dudek EM. Activators of protein kinase C increase the phosphorylation of the synapsins at sites phosphorylated by cAMP‐dependent and Ca2+/calmodulin‐dependent protein kinases in the rat hippocampal slice. Synapse 10: 62–70, 1992. Burrone J, Li Z, Murthy VN. Studying vesicle cycling in presynaptic terminals using the genetically encoded probe synaptopHluorin. Nat Protoc 1: 2970–2978, 2007. Bykhovskaia M. Synapsin regulation of vesicle organization and functional pools. Semin Cell Dev Biol 22: 387–392, 2011. Cai X, Lytton J. The cation/calcium exchanger superfamily: Phylogenetic analysis and structural implications. Mol Biol Evol 21: 1692–1703, 2004. Carafoli E. Mitochondrial uptake of calcium ions and the regulation of cell function. Biochem Soc Symp 39: 89–109, 1974. Carafoli E, Santella L, Branca D, Brini M. Generation, control, and processing of cellular calcium signals. Crit Rev Biochem Mol Biol 36: 107–260, 2001. del Castillo J, Katz B. Quantal components of the end‐plate potential. J Physiol 124: 560–573, 1954. Castillo PE. Presynaptic LTP and LTD of excitatory and inhibitory synapses. Cold Spring Harb Perspect Biol 4, 2012. Cesca F, Baldelli P, Valtorta F, Benfenati F. The synapsins: Key actors of synapse function and plasticity. Prog Neurobiol 91: 313–348, 2010. Chamberland S, Tóth K. Functionally heterogeneous synaptic vesicle pools support diverse synaptic signalling. J Physiol 594: 825–835, 2016. Chanaday NL, Cousin MA, Milosevic I, Watanabe S, Morgan JR. The Synaptic Vesicle Cycle Revisited: New Insights into the Modes and Mechanisms. J Neurosci 39: 8209–8216, 2019. Chang JY, Nakahata Y, Hayano Y, Yasuda R. Mechanisms of Ca2+/calmodulin-dependent kinase II activation in single dendritic spines. Nat Commun 10: 1–12, 2019. Charlton MP, Bittner GD. Facilitation of transmitter release at squid synapses. J Gen Physiol 72: 471–86, 1978. Charlton MP, Smith SJ, Zucker RS. Role of presynaptic calcium ions and channels in synaptic facilitation and depression at the squid giant synapse. J Physiol 323: 173–193, 1982.

114

Chater TE, Goda Y. The role of AMPA receptors in postsynaptic mechanisms of synaptic plasticity. Front Cell Neurosci 8: 1–14, 2014. Chaudhuri D, Sancak Y, Mootha VK, Clapham DE. MCU encodes the pore conducting mitochondrial calcium currents. Elife 2013: 4–11, 2013. Chavan V, Willis J, Walker SK, Clark HR, Liu X, Fox MA, Srivastava S, Mukherjee K. Central presynaptic terminals are enriched in ATP but the majority lack mitochondria. PLoS One 10: 1–19, 2015. Chen T, Wardill TJ, Sun Y, Pulver SR, Renninger SL, Baohan A, Schreiter ER, Kerr RA, Orger MB, Jayaraman V, Looger LL, Svoboda K, Kim DS. Ultra-sensitive fluorescent proteins for imaging neuronal activity. Nature 499: 295–300, 2013. Chi P, Greengard P, Ryan TA. Synapsin dispersion and reclustering during synaptic activity. Nat Neurosci 4: 1187–1193, 2001. Chi P, Greengard P, Ryan TA. Synaptic vesicle mobilization is regulated by distinct synapsin I phosphorylation pathways at different frequencies. Neuron 38: 69–78, 2003. Coleman WL, Bykhovskaia M. Synapsin I accelerates kinetics of neurotransmitter release in mouse motor terminals William. Synapse 63: 531–533, 2016. Collins TJ, Lipp P, Berridge MJ, Bootman MD. Mitochondrial Ca2+ Uptake Depends on the Spatial and Temporal Profile of Cytosolic Ca2+ Signals. J Biol Chem 276: 26411–26420, 2001. Cousin MA, Hurst H, Nicholls DG. Presynaptic calcium channels and field-evoked transmitter exocytosis from cultured cerebellar granule cells. Neuroscience 81: 151–61, 1997. Cremona O, De Camilli P. Synaptic vesicle endocytosis. Curr Opin Neurobiol 7: 323–330, 1997. Darcy KJ, Staras K, Collinson LM, Goda Y. Constitutive sharing of recycling synaptic vesicles between presynaptic boutons. Nat Neurosci 9: 315–321, 2006. Denker A, Rizzoli SO. Synaptic vesicle pools: An update. Front. Synaptic Neurosci. 2010. Devine MJ, Kittler JT. Mitochondria at the neuronal presynapse in health and disease. Nat Rev Neurosci 19: 63–80, 2018. Drago I, Pizzo P, Pozzan T. After half a century mitochondrial calcium in-and efflux machineries reveal themselves. EMBO J 30: 4119–4125, 2011. Duchen MR. Calcium-dependent changes in the mitochondrial energetics in single dissociated mouse sensory neurons. Biochem J 283: 41–50, 1992. Duchen MR. Contributions of mitochondria to animal physiology: from homeostatic sensor to calcium signalling and cell death. J Physiol 516 ( Pt 1: 1–17, 1999. Duchen MR. Mitochondria and calcium: From cell signalling to cell death. J Physiol 529: 57– 68, 2000.

115

Dyck BA, Skoblenick KJ, Castellano JM, Ki K, Thomas N, Mishra RK. Behavioral abnormalities in synapsin II knockout mice implicate a causal factor in schizophrenia. Synapse 63: 662–672, 2009. Esser L, Wang CR, Hosaka M, Smagula CS, C.südhof T, Deisenhofer J. Synapsin I is structurally similar to ATP-utilizing enzymes. EMBO J 17: 977–984, 1998. Feliciano P, Matos H, Andrade R, Bykhovskaia M. Synapsin II regulation of GABAergic synaptic transmission is dependent on interneuron subtype. J Neurosci 37: 1757–1771, 2017. Ferreira A, Han HQ, Greengard P, Kosik KS. Suppression of synapsin II inhibits the formation and maintenance of synapses in hippocampal culture. Proc Natl Acad Sci U S A 92: 9225–9229, 1995. Fornasiero EF, Raimondi A, Guarnieri FC, Orlando M, Fesce R, Benfenati F, Valtorta F. Synapsins Contribute to the Dynamic Spatial Organization of Synaptic Vesicles in an Activity- Dependent Manner. J Neurosci Off J Soc Neurosci Neurosci 32: 12214–12227, 2012. Foskett JK, White C, Cheung KH, Mak DOD. Inositol trisphosphate receptor Ca2+ release channels. Physiol Rev 87: 593–658, 2007. Fourcaudot E, Gambino F, Humeau Y, Casassus G, Shaban H, Poulain B, Lüthi A. cAMP/PKA signaling and RIM1α mediate presynaptic LTP in the lateral amygdala. Proc Natl Acad Sci 105: 15130 LP – 15135, 2008. Fowler MW, Staras K. Synaptic vesicle pools: Principles, properties and limitations. Exp Cell Res 335: 150–156, 2015. Gabriel T, García-Pérez E, Mahfooz K, Goñi J, Martínez-Turrillas R, Pérez-Otaño I, Lo DC, Wesseling JF. A new kinetic framework for synaptic vesicle trafficking tested in synapsin knock-outs. J Neurosci 31: 11563–11577, 2011. Gaffield MA, Betz WJ. Imaging synaptic vesicle exocytosis and endocytosis with FM dyes. Nat Protoc 1: 2916–2921, 2006. Garcia-Perez E, Lo DC, Wesseling JF. Kinetic isolation of a slowly recovering component of short-term depression during exhaustive use at excitatory hippocampal synapses. J Neurophysiol 100: 781–795, 2008. Garcia CC, Blair HJ, Seager M, Coulthard A, Tennant S, Buddles M, Curtis A, Goodship JA. Identification of a mutation in synapsin I, a synaptic vesicle protein, in a family with epilepsy. J Med Genet 41: 183–187, 2004. Gautheron DC. Mitochondrial Oxidative Phosphorylation and Respiratory Chain: Review. J Inherit Metab Dis 7: 57–61, 1984. Gitler D, Cheng Q, Greengard P, Augustine GJ. Synapsin IIa Controls the Reserve Pool of

116

Glutamatergic Synaptic Vesicles. J Neurosci 28: 10835–10843, 2008. Gitler D, Takagishi Y, Feng J, Ren Y, Rodriguiz RM, Wetsel WC, Greengard P, Augustine GJ. Different presynaptic roles of synapsins at excitatory and inhibitory synapses. J Neurosci 24: 11368–11380, 2004a. Gitler D, Xu Y, Kao H-T, Lin D, Lim S, Feng J, Greengard P, Augustine GJ. Molecular determinants of synapsin targeting to presynaptic terminals. J Neurosci 24: 3711–20, 2004b. Granseth B, Odermatt B, Royle SJ, Lagnado L. Clathrin-Mediated Endocytosis Is the Dominant Mechanism of Vesicle Retrieval at Hippocampal Synapses. Neuron 51: 773–786, 2006. Greengard P, Valtorta F, Czernik AJ, Benfenati F. Synaptic vesicle phosphoproteins and regulation of synaptic function. Science (80- ) 259: 780 LP – 785, 1993. Groh A, De Kock CPJ, Wimmer VC, Sakmann B, Kuner T. Driver or coincidence detector: Modal switch of a corticothalamic giant synapse controlled by spontaneous activity and short-term depression. J Neurosci 28: 9652–9663, 2008. Gundlfinger A, Breustedt J, Sullivan D, Schmitz D. Natural spike trains trigger short- and long lasting dynamics at hippocampal mossy fiber synapses in rodents. PLoS One 5: 1–9, 2010. Gustafsson B, Ma R, Hanse E. The Small and Dynamic Pre-primed Pool at the Release Site; A Useful Concept to Understand Release Probability and Short-Term Synaptic Plasticity? [Online]. Front. Synaptic Neurosci. 11: 7, 2019.https://www.frontiersin.org/article/10.3389/fnsyn.2019.00007. Halestrap AP, Davidson AM. Inhibition of Ca2+-induced large-amplitude swelling of liver and heart mitochondria by cyclosporin is probably caused by the inhibitor binding to mitochondrial-matrix peptidyl-prolyl cis-trans isomerase and preventing it interacting with the adenine nucle. Biochem J 268: 153–160, 1990. Hall CN, Klein-Flügge MC, Howarth C, Attwell D. Oxidative phosphorylation, not glycolysis, powers presynaptic and postsynaptic mechanisms underlying brain information processing. J Neurosci 32: 8940–8951, 2012. Harris JJ, Jolivet R, Attwell D. Synaptic Energy Use and Supply. Neuron 75: 762–777, 2012. Hebb DO. The organisation of behaviour: a neuropsychological theory. Wiley New York, 1952. Herring BE, Nicoll RA. Long-Term Potentiation: From CaMKII to AMPA Receptor Trafficking. Annu Rev Physiol 78: 351–365, 2016. Herrington J, Park YB, Babcock DF, Hille B. Dominant role of mitochondria in clearance of large Ca2+ loads from rat adrenal chromaffin cells. Neuron 16: 219–228, 1996. Heytler PG, Prichard WW. A new class of uncoupling agents - Carbonyl cyanide phenylhydrazones. Biochem Biophys Res Commun 7: 272–275, 1962.

117

Hilfiker S, Benfenati F, Doussau F, Nairn AC, Czernik AJ, Augustine GJ, Greengard P. Structural domains involved in the regulation of transmitter release by synapsins. J Neurosci 25: 2658– 2669, 2005. Hilfïker S, Schweizer FE, Kao HT, Czernik AJ, Greengard P, Augustine GJ. Two sites of action for synapsin domain E in regulating neurotransmitter release. Nat Neurosci 1: 29–35, 1998. Hirabayashi Y, Kwon SK, Paek H, Pernice WM, Paul MA, Lee J, Erfani P, Raczkowski A, Petrey DS, Pon LA, Polleux F. ER-mitochondria tethering by PDZD8 regulates Ca2+ dynamics in mammalian neurons. Science (80- ) 358: 623–630, 2017. Horikawa K. Recent progress in the development of genetically encoded Ca2+indicators. J Med Investig 62: 24–28, 2015. Hosaka M, Hammer RE, Su TC. A Phospho-Switch Controls the Dynamic Association.pdf. Neuron 24: 377–387, 1999. Hosaka M, Südhof TC. Synapsins I and II are ATP-binding proteins with differential calcium regulation. J Biol Chem 273: 1425–1429, 1998a. Hosaka M, Südhof TC. Synapsin III, a novel synapsin with an unusual regulation by Calcium. J Biol Chem 273: 13371–13374, 1998b. Hosaka M, Südhof TC. Homo- and heterodimerization of synapsins. J Biol Chem 274: 16747– 16753, 1999. Hubley MJ, Locke BR, Moerland TS. The effects of temperature, pH, and magnesium on the diffusion coefficient of ATP in solutions of physiological ionic strength. Biochim Biophys Acta - Gen Subj 1291: 115–121, 1996. Hubley MJ, Moerland TS, Rosanske RC. Diffusion coefficients of atp and creatine phosphate in isolated muscle: pulsed gradient 31p nmr of small biological samples. NMR Biomed 8: 72– 78, 1995. Humeau Y, Doussau F, Vitiello F, Greengard P, Benfenati F, Poulain B. Synapsin controls both reserve and releasable synaptic vesicle pools during neuronal activity and short-term plasticity in Aplysia. J Neurosci 21: 4195–4206, 2001. Igbavboa U, Pfeiffer DR. EGTA Inhibits Reverse Uniport-dependent Ca2+ Release from Uncoupled Mitochondria. J Biol Chem 263: 1405–1412, 1988. Imamura H, Huynh Nhat KP, Togawa H, Saito K, Iino R, Kato-Yamada Y, Nagai T, Noji H. Visualization of ATP levels inside single living cells with fluorescence resonance energy transfer-based genetically encoded indicators. Proc Natl Acad Sci 106: 15651–15656, 2009. Jackman SL, Regehr WG. The Mechanisms and Functions of Synaptic Facilitation. Neuron 94: 447–464, 2017.

118

Jackman SL, Turecek J, Belinsky JE, Regehr WG. The calcium sensor synaptotagmin 7 is required for synaptic facilitation. Nature 529: 88–91, 2016. Jae SL, Kim MH, Ho WK, Lee SH. Presynaptic release probability and readily releasable pool size are regulated by two independent mechanisms during posttetanic potentiation at the calyx of held synapse. J Neurosci 28: 7945–7953, 2008. Jahn R, Fasshauer D. Molecular machines governing exocytosis of synaptic vesicles. Nature 490: 201–207, 2012. Jalil SJ, Sacktor TC, Shouval HZ. Atypical PKCs in memory maintenance: The roles of feedback and redundancy. Learn Mem 22: 344–353, 2015. Jang SR, Nelson JC, Bend EG, Rodríguez-Laureano L, Tueros FG, Cartagenova L, Underwood K, Jorgensen EM, Colón-Ramos DA. Glycolytic Enzymes Localize to Synapses under Energy Stress to Support Synaptic Function. Neuron 90: 278–291, 2016. Johnston PA, Jahn R, Sudhof TC. Transmembrane topography and evolutionary conservation of synaptophysin. J Biol Chem 264: 1268–1273, 1989. Jovanovic JN, Sihra TS, Nairn AC, Hemmings HC, Greengard P, Czernik AJ. Opposing changes in phosphorylation of specific sites in synapsin I during Calcium-dependent glutamate release in isolated nerve terminals. J Neurosci 21: 7944–7953, 2001. Jung DW, Baysal K, Brierley GP. The sodium-calcium antiport of heart mitochondria is not electroneutral. J. Biol. Chem. 270: 672–678, 1995. Kaeser PS, Regehr WG. The readily releasable pool of synaptic vesicles. Curr Opin Neurobiol 43: 63–70, 2017. Kanaani J, Lissin D, Kash SF, Baekkeskov S. The hydrophilic isoform of glutamate decarboxylase, GAD67, is targeted to membranes and nerve terminals independent of dimerization with the hydrophobic membrane-anchored isoform, GAD65. J Biol Chem 274: 37200–37209, 1999. Kang J, Lim L, Song J. ATP enhances at low concentrations but dissolves at high concentrations liquid-liquid phase separation (LLPS) of ALS/FTD-causing FUS. Biochem Biophys Res Commun 504: 545–551, 2018. Kang JS, Tian JH, Pan PY, Zald P, Li C, Deng C, Sheng ZH. Docking of Axonal Mitochondria by Syntaphilin Controls Their Mobility and Affects Short-Term Facilitation. Cell 11: 137–148, 2008. Kao HT, Porton B, Hilfiker S, Stefani G, Pieribone VA, DeSalle R, Greengard P. Molecular evolution of the synapsin gene family. J Exp Zool 285: 360–377, 1999. Katz B, Miledi R. Spontaneous and evoked activity of motor nerve endings in calcium Ringer.

119

J Physiol 203: 689–706, 1969. Kay AR, Alfonso A, Alford S, Cline HT, Holgado AM, Sakmann B, Snitsarev VA, Stricker TP, Takahashi M, Wu L-G. Imaging Synaptic Activity in Intact Brain and Slices with FM1-43 in C. elegans, Lamprey, and Rat. Neuron 24: 809–817, 1999. Kerchner GA, Nicoll RA. Silent synapses and the emergence of a postsynaptic mechanism for LTP. Nat Rev Neurosci 9: 813–825, 2008. Kim SH, Ryan TA. CDK5 Serves as a Major Control Point in Neurotransmitter Release. Neuron 67: 797–809, 2010. Kim SH, Ryan TA. Balance of Calcineurin A and CDK5 Activities Sets Release Probability at Nerve Terminals. J Neurosci 33: 8937–8950, 2013. Kirichok Y, Krapivinsky G, Clapham DE. The mitochondrial calcium uniporter is a highly selective ion channel. Nature 427: 360–364, 2004. Klenchin VA, Martin TFJ. Priming in exocytosis: Attaining fusion-competence after vesicle docking. Biochimie 82: 399–407, 2000. Korogod N, Lou X, Schneggenburger R. Presynaptic Ca2+ requirements and developmental regulation of posttetanic potentiation at the calyx of Held. J Neurosci 25: 5127–5137, 2005. Kostic M, Katoshevski T, Sekler I. Allosteric Regulation of NCLX by Mitochondrial Membrane Potential Links the Metabolic State and Ca2+ Signaling in Mitochondria. Cell Rep 25: 3465- 3475.e4, 2018. Kügler S, Kilic E, Bähr M. Human synapsin 1 gene promoter confers highly neuron-specific long-term transgene expression from an adenoviral vector in the adult rat brain depending on the transduced area. Gene Ther 10: 337–347, 2003. Kuiper JWP, Oerlemans FTJJ, Fransen JAM, Wieringa B. Creatine kinase B deficient neurons exhibit an increased fraction of motile mitochondria. BMC Neurosci 9: 1–10, 2008. Kuromi H, Honda A, Kidokoro Y. Ca2+ Influx through Distinct Routes Controls Exocytosis and Endocytosis at Drosophila Presynaptic Terminals. Neuron 41: 101–111, 2004. Laviv T, Riven I, Dolev I, Vertkin I, Balana B, Slesinger PA, Slutsky I. Basal GABA regulates GABABR conformation and release probability at single hippocampal synapses. Neuron 67: 253–267, 2010. Lee SH, Schwaller B, Neher E. Kinetics of Ca2+ binding to parvalbumin in bovine chromaffin cells: Implications for [Ca2+] transients of neuronal dendrites. J Physiol 525: 419–432, 2000. Levy S, Wilms CD, Brumer E, Kahn J, Pnueli L, Arava Y, Eilers J, Gitler D. SpRET: Highly sensitive and reliable spectral measurement of absolute FRET efficiency. Microsc Microanal 17: 176– 190, 2011.

120

Li H, Chen Y, Jones AF, Sanger RH, Collis LP, Flannery R, Mcnay EC, Yu T, Schwarzenbacher R, Bossy B, Bossy-Wetzel E, Bennett MVL, Pypaert M, Hickman JA, Smith PJS, Hardwick JM, Jonas EA. Bcl-x L induces Drp1-dependent synapse formation in cultured hippocampal neurons. PNAS 105: 2169–2174, 2008. Li Z, Murthy VN. Visualizing postendocytic traffic of synaptic vesicles at hippocampal synapses [Online]. Neuron 31: 593–605, 2001papers3://publication/uuid/A6309E80-224D-4D0F-9B37- 1F4AB3612019. Liao D, Hessler NA, Malinow R. Activation of postsynaptically silent synapses during pairing- induced LTP in CA1 region of hippocampal slice. Nature 375: 400–404, 1995. Lisman J, Yasuda R, Raghavachari S. Mechanisms of CAMKII action in long-term potentiation. Nat Rev Neurosci 13: 169–182, 2014. Litsky ML, Pfeiffer DR. Regulation of the mitochondrial Ca2+ uniporter by external adenine nucleotides: The uniporter behaves like a gated channel which is regulated by nucleotides and divalent cations. Biochemistry 36: 7071–7080, 1997. Llinas R, McGuinness TL, Leonard CS, Sugimori M, Greengard P. Intraterminal injection of synapsin I or calcium/calmodulin-dependent protein kinase II alters neurotransmitter release at the squid giant synapse. Proc. Natl. Acad. Sci. U. S. A. 82: 3035–3039, 1985. Llinas R, Sugimori M, Silver RB. Microdomains of high calcium concentration in a presynaptic terminal. Science (80- ) 256: 677 LP – 679, 1992. Lu S, Huang W, Wang Q, Shen Q, Li S, Nussinov R, Zhang J. The Structural Basis of ATP as an Allosteric Modulator. PLoS Comput Biol 10, 2014. Lu YF, Kandel ER, Hawkins RD. Nitric oxide signaling contributes to late-phase LTP and CREB phosphorylation in the hippocampus. J Neurosci 19: 10250–10261, 1999. Luciani DS, Ao P, Hu X, Warnock GL, Johnson JD. Voltage-gated Ca2+ influx and insulin secretion in human and mouse β-cells are impaired by the mitochondrial Na+/Ca2+ exchange inhibitor CGP-37157. Eur J Pharmacol 576: 18–25, 2007. Lukas W, Jones KA. Cortical neurons containing calretinin are selectively resistant to calcium overload and excitotoxicity in vitro. Neuroscience 61: 307–316, 1994. Lukyanetz IA, Kostyuk PG, Lukyanetz EA. The involvement of calcium transport systems of the plasma membrane in calcium exchange in neurons of the carassius gibelio cerebellum. Neurophysiology 41: 231–237, 2009. Luongo TS, Lambert JP, Gross P, Nwokedi M, Lombardi AA, Shanmughapriya S, Carpenter AC, Kolmetzky D, Gao E, Van Berlo JH, Tsai EJ, Molkentin JD, Chen X, Madesh M, Houser SR, Elrod JW. The mitochondrial Na+/Ca2+ exchanger is essential for Ca2+ homeostasis and

121 viability. Nature 545: 93–97, 2017. Lynch G, Kramár EA, Gall CM. Protein synthesis and consolidation of memory-related synaptic changes. Brain Res 1621: 62–72, 2015. Lynch G, Larson J, Kelso S, Barrionuevo G, Schottler F. Intracellular injections of EGTA block induction of hippocampal long-term potentiation. Nature 305: 719–721, 1983. MacAskill AF, Atkin TA, Kittler JT. Mitochondrial tracking and the provision of energy and calcium buffering at excitatory synapses. Eur J Neurosci 32: 231–240, 2010. Madison D V., Malenka RC, Nicoll RA. Mechanisms Underlying Long-Term Potentiation Of Synaptic Transmission. Annu Rev Neurosci 14: 379–397, 1991. Malenka RC, Bear MF. LTP and LTD: An embarrassment of riches. Neuron 44: 5–21, 2004. Maltsev A V., Bal N V., Balaban PM. LTP suppression by protein synthesis inhibitors is NO- dependent. Neuropharmacology 146: 276–288, 2019. Marra V, Burden JJ, Thorpe JR, Smith IT, Smith SL, Häusser M, Branco T, Staras K. A Preferentially Segregated Recycling Vesicle Pool of Limited Size Supports Neurotransmission in Native Central Synapses. Neuron 76: 579–589, 2012. Martin KC, Michael D, Rose JC, Barad M, Casadio A, Zhu H, Kandel ER. MAP kinase translocates into the nucleus of the presynaptic cell and is required for long-term facilitation in Aplysia. Neuron 18: 899–912, 1997. Matthews EA, Schoch S, Dietrich D. Tuning local calcium availability: Cell-type-specific immobile calcium buffer capacity in hippocampal neurons. J Neurosci 33: 14431–14445, 2013. McCormack JG, Halestrap AP, Denton RM. Role of calcium ions in regulation of mammalian intramitochondrial metabolism. Physiol Rev 70: 391–425, 1990. Medrihan L, Cesca F, Raimondi A, Lignani G, Baldelli P, Benfenati F. Synapsin II desynchronizes neurotransmitter release at inhibitory synapses by interacting with presynaptic calcium channels. Nat Commun 4: 1512–1513, 2013. Medrihan L, Ferrea E, Greco B, Baldelli P, Benfenati F. Asynchronous GABA release is a key determinant of tonic inhibition and controls neuronal excitability: A study in the synapsin II-/- mouse. Cereb Cortex 25: 3356–3368, 2015. Menegon A, Bonanomi D, Albertinazzi C, Lotti F, Ferrari G, Kao HT, Benfenati F, Baldelli P, Valtorta F. Protein kinase A-mediated synapsin I phosphorylation is a central modulator of Ca2+-dependent synaptic activity. J Neurosci 26: 11670–11681, 2006. Milovanovic D, Wu Y, Bian X, De Camilli P. A Liquid Phase of Synapsin and Lipid Vesicles. Science (80- ) 361: 604–607, 2018. Mitchell P, Moyle J. Chemiosmotic Hypothesis of Oxidative Phosphorylation. Nature 213:

122

137–139, 1967. Mochida S. Presynaptic calcium channels. Neurosci Res 127: 33–44, 2018. Montero M, Alonso MT, Albillos A, García-Sancho J, Alvarez J. Mitochondrial Ca2+-induced Ca2+ release mediated by the Ca2+ uniporter. Mol Biol Cell 12: 63–71, 2001. Morimoto K, Sato K, Sato S, Yamada N, Hayabara T. Time-dependent changes in rat hippocampal synapsin I mRNA expression during long-term potentiation. Brain Res 783: 57– 62, 1998. Müller M, Felmy F, Schwaller B, Schneggenburger R. Parvalbumin is a mobile presynaptic Ca2+ buffer in the calyx of held that accelerates the decay of Ca2+ and short-term facilitation. J Neurosci 27: 2261–2271, 2007. Murthy VN, Camilli P De. Cell Biology of the Presynaptic Terminal. Annu Rev Neurosci 26: 701– 728, 2003. Nakai J, Ohkura M, Imoto K. A high signal-to-noise ca2+ probe composed of a single green fluorescent protein. Nat Biotechnol 19: 137–141, 2001. Narita K, Akita T, Hachisuka J, Huang SM, Ochi K, Kuba K. Functional coupling of Ca2+ channels to ryanodine receptors at presynaptic terminals: Amplification of exocytosis and plasticity. J Gen Physiol 115: 519–532, 2000. Nayak AS, Moore CI, Browning MD. Ca2+/calmodulin-dependent protein kinase II phosphorylation of the presynaptic protein synapsin I is persistently increased during long- term potentiation. Proc Natl Acad Sci U S A 93: 15451–15456, 1996. Neher E. Introduction: Regulated exocytosis. Cell Calcium 52: 196–198, 2012. Neher E. Merits and Limitations of Vesicle Pool Models in View of Heterogeneous Populations of Synaptic Vesicles. Neuron 87: 1131–1142, 2015. Neher E, Brose N. Dynamically Primed Synaptic Vesicle States: Key to Understand Synaptic Short-Term Plasticity. Neuron 100: 1283–1291, 2018. Neumann JT, Diaz-Sylvester PL, Fleischer S, Copello JA. CGP-37157 inhibits the sarcoplasmic reticulum Ca 2+ ATPase and activates ryanodine receptor channels in striated muscle. Mol Pharmacol 79: 141–147, 2011. Nguyen P V., Marin L, Atwood HL. Synaptic physiology and mitochondrial function in crayfish tonic and phasic motor neurons. J Neurophysiol 78: 281–294, 1997. Nicholls DG, Budd SL. Mitochondria and Neuronal Survival [Online]. Physiol Rev 80: 315–360, 2000http://physrev.physiology.org/content/80/1/315.abstract. Nicoll RA, Schmitz D. Synaptic plasticity at hippocampal mossy fibre synapses. Nat Rev Neurosci 6: 863–876, 2005.

123

Nita LI, Hershfinkel M, Sekler I. Life after the birth of the mitochondrial Na+/Ca2+ exchanger, NCLX. Sci China Life Sci 58: 59–65, 2015. Ohana E, Elgazar V, Palty R, Beharier O, Hershfinkel M, Volokita M, Sekler I, Argaman M, Silverman WF. Lithium-Calcium Exchange Is Mediated by a Distinct Potassium-independent Sodium-Calcium Exchanger. J Biol Chem 279: 25234–25240, 2004. Orenbuch A, Shalev L, Marra V, Sinai I, Lavy Y, Kahn J, Burden JJ, Staras K, Gitler D. Synapsin Selectively Controls the Mobility of Resting Pool Vesicles at Hippocampal Terminals. J Neurosci 32: 3969–3980, 2012a. Orenbuch A, Shalev L, Marra V, Sinai I, Lavy Y, Kahn J, Burden JJ, Staras K, Gitler D. Synapsin Selectively Controls the Mobility of Resting Pool Vesicles at Hippocampal Terminals. J Neurosci 32: 3969–3980, 2012b. Orenbuch A, Shulman Y, Lipstein N, Bechar A, Levy Y, Brumer E, Vasilevs M, Kahn J, Barki- Harrington L, Kuner T, Gitler D. Inhibition of exocytosis or endocytosis blocks activity- dependent redistribution of synapsin. J Neurochem 120: 248–258, 2012c. Orlando M, Lignani G, Maragliano L, Fassio A, Onofri F, Baldelli P, Giovedí S, Benfenati F. Functional role of ATP binding to synapsin I in synaptic vesicle trafficking and release dynamics. J Neurosci 34: 14752–14768, 2014. Palty R, Silverman WF, Hershfinkel M, Caporale T, Sensi SL, Parnis J, Nolte C, Fishman D, Shoshan-Barmatz V, Herrmann S, Khananshvili D, Sekler I. NCLX is an essential component of mitochondrial Na+/Ca2+ exchange. Proc Natl Acad Sci 107: 436–441, 2010. Pan B, Zucker RS. A General Model of Synaptic Transmission and Short-Term Plasticity. Neuron 62: 539–554, 2009. Parfitt KD, Hoffer BJ, Browning MD. Norepinephrine and isoproterenol increase the phosphorylation of synapsin I and synapsin II in dentate slices of young but not aged Fisher 344 rats. Proc Natl Acad Sci U S A 88: 2361–2365, 1991. Patel A, Malinovska L, Saha S, Wang J, Alberti S, Krishnan Y, Hyman AA. Biochemistry: ATP as a biological hydrotrope. Science (80- ) 356: 753–756, 2017. Pathak D, Shields LY, Mendelsohn BA, Haddad D, Lin W, Gerencser AA, Kim H, Brand MD, Edwards RH, Nakamura K. The role of mitochondrially derived ATP in synaptic vesicle recycling. J Biol Chem , 2015. doi:10.1074/jbc.M115.656405. Perkel DJ, Petrozzino JJ, Nicoll RA, Connor JA. The role of Ca2+ entry via synaptically activated NMDA receptors in the induction of long-term potentiation. Neuron 11: 817–823, 1993. Perkins GA, Tjong J, Brown JM, Poquiz PH, Scott RT, Kolson DR, Ellisman MH, Spirou GA. The micro-architecture of mitochondria at active zones: Electron tomography reveals novel

124 anchoring scaffolds and cristae structured for high-rate metabolism. J Neurosci 30: 1015– 1026, 2010. Perocchi F, Gohil VM, Girgis HS, Bao XR, McCombs JE, Palmer AE, Mootha VK. MICU1 encodes a mitochondrial EF hand protein required for Ca2+ uptake. Nature 467: 291–296, 2010. Pfeiffer T, Schuster S, Bonhoeffer S. Erratum: Cooperation and competition in the evolution of ATP-producing pathways (Science (April 20) (504)). Science (80- ) 293: 1436, 2001. Plášek J, Vojtíšková A, Houštěek J. Flow-cytometric monitoring of mitochondrial depolarisation: From fluorescence intensities to millivolts. J Photochem Photobiol B Biol 78: 99–108, 2005. Rangaraju V, Calloway N, Ryan TA. Activity-driven local ATP synthesis is required for synaptic function. Cell , 2014. doi:10.1016/j.cell.2013.12.042. Rangaraju V, Lauterbach M, Schuman EM. Spatially Stable Mitochondrial Compartments Fuel Local Translation during Plasticity. Cell 176: 73-84.e15, 2019a. Rangaraju V, Lewis TL, Hirabayashi Y, Bergami M, Motori E, Cartoni R, Kwon SK, Courchet J. Pleiotropic Mitochondria: The Influence of Mitochondria on Neuronal Development and Disease. J Neurosci 39: 8200–8208, 2019b. Regehr WG. Short-term presynaptic plasticity. Cold Spring Harb Perspect Biol 4: 1–19, 2012. Regehr WG, Abbott LF. Synaptic computation. Nature 431: 796–803, 2004. Regehr WG, Delaney KR, Tank DW. The role of presynaptic calcium in short-term enhancement at the hippocampal mossy fiber synapse. J Neurosci 14: 523–537, 1994. RENGEL Z. The role of calcium in salt toxicity. Plant. Cell Environ. 15: 625–632, 1992. Rizzoli SO, Betz WJ. Synaptic vesicle pools. Nat Rev Neurosci 6: 57–69, 2005. De Robertis ED, Bennett HS. Some features of the submicroscopic morphology of synapses in frog and earthworm. J Biophys Biochem Cytol 1: 47–58, 1955. Roe MW, Lemasters JJ, Herman B. Assessment of Fura-2 for measurements of cytosolic free calcium. Cell Calcium 11: 63–73, 1990. Rosahl TW, Geppert M, Spillane D, Herz J, Hammer RE, Malenka RC, Südhof TC. Short-term synaptic plasticity is altered in mice lacking synapsin I. Cell 75: 661–670, 1993. Rosahl TW, Spillane D, Missler M, Herz J, Selig DK, Wolff JR, Hammer RE, Malenka RC, Südhof TC. Essential functions of synapsins I and II in synaptic vesicle regulation. Nature 375: 488– 493, 1995. Rosenberg T, Gal-Ben-Ari S, Dieterich DC, Kreutz MR, Ziv NE, Gundelfinger ED, Rosenblum K. The roles of protein expression in synaptic plasticity and memory consolidation. Front Mol Neurosci 7: 1–14, 2014.

125

Rosenmund C, Westbrook GL. Rundown of N‐methyl‐D‐aspartate channels during whole‐cell recording in rat hippocampal neurons: role of Ca2+ and ATP. J Physiol 470: 705–729, 1993. Rottenberg H, Scarpa A. Calcium uptake and membrane potential in mitochondria. Biochemistry 13: 4811–4817, 1974. Rowland KC, Irby NK, Spirou GA. Specialized synapse-associated structures within the calyx of Held. J Neurosci 20: 9135–44, 2000. Ryan TA, Li L, Chin LS, Greengard P, Smith SJ. Synaptic vesicle recycling in synapsin I knockout mice. J Cell Biol 134: 1219–1227, 1996. Ryan TA, Reuter H, Wendland B, Schweizer FE, Tsien RW, Smith SJ. The kinetics of synaptic vesicle recycling measured at single presynaptic boutons. Neuron 11: 713–724, 1993. Saitoh T, Schwartz JH. Phosphorylation-dependent subcellular translocation of a Ca2+/Calmodulin-dependent protein kinase produces an autonomous enyzme in Aplysia neurons. J Cell Biol 100: 835–842, 1985. Sancak Y, Markhard AL, Kitami T, Kovács-Bogdán E, Kamer KJ, Udeshi ND, Carr SA, Chaudhuri D, Clapham DE, Li AA, Calvo SE, Goldberger O, Mootha VK. EMRE is an essential component of the mitochondrial calcium uniporter complex. Science (80- ) 342: 1379–1382, 2013. Sankaranarayanan S, De Angelis D, Rothman JE, Ryan TA. The Use of pHluorins for Optical Measurements of Presynaptic Activity. Biophys J 79: 2199–2208, 2000. Sankaranarayanan S, Ryan TA. Real-time measurements of vesicle- SNARE recycling in synapses of the central nervous system. Nat Cell Biol 2: 197–204, 2000a. Sankaranarayanan S, Ryan TA. Real-time measurements of vesicle-SNARE recycling in synapses of the central nervous system. Nat Cell Biol 2: 197–204, 2000b. Santamaria F, Wils S, De Schutter E, Augustine GJ. Anomalous Diffusion in Purkinje Cell Dendrites Caused by Spines. Neuron 52: 635–648, 2006. Scaduto RC, Grotyohann LW. Measurement of mitochondrial membrane potential using fluorescent rhodamine derivatives. Biophys J 76: 469–477, 1999. Scheff SW, Price DA, Ansari MA, Roberts KN, Schmitt FA, Ikonomovic MD, Mufson EJ. Synaptic change in the posterior cingulate gyrus in the progression of Alzheimer’s disease. J Alzheimers Dis 43: 1073–1090, 2015. Schneggenburger R, Neher E. Presynaptic calcium and control of vesicle fusion. Curr Opin Neurobiol 15: 266–274, 2005. Sekler I, Fishman D, Shoshan-Barmatz V, Parnis J, Palty R, Caporale T, Hershfinkel M, Silverman WF, Herrmann S, Nolte C, Khananshvili D, Sensi SL. NCLX is an essential component of mitochondrial Na+/Ca2+ exchange. Proc Natl Acad Sci 107: 436–441, 2009.

126

Selivanov VA, Krause S, Roca J, Cascante M. Modeling of spatial metabolite distributions in the cardiac sarcomere. Biophys J 92: 3492–3500, 2007. Shahrezaei V, Delaney KR. Brevity of the Ca2+ microdomain and active zone geometry prevent Ca2+-sensor saturation for neurotransmitter release. J Neurophysiol 94: 1912–1919, 2005. Sheng ZH, Cai Q. Mitochondrial transport in neurons: Impact on synaptic homeostasis and neurodegeneration. Nat Rev Neurosci 13: 77–93, 2012. Shi S-H, Hayashi Y, Petralia RS, Zaman SH, Wenthold RJ, Svoboda K, Malinow R. Rapid Spine Delivery and Redistribution of AMPA Receptors After Synaptic NMDA Receptor Activation. Science (80- ) 284: 1811 LP – 1816, 1999. Shulman Y, Stavsky A, Fedorova T, Mikulincer D, Atias M, Radinsky I, Kahn J, Slutsky I, Gitler D. ATP Binding to Synaspsin IIa Regulates Usage and Clustering of Vesicles in Terminals of Hippocampal Neurons. J Neurosci 35: 985–98, 2015. Shupliakov O, Haucke V, Pechstein A. How synapsin I may cluster synaptic vesicles. Semin Cell Dev Biol 22: 393–399, 2011. Siksou L, Rostaing P, Lechaire JP, Boudier T, Ohtsuka T, Fejtová A, Kao HT, Greengard P, Gundelfinger ED, Triller A, Marty S. Three-dimensional architecture of presynaptic terminal cytomatrix. J Neurosci 27: 6868–6877, 2007. Silva AJ, Rosahl TW, Chapman PF, Marowitz Z, Friedman E, Frankland PW, Cestari V, Cioffi D, Südhof TC, Bourtchuladze R. Impaired learning in mice with abnormal short-lived plasticity. Curr Biol 6: 1509–1518, 1996. Smith HL, Bourne JN, Cao G, Chirillo MA, Ostroff LE, Watson DJ, Harris KM. Mitochondrial support of persistent presynaptic vesicle mobilization with age-dependent synaptic growth after LTP. Elife 5: 1–30, 2016. Sobieski C, Fitzpatrick MJ, Mennerick SJ. Differential Presynaptic ATP Supply for Basal and High-Demand Transmission. J Neurosci 37: 1888–1899, 2017. Song S-H, Augustine GJ. Synapsin Isoforms Regulating GABA Release from Hippocampal Interneurons. J Neurosci 36: 6742–57, 2016. Song X, Jensen MO, Jogini V, Stein RA, Lee CH, McHaourab HS, Shaw DE, Gouaux E. Mechanism of NMDA receptor channel block by MK-801 and memantine. Nature 556: 515– 519, 2018. Spillane DM, Rosahl TW, Südhof TC, Malenka RC. Long-term potentiation in mice lacking synapsins. Neuropharmacology 34: 1573–1579, 1995. Stanton PK, Schanne FAX. Hippocampal long-term potentiation increases mitochondrial

127 calcium pump activity in rat. Brain Res 382: 185–188, 1986. Staras K, Branco T, Burden JJ, Pozo K, Darcy K, Marra V, Ratnayaka A, Goda Y. A Vesicle Superpool Spans Multiple Presynaptic Terminals in Hippocampal Neurons. Neuron 66: 37–44, 2010. Staras K, Mikulincer D, Gitler D. Monitoring and quantifying dynamic physiological processes in live neurons using fluorescence recovery after photobleaching. J Neurochem 126: 213–222, 2013. De Stefani D, Raffaello A, Teardo E, Szabó I, Rizzuto R. A forty-kilodalton protein of the inner membrane is the mitochondrial calcium uniporter. Nature 476: 336–340, 2011. Stein A, Radhakrishnan A, Riedel D, Fasshauer D, Jahn R. Synaptotagmin activates membrane fusion through a calcium-dependent trans interaction with phospholipids. Nat Struct Mol Biol 14: 904–911, 2007. Stevens CF, Wesseling JF. Augmentation is a potentiation of the exocytotic process. Neuron 22: 139–146, 1999. Strehler E, Treiman M. Calcium Pumps of Plasma Membrane and Cell Interior. Curr Mol Med 4: 323–335, 2005. Südhof TC. the Synaptic Vesicle Cycle. Annu Rev Neurosci 27: 509–547, 2004. Südhof TC. Neurotransmitter release: The last millisecond in the life of a synaptic vesicle. Neuron 80: 675–690, 2013. Südhof TC, Czernik AJ, Kao HT, Takei K, Johnston PA, Horiuchi A, Kanazir SD, Wagner MA, Perin MS, De Camilli P, Greengard P. Synapsins: Mosaics of shared and individual domains in a family of synaptic vesicle phosphoproteins. Science (80- ) 245: 1474–1480, 1989. Sun T, Qiao H, Pan PY, Chen Y, Sheng ZH. Motile axonal mitochondria contribute to the variability of presynaptic strength. Cell Rep 413–149, 2013. Szabadkai G, Duchen MR. Mitochondria: The hub of cellular Ca2+ signaling. Physiology 23: 84–94, 2008. Szydlowska K, Tymianski M. Calcium, ischemia and excitotoxicity. Cell Calcium 47: 122–129, 2010. Tang YG, Zucker RS. Mitochondrial involvement in post-tetanic potentiation of synaptic transmission. Neuron 18: 483–491, 1997. Tanida I, Ueno T, Uchiyama Y. A Super-Ecliptic, pHluorin-mKate2, Tandem Fluorescent Protein-Tagged Human LC3 for the Monitoring of Mammalian Autophagy. PLoS One 9: e110600, 2014. Tevet Y, Gitler D. Using FRAP or FRAPA to Visualize the Movement of Fluorescently Labeled

128

Proteins or Cellular Organelles in Live Cultured Neurons Transformed with Adeno-Associated Viruses BT - High-Resolution Imaging of Cellular Proteins: Methods and Protocols. In: , edited by Schwartzbach SD, Skalli O, Schikorski T. Springer New York, p. 125–151. Thiel G, Sudhof CT, Greengard P. Synapsin II. Mapping ofa domain in the NH2-terminal region which binds to small synaptic vesicles. J Biol Chem 265: 16527–16533, 1990. Todorova V, Blokland A. Mitochondria and Synaptic Plasticity in the Mature and Aging Nervous System. Curr Neuropharmacol 15: 166–173, 2017. Tsuyama T, Kishikawa JI, Han YW, Harada Y, Tsubouchi A, Noji H, Kakizuka A, Yokoyama K, Uemura T, Imamura H. In Vivo Fluorescent Adenosine 5′-Triphosphate (ATP) Imaging of Drosophila melanogaster and Caenorhabditis elegans by Using a Genetically Encoded Fluorescent ATP Biosensor Optimized for Low Temperatures. Anal Chem 85: 7889–7896, 2013. Valero RA, Senovilla L, Núñez L, Villalobos C. The role of mitochondrial potential in control of calcium signals involved in cell proliferation. Cell Calcium 44: 259–269, 2008. Vasin A, Zueva L, Torrez C, Volfson D, Troy Littleton J, Bykhovskaia M. Synapsin regulates activity-dependent outgrowth of synaptic boutons at the Drosophila neuromuscular junction. J Neurosci 34: 10554–10563, 2014. Verstegen AMJ, Tagliatti E, Lignani G, Marte A, Stolero T, Atias M, Corradi A, Valtorta F, Gitler D, Onofri F, Fassio A, Benfenati F. Phosphorylation of synapsin I by cyclin-dependent kinase- 5 sets the ratio between the resting and recycling pools of synaptic vesicles at hippocampal synapses. J Neurosci 34: 7266–7280, 2014. Vigil FA, Giese KP. Calcium/calmodulin-dependent kinase II and memory destabilization: a new role in memory maintenance. J Neurochem 147: 12–23, 2018. Vos M, Lauwers E, Verstreken P. Synaptic mitochondria in synaptic transmission and organization of vesicle pools in health and disease. Front Synaptic Neurosci 2: 1–10, 2010. Walter AM, Pinheiro PS, Verhage M, Sørensen JB. A Sequential Vesicle Pool Model with a Single Release Sensor and a Ca2+-Dependent Priming Catalyst Effectively Explains Ca2+- Dependent Properties of Neurosecretion. PLoS Comput Biol 9, 2013. Waters J, Smith SJ. Mitochondria and release at hippocampal synapses. Pflugers Arch Eur J Physiol 447: 363–370, 2003. Weisskopf MG, Castillo PE, Zalutsky RA, Nicoll RA. Mediation of hippocampal mossy fiber long-term potentiation by cyclic AMP. Science (80- ) 265: 1878 LP – 1882, 1994. Wu XS, Wu LG. The yin and yang of calcium effects on synaptic vesicle endocytosis. J Neurosci 34: 2652–2659, 2014. Wyss M, Kaddurah-Daouk R. Creatine and creatinine metabolism. Physiol Rev 80: 1107–1213,

129

2000. Yao C-K, Lin YQ, Ly C V., Ohyama T, Haueter CM, Moiseenkova-Bell VY, Wensel TG, Bellen HJ. A Synaptic Vesicle-Associated Ca2+ Channel Promotes Endocytosis and Couples Exocytosis to Endocytosis. Cell 138: 947–960, 2009. Yellen G. Fueling thought: Management of glycolysis and oxidative phosphorylation in neuronal metabolism. J Cell Biol 217: 2235–2246, 2018. Ying SW, Futter M, Rosenblum K, Webber MJ, Hunt SP, Bliss TVP, Bramham CR. Brain-derived neurotrophic factor induces long-term potentiation in intact adult hippocampus: Requirement for ERK activation coupled to CREB and upregulation of Arc synthesis. J Neurosci 22: 1532– 1540, 2002. Zhang B-B, Jin H, Bing Y-H, Zhang X-Y, Chu C-P, Li Y-Z, Qiu D-L. A Nitric Oxide-Dependent Presynaptic LTP at Glutamatergic Synapses of the PVN Magnocellular Neurosecretory Cells in vitro in Rats [Online]. Front. Cell. Neurosci. 13: 283, 2019.https://www.frontiersin.org/article/10.3389/fncel.2019.00283. Zheng CY, Petralia RS, Wang YX, Kachar B. Fluorescence recovery after photobleaching (FRAP) of fluorescence tagged proteins in dendritic spines of cultured hippocampal neurons. J Vis Exp 9: 9–11, 2011. Zhou Q, Petersen CCH, Nicoll RA. Effects of reduced vesicular filling on synaptic transmission in rat hippocampal neurones. J Physiol 525: 195–206, 2000. Zucker RS. Calcium- and activity-dependent synaptic plasticity. Curr Opin Neurobiol 9: 305– 313, 1999. Zucker RS, Regehr WG. Short-Term Synaptic Plasticity. Annu Rev Physiol 64: 355–405, 2002.

130

תקציר

תקשורת סינפטית הינה תהליך הצורך אנרגיה רבה. ההנחה המקובלת היא שהעברה סינפטית דורשת אנרגיה, ולפיכך אמורה להשפיע על האיזון האנרגטי בתא. רק לאחרונה הראו כי פעילות סינפטית מוגברת גורמת לירידה בריכוז ה- ATP בטרמינל הפרהסינפטי בנוירונים במערכת העצבים המרכזית. מיטוכונדריות הן ספקיות האנרגיה המרכזיות בנוירונים ע"י שימוש בתהליך הזרחון החמצוני, ולפיכך אין זה מפתיע שלפחות חצי מהטרמינאלים הפרהסינפטיים מכילים מיטוכונדריות. בעבר הניחו כי יש לכך השפעה על הדינמיות האנרגטית בסינפסות, דבר שאמור להתבטא בשינוי ההולכה הסינפטית והפלסטיות קצרת הטווח בסינפסות אינדיבידואליות. סינפסינים הינם משפחה של פוספו-חלבונים סינפטיים אשר עוברים אינטראקציה עם פני השטח של הוסיקולות הסינפטיות. לפני יותר מעשור אופיין בהם אתר קישור ל ATP -שפעילותו לא הוגדרה היטב. ההיפותזה שלנו היא שבנוסף להיותו מקור אנרגיה, ATP מבקר ישירות את פעילות הסינפסינים ואת ההולכה הנוירונאלית התלויה בהם . על מנת לבחון את ההיפותזה שלנו, יצרנו מוטציה (K270Q) בסינפסין IIa המונעת ממנו לקשור ATP. ביטאנו את הסינפסין המוטנטי בנוירונים היפוקמפאליים שהופקו מעכבר שחסרים לו הגנים המקודדים לכל האיזופורמים במשפחת הסינפסינים. התוצאות שהתקבלו מראות כי הקשירה של ATP לסינפסין IIa משחקת תפקיד מרכזי בוויסות הפעילות שלו, ושהיא בעלת חשיבות לפלסטיות הסינפטית קצרת-הטווח.

בנוסף חקרנו את היחסים בין מיטוכונדריה, רמות ATP ואגירת וסיקולות בטרמינלים, והאם הקשר ביניהם הוא תלוי סינפסין. בכך העמקנו את ההבנה שלנו בנושא החשוב אך המורכב של גלגולי אנרגיה בסינפסה. מצאנו כי בטרמינלים פרהסינפטיים בהם מצויות מיטוכונדריות ישנו מאגר גדול יותר של וסיקולות. תופעה זו מתקיימת רק כאשר חלבוני הסינפסין מצויים בתאים אלו. בנוסף, מצאנו כי רמות ה- ATP בסינפסות המכילות מיטוכונדריה בתוכן גבוהות יותר בתלות בנוכחות סינפסינים בתא. תוצאות אלו מרמזות כי סינפסינים יכולים לווסת את רמות ה- ATP בתא, על-ידי קשירה או ניצול של מולקולות ATP, וגם כי ישנה בקרה שונה על איזופורמים שונים של סינפסין בנושא זה.

לסידן ישנו תפקיד מרכזי בתהליך ההעברה הסינפטית. גורמים אשר משפיעים על משק הסידן בציטופלסמת התא ישפיעו על תהליכים כמו ההסתברות לשחרור, מחזור הוסיקולה וצריכת אנרגיה. למיטוכונדריות יש תפקיד מרכזי בתהליכים אלו כי הן יכולות לאגור כמויות גדולות של סידן וגם לשחרר אותו לציטופלסמת התא. מאזן הסידן מווסת על-ידי איזון בין מנגנוני כניסת סידן והוצאת סידן. הוצאת הסידן מתווכת ע"י משחלף נתרן\סידן\ליתיום (NCLX)

131

מיטוכונדריאלי. שמנו לנו למטרה לחקור כיצד משחלף זה משפיע על תכונות סינפטיות, שיווי- משקל של סידן במיטוכונדריות ובציטופלסמת התא, ופעילות סינפטית. בעזרת שימוש בעכברים בהם הגן המקודד ל-NCLX חסר, יכולנו לחקור היפותזה זו. לפי התוצאות שהתקבלו NCLX הוא שחקן חשוב בשמירת מאזן סידן תקין במיטוכונדריות, ובכך הוא גם משפיע על מאזן הסידן בציטופלסמת התא. בכך, המשחלף מבקר פעילות סינפטית ותהליכים של פלסטיות סינפטית קצרת וארוכת-טווח. בנוסף, אנו משערים כי ה-NCLX משפיע על תכונות פרהסינפטיות המתווכות על-ידי סינפסין I, היות וסינפסינים מבוקרים על-ידי סידן בנוסף לבקרה עליהם על-ידי רמות ATP בתא והזירחון שלהם.

התוצאות המובאות בתזה זו פורסמו בחלקן )Shulman, Stavsky et al., 2015(, וחלקים אחרים נמצאים בשלבים מתקדמים של הכנה לפרסום.

132

גלגולי אנרגיה בסינפסה: חקר ניצול ATP בטרמינל הפרסינפטי והפוסטסינפטי, ותפקידו של קישור ATP לחלבון הפרסינפטי סינפסין

מחקר לשם מילוי חלקי של הדרישות לקבלת תואר "דוקטור לפילוסופיה"

מאת

סטבסקי אלכסנדרה

הוגש לסינאט אוניברסיטת בן גוריון בנגב

05.10.2019 ז' בחשוון תש"פ

באר שבע

מחקר לשם מילוי חלקי של הדרישות לקבלת תואר "דוקטור לפילוסופיה"

מאת

סטבסקי אלכסנדרה

הוגש לסינאט אוניברסיטת בן גוריון בנגב

אישור המנחה ______

אישור דיקן בית הספר ללימודי מחקר מתקדמים ע"ש קרייטמן ______

05.10.2019 ז' בחשוון תש"פ

באר שבע

העבודה נעשתה בהדרכת דר' דניאל גיטלר ופרופ' יעל אמיתי

במחלקה לפיסיולוגיה וביולוגיה של התא

בפקולטה למדעי הבריאות

הצהרת תלמיד המחקר עם הגשת עבודת הדוקטור לשיפוט

אני החתום מטה מצהיר/ה בזאת: )אנא סמן(:

X חיברתי את חיבורי בעצמי, להוציא עזרת ההדרכה שקיבלתי מאת מנחה/ים.

X החומר המדעי הנכלל בעבודה זו הינו פרי מחקרי מתקופת היותי תלמיד/ת מחקר.

X בעבודה נכלל חומר מחקרי שהוא פרי שיתוף עם אחרים, למעט עזרה טכנית הנהוגה בעבודה ניסיונית. לפי כך מצורפת בזאת הצהרה על תרומתי ותרומת שותפי למחקר, שאושרה על ידם ומוגשת בהסכמתם.

תאריך 05.11.2019 שם התלמיד/ה אלכסנדרה סטבסקי חתימה