VU Research Portal

VU Research Portal

The Role of Adhesion Molecule SALM1 in Synapse Development Brouwer, M.

2020

document version

Publisher's PDF, also known as Version of record

Link to publication in VU Research Portal

citation for published version (APA)

Brouwer, M. (2020). The Role of Adhesion Molecule SALM1 in Synapse Development.

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal ?

Take down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

E-mail address:

vuresearchportal.ub@vu.nl

The Role of Adhesion Molecule SALM1

in Synapse Development

Author: Marinka Brouwer

Printed by: Ipskamp Printing, Enschede ISBN: 978-94-028-1987-8

© 2020 Marinka Brouwer. All rights reserved.

Cover, layout & illustrations: Marinka Brouwer

About the cover: Confocal micrograph of a cultured mouse hippocampal excitatory

neuron expressing the SALM1-pHl construct. Surface SALM1-pHl is shown in green, presynaptic neurotransmitter vesicle marker VGluT1 is shown in red and cytoskeleton component F-actin (stained using Phalloidin) is shown in blue.

VRIJE UNIVERSITEIT

T

R

A

M

SALM1

S

D

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad Doctor of Philosophy aan de Vrije Universiteit Amsterdam,

op gezag van de rector magnificus prof.dr. V. Subramaniam, in het openbaar te verdedigen ten overstaan van de promotiecommissie

van Faculteit der Geneeskunde op vrijdag 1 mei 2020 om 11.45 uur

in de aula van de universiteit, De Boelelaan 1105

door Marinka Brouwer geboren te Rheden

promotor: prof.dr. M. Verhage

Contents

Chapter 1

General Introduction

1.1 Synapse Development

1.2 Regulators of Synapse Development 1.3 Aim of the Thesis

Chapter 2

Adhesion molecule SALM1 binds the presynaptic

CASK/Mint1/Lin7b complex and localizes to

hippocampal pre- and postsynapses.

2.1 Abstract 2.2 Introduction 2.3 Results 2.4 Discussion

2.5 Materials and Methods 2.6 Supplemental Figures

Chapter 3

SALM1 regulates Neurexin1β cis-clusters at the

cell surface via PIP2/F-actin microdomains.

3.1 Abstract 3.2 Introduction 3.3 Results 3.4 Discussion

3.5 Materials and Methods 3.6 Supplemental Figures

Chapter 4

SALM1 promotes synaptogenesis by regulating

Neurexin1β surface levels in hippocampal neurons.

4.1 Abstract 4.2 Introduction 4.3 Results 4.4 Discussion

4.5 Materials and Methods 4.6 Supplemental Figures

9

25

49

79

Chapter 5

SALM1 depletion impairs early stage synapse

formation, synaptic vesicle clustering and

synaptic transmission.

5.1 Abstract 5.2 Introduction 5.3 Results 5.4 Discussion

5.5 Materials and Methods 5.6 Supplemental Figures

Chapter 6

General Discussion

6.1 SALMs differentially localize to synaptic terminals of hippocampal neurons.

6.2 SALM1 mediates synapse development and synaptic function.

6.3 SALM1 controls synapse development via F-actin/PIP2- dependent Neurexin cis-oligomerization.

6.4 A general mechanism for synapse development? 6.5 Future directions

Bibliography

Summary

Acknowledgements

99

121

134

146

148

Chapter 1

10

The complexity of the human brain is a unique result of evolution that hallmarks the human species and allows higher order cognitive functions such as learning and memory, perception and information processing, decision-making and self-consciousness. The human brain is estimated to contain ~160 billion brain cells which are roughly divided into two major classes: glial cells and neurons (Azevedo et al., 2009; Andrade-Moraes et al., 2013; von Bartheld et al., 2016). Glial cells have various essential functions in the brain such as facilitating communication between neurons, providing support to neurons and providing immune defense. Neurons are polarized cells consisting of a cell body (the soma) from which protrusions originate that are distinguished as long thin axons or shorter and thicker dendrites. Neurons transfer electrical signals via their axons onto the dendrites of other neurons. This neuronal communication occurs via specialized contact sites termed the synapses and forms the basis of cognitive brain functioning (Fig. 1).

Synapses typically exist of a presynapse (which is present on axons) and a postsynapse (present on dendrites) which are separated by a gap called the synaptic cleft (Fig. 1C-D). Presynapses are hallmarked by the presence of small vesicles containing neurotransmitter molecules. Different neurotransmitters exist and the type of neurotransmitter secreted by a neuron is generally used to discriminate subpopulations of neurons (e.g. glutamate is the main neurotransmitter secreted by excitatory neurons while GABA is the main neurotransmitter secreted by inhibitory neurons). When an electrical signal travels through an axon and reaches a presynapse, the signal triggers release of neurotransmitters into the synaptic cleft. The neurotransmitters then bind to specialized receptor molecules (the neurotransmitter receptors) that are present on the postsynapses. The binding of neurotransmitters to neurotransmitter receptors on the postsynapse induces an electrical signal in the postsynapse that can travel into the dendrite and further to the soma. Hence, synapses are essential for communication between neurons.

The synapses found in the brain are highly diverse in synaptic protein content resulting in a high diversity in physiological properties amongst synapses which underlies the complex connectivity of the brain (Fuccillo et al., 2015; Foldy et al., 2016; Zhu et al., 2018). Synapses are extensively formed during early development (around birth and during the first two decades in humans) and, to a lesser extent, throughout life. Synapse formation between neurons involves tightlyregulated targeting of subsets

11

Figure 1. Neurons form specialized neural circuits and communicate via specialized contact sites called synapses. A) Schematic side view of the rodent brain. Blue box highlights the

hippocampus. B) Schematic representation of the rodent hippocampal microcircuit implicated in learning and memory. Signals enter hippocampal neurons (orange) in the Dentate Gyrus (DG) and subsequently travel via the axons of DG neurons to neurons (purple) in the CA3 subfield via synaptic contact (highlighted by the green box). Signals further propagate from the CA3 neurons to CA1 neurons (red) and subsequently travels to different brain areas. Black arrows indicate direction of signal propagation. C) Schematic representation of a synapse. Presynapses are present on axons and are hallmarked by the presence synaptic vesicles (SV) filled with neurotransmitter (black dots). Signals travelling through an axon can trigger fusion of SVs with the synaptic membrane to release neurotransmitters into the synaptic cleft. Postsynapses are present on dendrites and are hallmarked by the presence of neurotransmitter receptors (NTR). Binding of neurotransmitters to NTRs induces a signal that propagates further into to the dendrites. D) Electron micrograph of a synapse from cultured mouse hippocampal neurons. Presynapse is indicated in orange, postsynapse is indicated in purple. Red arrow indicates the synaptic cleft, blue arrow indicates a synaptic vesicle and green arrow indicates the postsynaptic density which is packed with postsynaptic proteins such as NTRs. Scale bar (bottom left corner) = 100nm

12

of neurons to establish neural circuits allowing specific cognitive functions (e.g. the hippocampal circuit for learning and memory; Fig. 1A-B). Though research in the past decades has provided insight on the diversity of molecules involved in synapse development, the precise molecular mechanisms underlying synapse development and synaptic specificity are still largely unknown. Various molecules involved in synapse development have been linked to neurological disorders including Autism Spectrum Disorders (ASD), intellectual disability and Schizophrenia. Understanding the molecular mechanisms underlying synapse development is therefore also crucial to help understand the malfunctions in different neurological disorders and may provide a basis for the development of treatments.

1.1 Synapse development

To establish specified neural circuits, axons of developing neurons grow towards their target cells through a process called “axon guidance” or “axon pathfinding”. During this process growing axons detect specialized guidance proteins in the environment which form repelling or attracting cues that guide the axon to the target cell (O'Donnell et al., 2009). Once the axon reaches the dendrites of a target cell, synapses are formed through complex stepwise mechanisms which can be roughly divided in three stages: synapse initiation, synapse assembly and synapse maturation (Fig. 2).

1.1.1 Synapse initiation

Upon reaching the area of the target cell, axons need to discriminate the target cell from other neighboring cells to initiate synapse development. How synapse formation is initiated is still poorly understood. Target recognition and subsequent synapse initiation is thought to arise from direct interaction of specific subsets of cell adhesion molecules (CAMs) that are present on the membranes of axons and dendrites (Fig. 2A) (Akins and Biederer, 2006; Sudhof, 2017; Kurshan and Shen, 2019). These concepts arose from experimental evidence showing that presentation of specific CAMs to an axon or dendrite is sufficient to induce pre- or postsynapse development respectively (Biederer and Scheiffele, 2007; Czondor et al., 2013). In addition, by expressing specific subsets of the myriad of CAMs neurons are thought to form molecular codes that allows neurons to discriminate each other and promote correct target recognition (Fuccillo et al., 2015; Foldy et al., 2016; Schroeder et al., 2018).

13

However, little is known about the organization of synaptic CAMs and other synaptic proteins before axodendritic contact is established. Several studies showed that homomeric clustering of CAMs promoted trans-cellular contact and synapse formation in artificial assays (Dean et al., 2003; Fogel et al., 2011; Shipman and Nicoll, 2012) suggesting that CAM pre-clustering may contribute to synapse formation. However,

Figure 2. Synapses develop through stepwise mechanisms roughly divided in three stages: synapse initiation, assembly and maturation. A) Synapse initiation is the first step in synapse

development and is hallmarked by direct contact between axons and dendrites via CAMs. B) Initiation is followed by synapse assembly and involves the recruitment of various synaptic components (e.g. NMDARs and SVs) to the developing synapse. C) The last step in synapse development is the maturation of the synapse which involves further fine-tuning of the synaptic material and recruit AMPARs to establish a functional synapse. D) Synaptic transmembrane proteins (TMP) are thought to partition into lipid rafts (green) and/or PIP2 (red) microdomains.

14

whether multiple different CAMs form heteromeric pre-clustered complexes prior to axodendritic contact is unknown.

1.1.2 Synapse assembly

After initial contact between axons and dendrites, various essential synaptic components are recruited to the developing synapse including scaffolding proteins, the presynaptic release machinery, presynaptic vesicles and postsynaptic neurotransmitter receptors (Fig. 2B). How initial contact between axonal and dendritic CAMs trigger the assembly of other proteins is poorly understood. Many CAMs bind intracellular scaffolding proteins including postsynaptic PSD95 and SAP102 or presynaptic CASK and Liprin‐α which recruit and stabilize other synaptic proteins and are proposed to link cell adhesion to synapse assembly (Dalva et al., 2007). The recruitment of these synaptic components to the developing synapse is temporally regulated. Presynaptic components such as scaffolding protein Bassoon and presynaptic vesicles are recruited faster than postsynaptic components including PSD95 and the postsynaptic glutamate neurotransmitter receptors AMPARs and NMDARs (Friedman et al., 2000; Okabe et al., 2001; Bresler et al., 2004). However, the mechanisms that control the sequential recruitment of different synaptic proteins is poorly understood.

1.1.3 Synapse maturation

Synapse maturation involves the fine-tuning of the synaptic material to a fully functional synapse as part of a neural circuit (Fig. 2C). Whereas synapse initiation and synapse assembly are activity independent, synapse maturation is thought to be largely activity-dependent (Verhage et al., 2000; Sando et al., 2017; Sigler et al., 2017; Sudhof, 2018). Several key events have been described that mark synapse maturation including a developmental switch in synaptic protein expression, synapse unsilencing and synaptic pruning. For several pre- and postsynaptic proteins, a developmental switch has been described in the expression levels of different variants (isoforms) of these proteins. For example, the NR2b subunit of the postsynaptic NMDA neurotransmitter receptor is the predominant NR2 isoform expressed in immature synapses. However, during maturation expression levels of the NR2a isoform drastically increase resulting in a switch where NR2a becomes the predominant NR2 isoform incorporated in NMDA receptors (Williams et al., 1993; Monyer et al., 1994). In addition to the NMDA receptor, developmental switching in expression levels of protein isoforms has also been

15

observed for other synaptic proteins including the postsynaptic NKCC1/KCC2 chloride transporters (involved in switching GABA responses from excitatory to inhibitory responses) and presynaptic release machinery proteins including SNAP25a/SNAP25b and Syntaptotagmin1/Synaptotagmin2 (Bark et al., 1995; Rivera et al., 1999; Kochubey et al., 2016). Impairment of the developmental switching and/or expression of these proteins results in aberrant synapse development, synapse function and/or lethality (Hubner et al., 2001; Bark et al., 2004; Gambrill and Barria, 2011; Kochubey et al., 2016). Together, these findings suggest that developmental switching in the expression of synaptic proteins is tightly temporally regulated and is important for synapse maturation.

Another key event in glutamatergic synapse maturation involves the upregulation of AMPA receptor expression and recruitment to the postsynapse. During early development, the brain contains many ‘silent’ synapses (Liao et al., 1995; Durand et al., 1996; Isaac et al., 1997; Hanse et al., 2013). These immature synapses are marked by an absence of functional postsynaptic AMPA receptors. Neurotransmitters released by the presynapse that normally activate AMPA receptors can therefore not evoke a postsynaptic response in these synapses. Recruitment of AMPA receptors to silent synapses results in synapse unsilencing rendering a fully functional synapse (Liao et al., 1995; Durand et al., 1996; Isaac et al., 1997; Hanse et al., 2013).

Finally, brain maturation is marked by extensive elimination of synapses through a process called “synaptic pruning”. In humans, it is estimated that approximately half of all synapses in the brain are pruned during the first two decades of life, when the brain matures (Petanjek et al., 2011; Sudhof, 2018). Glial cells have been shown to play a major role in synaptic pruning through phagocytosis, a process where glial cells “eat” the excessive synapses (Paolicelli et al., 2011; Schafer et al., 2012). Defects in synaptic pruning during development leads to aberrant brain connectivity and abnormal social behavior (Kim et al., 2017; Filipello et al., 2018; Wilton et al., 2019).

1.2 Regulators of synapse development

Over the past decades, different cell intrinsic and extrinsic mechanisms have been described that promote and control the various steps in synapse development. Cell extrinsic factors that control synapse development include the extracellular matrix (a network of molecules that provides support to cells) and glial cells (Barros et al., 2011;

16

Chung et al., 2015). Cell intrinsic components that regulate synapse development include the cytoskeleton, membrane lipids, scaffolding proteins and cell adhesion molecules. For the purpose of this thesis, we will further focus on these cell intrinsic components.

1.2.1 Cytoskeleton

The complex shape of neurons (with its multitude of axonal and dendritic branches and hundreds of synapses) is established through the cytoskeleton. The neuronal cytoskeleton exists of a tightly regulated network of different types of filaments including microtubules, neurofilaments, actin filaments, septin filaments, and spectrin filaments. Of these subtypes, filamentous actin (F-actin) is considered the major cytoskeletal constituent of synapses (Fig. 2). F-actin is formed by polymerization of single monomeric actin proteins which is initiated by F-actin nucleation complexes such as the WAVE/WASP and ARP2/3 protein complexes (Machesky et al., 1999; Rohatgi et al., 1999). F-actin and its nucleation complexes are enriched in developing synapses (Zhang and Benson, 2001; Soderling et al., 2007; Wegner et al., 2008). Blocking F-actin polymerization using the drug Latrunculin A drastically reduces the number of synapses formed during early development (Zhang and Benson, 2001; Chia et al., 2012). Similarly, depleting or inhibiting specific proteins of actin nucleation complexes reduces synapse formation during early development (Soderling et al., 2007; Wegner et al., 2008; Chia et al., 2014). These findings suggest an important role for F-actin in the early stages of synapse development.

1.2.2 Membrane lipids

The cell membrane exists of a bilayer of different types of lipids that separates the extracellular and intracellular environments (Fig. 2D). Proteins such as cell adhesion molecules and the AMPA and NMDA glutamate receptors are integrated in the cell membrane allowing signal transduction between the extra- and intracellular environments. Membrane proteins are often organized in microdomains enriched for specific membrane proteins (Fig. 2D), cytoskeletal components, and lipids. Two types of microdomains have been described to regulate synapse development and function: lipid rafts and phospholipid microdomains. Several types of synaptic membrane proteins partition in lipid rafts which are enriched in cholesterol and sphingolipids (Thiele et al., 2000; Hering et al., 2003). Inhibition or stimulation of cholesterol synthesis during early development reduces or increases synapse numbers

17

respectively (Mauch et al., 2001; Hering et al., 2003; Suzuki et al., 2007). Phospholipid microdomains are enriched in phospholipids such as phosphatidylinositol-4,5-biphosphate (PIP2) which forms electrostatic interactions with positively charged polybasic domains of synaptic membrane proteins (van den Bogaart et al., 2011). PIP2 controls synaptic function by regulating vesicle exo- and endocytosis (Di Paolo et al., 2004; Honigmann et al., 2013; Milovanovic et al., 2016), but its role in synapse development is poorly understood. PIP2 microdomains furthermore reciprocally regulate F-actin polymerization by attracting negatively charged F-actin nucleation complexes (Papayannopoulos et al., 2005; Saarikangas et al., 2010; Chierico et al., 2014). Together, these findings indicate a role for membrane lipids in synapse development.

1.2.3 Scaffolding proteins

A variety of scaffolding proteins make up the presynaptic active zone (e.g. Liprinα, CASK and Bassoon) and the postsynaptic density (e.g. PSD95, SHANKs and SAP102). Scaffolding proteins are considered important regulators of synapse assembly as they simultaneously interact with different synaptic proteins including cell adhesion molecules, components of the release machinery, neurotransmitter receptors, F-actin binding proteins and/or other scaffolding proteins (Butz et al., 1998; Kim and Sheng, 2004). Depletion of scaffolding proteins including PSD95, Liprin-α and SHANKs alters synaptic structure and/or reduces synaptic expression levels of different synaptic proteins (Patel et al., 2006; Peca et al., 2011; Chen et al., 2015). These findings suggest an important role for scaffolding proteins in synapse assembly.

1.2.4 Cell adhesion molecules

Over the past decades, cell adhesion molecules (CAMs) have received much attention as essential regulators of synapse development and function (see also sections 1.1.1 and 1.1.2). A large collection of synaptic CAMs has been identified with various functions in the different steps of synapse development and synapse function. CAMs form homo‐ and/or heteromeric trans‐cellular interactions between their extracellular domains that are often sufficient to induce synapse formation (Scheiffele et al., 2000; Biederer et al., 2002; Graf et al., 2004; Yim et al., 2013). However, depleting individual CAMs often results in mild phenotypes with minor to no reduction in synaptic densities (Robbins et al., 2010; Anderson et al., 2015; Chanda et al., 2017). This may result from

18

the expression of multiple different CAMs in individual synapses (Schroeder et al., 2018) which may be redundant in their synaptogenic properties.

In addition to trans-interactions, CAMs form homo‐ and heteromeric cis‐ interactions via their extracellular domains that result in CAM oligomerization. Cis‐ and trans‐interactions can be competitive or cooperative resulting in a complex interplay between different CAMs and their downstream effectors in individual synapses (Aricescu and Jones, 2007; Taniguchi et al., 2007; Lie et al., 2016). In addition, by selectively expressing subsets of the myriad of CAMs, neurons may use the complex interplay between CAMs as a molecular code to establish synapse specificity and brain connectivity (Fuccillo et al., 2015; Foldy et al., 2016; Schroeder et al., 2018). Due to the large variety of synaptic CAMs, introducing all identified CAMs is beyond the scope of this thesis. This introduction will further focus on three major CAM subclasses: Neurexins and Neuroligins, the leucine rich repeat domain superfamily and the immunoglobulin domain superfamily (Fig. 3). Other classes of synaptic CAMs (e.g. Cadherins, Integrins and EphB), have been reviewed in excellent recent reviews (Missler et al., 2012; de Wit and Ghosh, 2016; Park and Goda, 2016).

Neurexins and Neuroligins

Presynaptic Neurexins and their postsynaptic binding partners Neuroligins are possibly the most extensively studied trans-interacting CAM pair in neuronal synapses. The mammalian genome contains three Neurexin genes (Nrxn1, 2 and 3) which each encode two classes of transcripts (the full α variant and a shorter β variant) due to the activity of two promoters in each gene (Ushkaryov et al., 1992; Ushkaryov and Sudhof, 1993; Ushkaryov et al., 1994). Nrxn1 furthermore contains a third promoter encoding a short γ variant (Sterky et al., 2017). Additional alternative splicing at six different splice sites (SS1 to SS6) in each of the Neurexin genes results in over 1000 predicted isoforms (Ullrich et al., 1995). The mammalian genome furthermore contains four Neuroligin genes (Nlgn1, 2, 3 and 4) which can be alternatively spliced at least two sites (splice site A and B) (Ichtchenko et al., 1995; Ichtchenko et al., 1996; Bolliger et al., 2001; Bolliger et al., 2008).

The Neurexin and Neuroligin isoforms are differentially expressed in different brain areas and different neuronal subtypes (Fuccillo et al., 2015; Foldy et al., 2016). For example, Nlgn1 localizes to excitatory synapses, whereas Nlgn2 localizes to inhibitory synapses (Song et al., 1999; Graf et al., 2004; Varoqueaux et al., 2004).

19

Figure 3. A variety of CAMs are present on pre- and postsynaptic membranes and form complex trans-interaction networks. Schematic representation of Neurexins, Neuroligins and a

selection of several well characterized LRR- and Ig-CAMs. Double-headed arrows indicate established trans-interactions between CAMs. Flrts trans-interact with postsynaptic G-protein coupled receptors latrophilins (LPHNs).

20

The trans-synaptic binding of Neuroligins to Neurexins is also isoform dependent. For example, Nlgn1 containing an insert at splice site B specifically interacts with Nrxn1β lacking the SS4 insert, but does not interact with α-Neurexins (Boucard et al., 2005). Nlgn1 lacking the splice site B insert binds both α- and β-Neurexins lacking or containing the SS4 insert (Boucard et al., 2005).

Expression of Neurexins and Neuroligins in heterologous cells is sufficient to induce synapse formation in co-cultured neurons through trans-synaptic signaling suggesting a function for these CAMs in synapse development (Scheiffele et al., 2000; Graf et al., 2004). This is further supported by the ability of Neurexins and Neuroligins to bind scaffolding proteins CASK and PSD95 respectively (Hata et al., 1996; Irie et al., 1997). However, depletion of specific Neurexin or Neuroligin isoforms results in a variety of synaptic phenotypes that affect synaptic function, but minimally affects synaptic density (Anderson et al., 2015; Chanda et al., 2017; Sudhof, 2017). Triple knockout (KO) of all α-Neurexin isoform is lethal likely due to severely impaired excitatory and inhibitory synaptic transmission as a result of defects in calcium signaling (Missler et al., 2003). In contrast, triple KO of all β-Neurexins does not impair viability and impairs only excitatory synaptic transmission due to a defect in postsynaptic endocannabinoid signaling (Anderson et al., 2015). Several lines of evidence furthermore suggest that individual Neurexin isoforms have distinct functions in different brain areas (Aoto et al., 2015; Chen et al., 2017). Together, these findings indicate a complex regulatory role for Neurexins and Neuroligins in synapse function, whereas the biological relevance of the synaptogenic properties of Neurexins and Neuroligins observed in co-culture assays is still poorly understood.

Leucine rich repeat domain superfamily

CAMs belonging to the Leucine Rich Repeat domain superfamily (LRR-CAMs) are hallmarked by the presence of a variable number (tandem repeats) of 20-30 amino acid stretches enriched for leucine (Kobe and Deisenhofer, 1994). These LRR domains are located in the extracellular domain of CAMs and function as protein-protein interaction domains (Kobe and Kajava, 2001). LRR-CAMs are differentially expressed in different subpopulations of neurons (Foldy et al., 2016; Shekhar et al., 2016; Paul et al., 2017). To date, Synaptic Adhesion Like Molecule 4 (SALM4) and Fibronectin Leucine Rich Repeat protein 3 (FLRT3) are the only LRR-CAMs known to be expressed in the presynapse (Seabold et al., 2008; Sando et al., 2019). In contrast,

21

multiple different postsynaptic LRR-CAMs have been identified (e.g. LRRTMs, Slitrks, NGLs and SALMs). Like for Neurexins and Neuroligins, the presence of individual LRR-CAMs (with the exception of FLRTs and several SALM isoforms) is sufficient to induce synaptogenesis through trans-synaptic signaling (Kim et al., 2006; Linhoff et al., 2009; Mah et al., 2010; Yim et al., 2013; Schroeder and de Wit, 2018). The cytoplasmic tails of LRR-CAMs furthermore often interact with scaffolding proteins, linking LRR-CAMs to synapse assembly (Kim et al., 2006; Ko et al., 2006; Linhoff et al., 2009). Depletion of specific LRR-CAM isoforms differentially alters synaptic density and/or synapse function (Schroeder and de Wit, 2018). For example, SALM3 KO reduces only excitatory synapse density and synaptic transmission, whereas SALM5 knockdown reduces both excitatory and inhibitory synapse density and synaptic transmission (Mah et al., 2010; Takahashi et al., 2012; Choi et al., 2016). Together, these findings indicate a role for LRR-CAMs in synapse development, though the precise molecular mechanisms are still poorly understood.

Immunoglobulin domain superfamily

CAMs belonging to the immunoglobulin superfamily are hallmarked by the presence of one or more immunoglobulin (Ig) like protein-interaction domains in the extracellular region. Ig domains mediate homophilic cis- and trans-interactions as well as heterophilic trans-interactions between CAMs providing a basis for CAM oligomerization (Seabold et al., 2008; Fogel et al., 2011; Takahashi and Craig, 2013; Sytnyk et al., 2017). A variety of synaptic Ig-CAMs have been identified including presynaptic LAR-PTPRs, postsynaptic NGLs and SALMs, and SynCAMs, L1 CAMs and NCAMs which are present at both pre- and postsynapses (Cameron and McAllister, 2018). Several synaptic CAMs (e.g. NGLs and SALMs) contain both Ig domains and LRR domains thereby belonging to both the Ig superfamily and the LRR superfamily. Like for LRR-CAMs, loss of specific Ig-CAM isoforms generally affects synapse density and/or function in different neuronal subpopulations (Cameron and McAllister, 2018). Presentation of individual Ig-CAMs to neurons furthermore is often sufficient to induce synapse formation (Biederer et al., 2002; Seabold et al., 2008; Woo et al., 2009). In addition, multiple Ig-CAMs regulate the dynamics of several cytoskeleton components including F-actin and spectrin (Leshchyns'ka and Sytnyk, 2016). Together, these findings suggest a role for Ig-CAMs in synapse development and function, though the underlying molecular mechanisms are still poorly understood.

22

Despite a large body of literature describing the functions of various synaptic CAMs, many questions remain on how CAMs control synapse development. Previous studies have mainly focused on investigating the functions of individual CAMs, but how multiple CAMs cooperate to control synapse formation and function is still poorly understood. In addition, functional research on CAMs has been mainly limited to the postsynapse where a variety of CAMs have been identified. The functional diversity of CAMs localized to the presynapse remains to be investigated. Though synaptic CAMs are thought to be important regulators of synapse development, little is still known about the downstream signaling pathways through which CAMs may induce synaptogenesis.

1.3 Aim of the thesis

Multiple different CAMs are thought to cis‐localize to individual synapses and cooperate to control synaptogenesis. How different CAMs are organized in individual synapses and how they cooperate with other synaptic regulators to organize pre‐ and postsynaptic specializations is still poorly understood. In addition, a variety of postsynaptic CAMs have been studied, whereas at the presynapse research has mainly been limited to Neurexins and RPTP receptors. Hence, the aim of this research was to identify new presynaptic CAMs and elucidate how these CAMs organize trans‐ synaptic signaling, link to synaptic regulatory proteins, and regulate synapse formation.

In Chapter 2, Synaptic Adhesion Like Molecule 1 (SALM1) was identified as an interactor of the proposed CASK/Mint1/Lin7b presynaptic organizer complex. Endogenous SALM1 was enriched at cultured mouse hippocampal excitatory synapses at different developmental stages. In addition, endogenous SALM1 localized to both pre- and postsynaptic compartments of synapses in adult mouse hippocampus. In Chapter 3, we developed short hairpin RNA’s to efficiently knock down SALM1. Knockdown of pre- or postsynaptic SALM1 impaired Neuroligin1AB and Neurexin1β mediated excitatory synaptogenesis, respectively. SALM1 clustered Neurexin on the surface of HEK293T cells in an F-actin/PIP2 dependent fashion. SALM1 recruited PIP2 through electrostatic interactions via the juxtamembrane polybasic domain.

In Chapter 4, we found that SALM1 regulates the surface expression of exogenous Neurexin1β in cultured mouse hippocampal excitatory neurons. Co-expression of Neurexin1β and SALM1 in cultured hippocampal neurons enhanced Neuroligin mediated excitatory synapse formation.

23

In Chapter 5, depletion of total pre- and postsynaptic SALM1 in single isolated mouse hippocampal excitatory neurons reduced synaptic density and synaptic transmission during early development, but not during late development. SALM1 depletion furthermore reduced synaptic vesicle clustering and synaptic vesicle fusion.

In Chapter 6, the main findings in this thesis are summarized and placed in context with existing literature. A working model of the molecular mechanism through which presynaptic SALM1 regulates synapse development is provided.

25

Chapter 2

Adhesion

molecule

SALM1

binds

the

presynaptic

CASK/Mint1/Lin7b complex and localizes to hippocampal

pre- and postsynapses.

Marinka Brouwer1_{, Fatima Farzana}2_{, Frank Koopmans}2,3_{, Ning Chen}2,3_{, Ka Wan}

Li3_{, Jan RT van Weering}1_{, August B Smit}3_{, Ruud F Toonen}2_{, Matthijs Verhage}1,2

1_{Department of Clinical Genetics, Center for Neurogenomics and Cognitive Research,}

Amsterdam Neuroscience, VU University Amsterdam and VU Medical Center, Amsterdam, The Netherlands

2_{Department of Functional Genomics, Center for Neurogenomics and Cognitive}

Research, Amsterdam Neuroscience, VU University Amsterdam and VU Medical Center, Amsterdam, The Netherlands

3_{Department of Molecular and Cellular Neurobiology, Center for Neurogenomics and}

Cognitive Research, Amsterdam Neuroscience, VU University Amsterda m and VU Medical Center, Amsterdam, The Netherlands

Results of this chapter have been published in the EMBO Journal. 2019; 38(17): e101289.

DOI: 10.15252/embj.2018101289 PMID: 31368584

26

2.1

Abstract

Synaptic cell adhesion molecules (CAMs) are essential regulators of synapse development and function. A variety of CAMs have been characterized at the postsynapse, but at the presynapse, research has mainly been limited to Neurexins and Protein tyrosine phosphatase receptors (PTPRs). In this study, we aimed to identify new presynaptic CAMs that interact with presynaptic organizer complexes involved in synapse function. We identified Synaptic Adhesion Like Molecule 1 (SALM1), previously described as a postsynaptic protein, as an interactor of the proposed CASK/Mint1/Lin7b presynaptic organizer complex in the adult mouse brain. SALM1 interacted directly with CASK via the PDZ binding domain. Endogenous SALM1 -5 family members showed differential subcellular localization in cultured hippocampal excitatory neurons. SALM1 was enriched at synapses of cultured mouse hippocampal excitatory neurons at different developmental stages. In addition, SALM1 preferentially localized to presynapses in adult mouse hippocampus. Together, these findings suggest a presynaptic functio n for SALM1 in excitatory hippocampal neurons.

27

2.2 Introduction

Establishing brain connectivity involves the precise targeting of billions of axons to their specific targets followed by the development of functional synapses. A large collection of cell adhesion molecules (CAMs) help orchestrate this complex connectivity and the assembly of synapses between specific populations of neurons (Robbins et al., 2010; Yim et al., 2013; Chen et al., 2017; Jiang et al., 2017). Multiple different CAMs are thought to cis‐localize to individual synapses and cooperate to control synaptogenesis.

Many postsynaptic CAMs have been identified to regulate synapse development and function. These CAMs form homo‐ and/or heteromeric cis‐ and trans‐interactions which can be competitive or cooperative resulting in a complex interplay between different CAMs and their downstream effectors in individual synapses (Aricescu and Jones, 2007; Taniguchi et al., 2007; Lie et al., 2016). In contrast, few presynaptic CAMs have been identified and functional studies have mainly been limited to Protein Tyrosine Phosphatase Receptors (PTPRs) and Neurexins (Nrxns) (Takahashi and Craig, 2013; Sudhof, 2017). Whether a high diversity in presynaptic CAMs and a complex cis/trans-interaction network also exists at the presynapse is still unknown.

The proposed presynaptic CAK/Mint1/Lin7b organizer complex was previously shown to link cell adhesion to synaptic vesicle release (Butz et al., 1998). In this study, we aimed to identify new presynaptic CAMs using this organizer complex as starting point in a proteomics screen. Synaptic Adhesion Like Molecule 1 (SALM1) was identified as an interactor of the CASK/Mint1/Lin7b complex through direct interaction with CASK via a PDZ binding domain. SALM1 is an LRR- and Ig-domain containing type I transmembrane protein belonging to a family of five proteins (SALM1-5) (Ko et al., 2006; Morimura et al., 2006; Wang et al., 2006). SALMs regulate synapse formation by postsynaptic mechanisms, though the function of SALM1 is unknown (Ko et al., 2006; Mah et al., 2010; Li et al., 2015; Lie et al., 2016). SALMs have previously been described as postsynaptic proteins (Lie et al., 2018), but their subcellular distribution has not yet been studied.

Here, we find differential subcellular distributions for SALM1-5 in cultured mouse hippocampal excitatory neurons using newly developed SALM1-5 specific antibodies. Endogenous SALM1 was enriched at excitatory synapses of cultured mouse hippocampal neurons at different developmental stages. SALM1 preferentially

28

localized to presynapses in adult mouse hippocampal neurons. Together, our data identify SALM1 as a presynaptic CAM and suggests a role for SALM1 in regulating presynaptic organization via the CASK/Mint1/Lin7b organizer complex.

2.3 Results

2.3.1 SALM1 binds the CASK/Mint1/Lin7b presynaptic organizer complex To identify presynaptic CAMs that regulate synapse organization, we performed an immunoprecipitation mass spectrometry (IP‐MS) proteomics screen using three components (CASK, Mint1, and Lin7b) of the proposed presynaptic organizer complex Munc18‐1/CASK/Mint1/Lin7b (Butz et al., 1998) as baits. Proteins that were detected at least 10‐fold higher as compared to the control (GluR2) IP‐MS and detected using

Figure 1. SALM1 interacts with the presynaptic CASK/Mint1/Lin7b complex. A) Heat map

showing 22 putative interactors of the Munc18‐1/CASK/Mint1/Lin7b presynaptic complex identified in a proteomics screen using CASK, Mint1, or Lin7b as bait. Detection of the putative interactors using SALM1 or GluR2 (control) as bait is also shown. Bar values indicate average log10 LFQ intensity of three replicates. Gray indicates no detection in the IP. B) Partial interactome of the putative SALM1/Munc18‐1/CASK/Mint1/Lin7b complex identified by IP‐MS analysis in (A). Blue lines indicate interactions identified in the IP‐MS screen using CASK, Mint1, or Lin7b as bait. Red lines indicate interactions identified by reverse IP‐MS with SALM1. Black lines indicate previously established interactions with the Munc18‐1/CASK/Mint1/Lin7b complex (Butz et al., 1998; Tabuchi et al., 2002). C) Co‐immunoprecipitation of V5‐tagged cytoplasmic SALM1, SALM1ΔPDZ, or an empty vector with CASK in HEK cells. The co‐IP was repeated 3 times.

29

at least two of the CASK, Mint1, and Lin7b baits were considered putative interactors of the presynaptic organizer complex (Fig. 1A and B, Fig. S1A). This resulted in 22 potential interactors including synaptic signaling proteins (e.g., Rab3GAP1 and Caskin1), cytoskeletal proteins (Actbl2 and Tubb4A), and liprin‐α isoforms. The proteomics screen identified several known interactors of the CASK/Mint1/Lin7b complex (e.g., Caskin1, liprin‐α) but did not detect other known interactors (e.g., Neurexins, Syncams, and Syndecans). The cell adhesion molecule SALM1, previously described as a postsynaptic protein (Lie et al., 2018), was the only adhesion molecule identified as a putative interactor of the CASK/Mint1/Lin7b presynaptic complex (Fig. 1A-B). In the reverse IP‐MS, using SALM1 as bait, CASK was one of the most abundantly detected molecules (Fig. 1A-B). Co‐precipitation in HEK cells further confirmed the interaction between SALM1 and CASK (Fig. 1C). Co‐precipitation was not observed after truncation of SALM1, removing its PDZ binding domain (Fig. 1C). Together, these data identify SALM1 as a novel binding partner of the CASK/Mint1/Lin7b presynaptic organizer complex, interacting directly with CASK, via its PDZ binding domain.

2.3.2 SALMs differentially distribute in mouse hippocampal neurons Interaction studies previously linked SALMs to the postsynapse (Ko et al., 2006; Morimura et al., 2006; Wang et al., 2006), but their subcellular distributions have not yet been studied. SALM1-5 are differentially expressed throughout the brain, but are all expressed in the hippocampus (Ko et al., 2006; Morimura et al., 2006). We therefore investigated the subcellular distribution of the five SALMs in mouse hippocampal neurons. Newly developed antibodies against individual SALMs specifically recognized exogenously expressed individual SALMs in HEK293T cells without cross-reacting with other SALMs (Fig. S1A-B). SALM1 antibody specificity was demonstrated after acute RNAi mediated knockdown of endogenous SALM1 in western blot and immunocytochemistry on mouse hippocampal primary neurons (Fig. S1 in Chapter 3). Multiple bands were detected in lysates of HEK cells expressing SALM1-5. Tunicamycin treatment, which blocks N-glycosylation, removed the top band of all SALMs suggesting that all five SALMs are highly glycosylated (Fig. S2).

30

Immunostaining for endogenous SALM1-5 using these new antibodies revealed the presence of all SALMs in both SMI-positive axons and MAP2-positive dendrites of mature mouse hippocampal primary neurons at 21 days in vitro (DIV21) (Fig. 2). SALM1 distribution was highly punctate, SALM2 and SALM3 distributions were partially punctate, and SALM4 and SALM5 appeared mostly diffuse with few puncta (Fig 2 and Fig 3). SALM1-5 puncta differentially overlapped with presynaptic marker VGluT1 and postsynaptic marker Homer (Fig 3B-C). SALM1 was highly enriched at

Figure 2. SALM1-5 localize to axons and dendrites of mouse hippocampal neurons. Example

images of sandwich‐cultured mouse hippocampal neurons stained at DIV21 for endogenous SALM1-5 (red), dendritic marker MAP2 (blue), or the axonal marker SMI‐312 (green). Boxes indicate area of zoom. Zooms are shown enlarged on the right. Bars = 10 μm in full neuron images. Bars = 5 μm in zoomed images.

31

Figure 3. SALM1-5 differentially localize to synapses of excitatory hippocampal neurons. A)

Example images of neurites from sandwich‐cultured mouse hippocampal neurons stained at DIV21 for endogenous SALM1-5 (red), dendritic marker MAP2 (blue), and presynaptic marker VGluT1 or postsynaptic marker Homer (green). Full neuron images are shown in Fig. S3 and S4. Arrows indicate puncta depicted in B. Bars = 5 μm. B) Examples of overlap between SALM1-5 puncta (red) and the presynaptic marker VGluT1 or postsynaptic marker Homer (green). C) Average Manders overlap coefficient ± SEM of SALM1-5 overlapping with synaptic markers VGluT1 or Homer. The n is indicated in the bars and represents the total number of zoomed images analyzed / the total number of neurons / the total number of independent cultures. Student’s t-tests with Bonferroni correction were used (P = 0.729 for SALM5). ***P < 0.0002, ns = not significant.

32

Figure 4. SALM1 is present in developing neurons. A) Example images of sandwich cultured

mouse hippocampal neurons stained at DIV9 or DIV16 for total endogenous SALM1 (green), dendritic marker MAP2 (blue) and the axonal marker SMI-312 (red), or synapse markers Homer or VGluT1 (red). Boxes indicate area of zoom. Zooms are shown enlarged on the right. Arrows indicate puncta depicted in B. Bars = 10μm in full neuron images. Bars = 5μm in zoomed images. B) Example images showing differential overlap of SALM1 with Homer or VGluT1 at DIV9 and DIV16. Bars = 1μm. C) Average Manders overlap coefficient ± SEM of SALM1 overlapping with synaptic markers VGluT1 or Homer at DIV9 and DIV16. The n is indicated in the bars and represents the total number of zoomed images analyzed / the total number of neurons / the total number of independent cultures. Student’s t-tests with Bonferroni correction were used, **P < 0.005, ***P < 0.0005.

33

synapses and overlapped more with VGluT1 compared to Homer. SALM2, 3 and 4 overlapped more with Homer compared to VGluT1 and SALM5 overlapped with Homer and VGluT1 equally. Together, these data demonstrate the specificity of the new antibodies and suggest a differential subcellular distribution of SALMs in hippocampal neurons.

2.3.3 SALM1 is present in developing neurons and preferentially localizes to presynaptic terminals

Since SALM1 was identified as interactor of the presynaptic CASK/Mint1/Lin7b organizer complex, we decided to further investigate the subcellular distribution of SALM1. SALM1 was present in both axons and dendrites of developing hippocampal neurons at DIV9 as well as in mature neurons at DIV16 (Fig. 4A). SALM1 distribution was similar throughout development with SALM1 overlapping better with presynaptic marker VGluT1 compared to postsynaptic marker Homer (Fig.4A-C).

Immunoelectron microscopy was used to examine the subsynaptic localization of SALM1 in adult mouse brain tissue sections (Fig. 5 and Fig. S5). SALM1 was mostly detected in presynaptic terminals (~60%) of P75 mouse hippocampus. SALM1 immunoreactivity was less abundantly detected (~40%) at postsynaptic terminals. Hence, in line with being a binding partner for the presynaptic organizer complex

Figure 5. SALM1 preferentially localizes to presynapses in adult mouse hippocampus. A)

Electron micrographs showing subsynaptic localization of endogenous SALM1 in mouse P75 hippocampal brain slices. SALM1 immunogold particles are detected in presynapses (orange and magenta arrows) and postsynapses (cyan and blue arrows). Bars = 100 nm. B) Mean percentage of gold particles ± SEM detected in pre‐ versus postsynapses of mouse hippocampal slices stained for SALM1. Percentages are based on detected gold particles in 32 synapses in hippocampal brain slices of three different animals (Mann–Whitney U‐tests with Bonferroni correction, ns = not significant, *P < 0.025).

34

CASK/Mint/Lin7b, SALM1 is preferentially localized to excitatory presynaptic terminals in mouse hippocampus.

2.4 Discussion

In this study, we identified adhesion molecule SALM1 as an interactor of the presynaptic CASK/Mint1/Lin7b organizer complex. Endogenous SALM1-5 differentially localized to synapses of mouse hippocampal excitatory neurons. SALM1 localization to synapses was consistent throughout development and preferentially localized to presynapses of adult mouse hippocampal neurons.

SALMs have collectively been described as postsynaptic proteins based on several interaction studies (Lie et al., 2018), but their subcellular distribution was not established. We detected all five SALMs in axons and dendrites of mature hippocampal neurons indicating that all SALMs may localize to both pre- and postsynapses. Consistently, all SALMs partially overlapped with both presynaptic marker VGluT1 and postsynaptic marker Homer, but in different ratios. SALM1 was mostly enriched at synapses as indicated by high overlap coefficients and overlapped better with VGluT1 than Homer suggesting that SALM1 is more abundant at presynapses compared to postsynapses. This was further confirmed by our immunoelectron microscopy data. In contrast, SALM2-4 overlapped more with Homer than VGluT1 suggesting a preferential postsynaptic localization for these proteins. SALM5 overlapped with VGluT1 and Homer equally. SALM2-5 furthermore localized more extrasynaptically compared to SALM1 (as indicated by overall lower overlap coefficients). Differential pre- versus postsynaptic distribution ratios have also been described for other synaptic proteins (e.g. Neurexins and NMDA receptors) (Bouvier et al., 2015; Banerjee et al., 2016; Ribeiro et al., 2019). These proteins function differently at pre- and postsynapses and their targeting is highly regulated. Hence, SALMs distribute to both pre- and postsynapses of hippocampal neurons with different ratios and may also function extrasynaptically.

We detected SALM1 in developing DIV9 hippocampal neurons consistent with previous findings that SALM1 is expressed early in development (Morimura et al., 2006; Wang et al., 2006). SALM1 consistently overlapped more with VGluT1 than Homer at different developmental time points (DIV9, DIV16 and DIV21). However, SALM1 localized more extrasynaptically in developing neurons as indicated by overall

35

lower overlap coefficients at DIV9 compared to DIV16 an DIV21 mature neurons. These findings suggest that SALM1 functions in synapses throughout development and may also function extrasynaptically especially during early development.

Immunoelectron microscopy showed preferential localization of SALM1 to presynapses of adult mouse hippocampal neurons. However, only a fraction of endogenous SALM1 molecules were detected on the synaptic membrane and the majority was detected in the presynaptic cytomatrix. Surface levels of SALM1 may be dynamically regulated between cytosolic and membrane pools. A tight regulation of membrane localization has also been described for other CAMs such as Neurexins, Neuroligins and L1-CAMs, which are important for regulating synaptic function and growth cone migration (Thoumine, 2008; Fu and Huang, 2010; Peixoto et al., 2012).

Our and previous data showed that SALMs are highly N-glycosylated proteins (our data and Ko et al. (2006); Mah et al. (2010). N-glycosylation is widely accepted to determine protein structure and function. In addition, increasing evidence suggests an important role for site-specific N-glycosylation in regulating trans-synaptic interactions between synaptic adhesion molecules. For example, N-glycosylation promotes SynCAM1 trans-interactions, but inhibits SynCAM2 trans-interactions (Fogel et al., 2010). N-glycosylation in Neuroligin promotes trans-interaction with Neurexins and regulates isoform-specific interactions between Neurexins and Neuroligins (Comoletti et al., 2003; Chih et al., 2006). Hence, N-glycosylation of SALMs may differentially regulate structure, interaction capacity and function of specific SALM isoforms.

In conclusion, three lines of evidence, based on protein–protein interactions, immunocytochemistry, and immunoelectron microscopy, indicate that endogenous SALM1 accumulates in pre‐ and postsynaptic locations of hippocampal neurons and is most abundant on the presynaptic side. This preferential localization to the presynaptic side has not been described for SALM proteins before. The function of SALM1 at the presynapse remains to be determined. The CASK/Mint1/Lin7b organizer complex was previously shown to link cell adhesion to synaptic vesicle release via interactions with Neurexins and Munc18 (Butz et al., 1998). Hence, our findings that SALM1 preferentially localizes to the presynapse and interacts with the CASK/Mint1/Lin7b organizer complex suggests a presynaptic role for SALM1 in linking cell adhesion to synaptic vesicle release.

36

2.5 Materials & Methods

Contact for reagent and resource sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Matthijs Verhage (matthijs@cncr.vu.nl).

Experimental model and subject details

Animals

Wild‐type E18 mouse embryos were obtained by cesarean section of pregnant female C57/Bl6 mice. Newborn P0‐P1 pups from pregnant female Wistar rats were used for glia preparations. All animal experiments were performed according to the Dutch legislations for the use of laboratory animals.

HEK cell cultures

Human Embryonic Kidney 293T cells (HEK cells) were cultured in DMEM/F12 medium with L‐glutamine supplemented with 10% FCS, 1% NEAA, and 1% Pen/Strep (all Gibco). Cells were plated in 6‐well culture plates (Greiner) at equal densities 1 day prior to transfection at 37°C, 5% CO2.

Method details

Proteomics

P2+microsome fractions from adult mice were solubilized with 1% detergent extraction buffer (1% n‐Dodecyl β‐D‐maltoside (DDM), 150 mM NaCl, 25 mM HEPES, pH 7.4, and protease inhibitor (Roche)). Extracts were incubated with antibodies against CASK, Lin7b, Mint1, SALM1, or GRIA2 at 4°C overnight on a mechanical rotator. 50 μl of protein A/G PLUS‐Agarose beads (Santa Cruz) was washed four times with washing buffer (0.1% DDM, 150 mM NaCl, 25 mM HEPES, pH 7.4) before it was added to the samples for 1 h at 4°C. The buffer was completely removed using an insulin syringe before storing the samples at −20°C until further use.

An SDS–PAGE LC‐MS/MS approach was used for protein identification as described previously (Chen et al., 2011). In short, after separation on the SDS–PAGE gel proteins were trypsin digested. The resulting peptides were separated on a capillary C18 column using a nano LC‐ultra 1D plus HPLC system (Eksigent) and analyzed online with an electrospray LTQ‐Orbitrap Discovery mass spectrometer (Thermo Fisher Scientific). MS‐MS spectra were searched against the UniProt proteomics

37

database (version 2013‐01‐06) with MaxQuant software (version 1.3.0.5). Methionine oxidation and protein N‐terminal acetylation were set as variable modifications, and cysteine alkylation with acrylamide was set as fixed modification. The maximum mass deviations of parent and fragment ions were set to 6 ppm and 0.5 Da, respectively. Trypsin was chosen as the digestion enzyme, and the maximum missed cleavage was set at 2. Each valid protein hit should contain at least one unique peptide. The false discovery rates of both peptides and proteins were set within a threshold value of 0.01. The MaxLFQ algorithm in MaxQuant was used to normalize the data.

Dissociated hippocampal sandwich cultures

To make sandwich cultures, flattened tweezers were heated and shortly placed on the bottom of wells of 12‐well plastic culture plates (Greiner) to create small extrusions.

Plates were sprayed using an airbrush containing a Poly‐D‐Lysine solution consisting of 0.5 mg/ml Poly‐D‐Lysine (Sigma), 3.66 mg/ml collagen (BD biosciences), and 17 mM acetic acid (Sigma). Plates were UV‐sterilized for 20 min before further use. Astrocytes were plated at 25K/well in pre‐warmed DMEM medium supplemented with 10% FCS, 1% NEAA, and 1% Pen/Strep (all Gibco). After 4–5 days, DMEM medium was replaced by Neurobasal medium supplemented with 2% B‐27, 1.8% HEPES, 0.25% glutamax, and 0.1% Pen/Strep (all Gibco) and neurons were added to the cultures.

Hippocampi were dissected from embryonic day 18 (E18) wild‐type C57/Bl6 mice and collected in ice‐cold Hanks Buffered Salt Solution (HBSS; Sigma), buffered with 7 mM HEPES (Invitrogen). Tissues were incubated in Hanks‐HEPES with 0.25% trypsin (Invitrogen) for 20 min at 37°C. After washing, neurons were triturated using a fire‐polished Pasteur pipette and counted in a Fuchs‐Rosenthal chamber. The cells were plated in Neurobasal medium supplemented with 2% B‐27, 1.8% HEPES, 0.25% glutamax, and 0.1% Pen/Strep (all Gibco) on Poly-L-Ornithine coated 18 mm glass coverslips (see below).

Glass coverslips (18 mm) were washed in 96% ethanol and flamed dry. Coverslips were incubated with sterile H2O (Baxter) containing 2.5 μg/ml Laminin

(Sigma) and 0.0005% Poly‐L‐Ornithine (Sigma) for 2 h at 37°C. Coverslips were then washed three times with sterile H2O (Baxter), and Neurobasal medium supplemented

with 2% B‐27, 1.8% HEPES, 0.25% glutamax, and 0.1% Pen/Strep (all Gibco) was added. Neurons were plated at a density of 15K per well and allowed to attach to the

38

coverslip for 24 h. DMEM medium on glia cultures was then replaced for Neurobasal medium supplemented with 2% B‐27, 1.8% HEPES, 0.25% glutamax, and 0.1% Pen/Strep (all Gibco). Glass coverslips containing the neurons were then placed on top of the extrusions with neurons faced down to allow a thin film of Neurobasal medium between neurons and glia.

SALM1-5 antibodies

We used rabbit polyclonal SALM1 antibody targeted against the 638–788 cytoplasmic amino acid fragment of mouse SALM1 developed by Synaptic Systems in all experiments labeling SALM1 except immunoelectron microscopy. For immunoEM, we used a different antibody (ProSci Cat#5067, RRID:AB_10906317) that recognized SALM1 with high specificity (Fig. S1C). The following antibodies against SALM2-5 were used: rabbit polyclonal SALM2 (Synaptic Systems Cat#294203), rabbit polyclonal SALM3 (Synaptic Systems Cat#294303), rabbit polyclonal SALM4 (Synaptic Systems Cat#294403) and rabbit polyclonal SALM5 (Synaptic Systems).

Plasmids

To investigate the interaction between SALM1 and CASK, full‐length CASK was picked up from a cDNA library and was subcloned with IRES‐mCherry. The

cytoplasmic part of SALM1 was obtained through PCR and was subcloned with a V5 tag at the N‐terminus (V5‐cytoplasmic SALM1). V5‐cytoplasmic SALM1ΔPDZ was obtained by PCR on V5‐cytoplasmic SALM1.

To test the specificity of the SALM1-5 antibodies, full‐length SALM1, SALM2, SALM3, SALM4, and SALM5 were cloned from a yeast two‐hybrid cDNA library. A mCherry tag was subcloned between amino acids 20 and 21 of SALM1, and a 3xFlag tag was placed at the N‐terminus of SALM2, SALM3, SALM4, and SALM5. All constructs were driven by CMV promoters.

HEK cell transfection

Human Embryonic Kidney 293T cells (HEK cells) were cultured in DMEM/F12 medium with L‐glutamine supplemented with 10% FCS, 1% NEAA, and 1% Pen/Strep (all Gibco). Cells were plated in 6‐well culture plates (Greiner) at equal densities 1 day prior to transfection at 37°C, 5% CO2. At the day of transfection, cell confluency

reached 80%. Cells were transfected using a calcium/phosphate transfection method. Briefly, the desired cDNA at a concentration of 1 μg/μl was diluted 1:40 (or 1:80 in case of co‐transfection) in HBS [140 mM NaCl, 1.5 mM Na2HPO4.H2O, and 50 mM HEPES

39

adjusted to pH 7.05 with NaOH (all Sigma)]. An equal volume of 250 mM CaCl2 was

added dropwise to the DNA‐HBS solution under constant mild vortexing. The DNA‐ HBS‐CaCl2 was added dropwise to wells containing HEK cells. Cells were then

incubated with the transfection mix for ~20 h at 37°C, 5% CO2. To investigate

posttranslational modification of SALM1, 2.5 μg/ml Tunicamycin (Sigma) or 0.25% DMSO (Sigma) was added to the medium after ~3 h of incubation with the transfection mix. Medium was replaced after the ~20‐h incubation period, and cells were incubated at 37°C, 5% CO2 for another ~5 h before further use.

Western blot analysis

Cultured HEK cells were lysed with SDS loading buffer and boiled at 90°C for 5 min. Co‐IP samples were boiled at 90°C for 5 min and spun down for 1 min at 12,000 g. Samples were loaded onto a 1 mm stacking gel consisting of 13.3% Acrylamide/Bis solution, 29:1 (30% w/v) (Serva Electrophoresis), 12.4% 1M Tris (pH 6.8) (AppliChem), 0.2% SDS (VWR Chemicals), 0.1% APS (AppliChem), and 0.01% TEMED (Electran, VWR Chemicals). PageRuler™ Prestained Protein Ladder (Thermo Scientific) was loaded as a reference for molecular weights. Loaded samples were then run through 8 or 10% running gels consisting of 26.6% (for 8% gels) or 33.2% (for 10% gels) Acrylamide/Bis solution, 29:1 (30% w/v) (Serva Electrophoresis), 24.9% 1.5M Tris (pH 8.8) (AppliChem), 0.13% SDS (VWR Chemicals), 0.07% APS (AppliChem), and 0.007% TEMED (Electran, VWR Chemicals) for 1 h at 50 mA. Proteins were transferred onto PVDF membranes (Immuno‐Blot, Bio Rad) at 4°C for 1.5 h at 35V. Blots were blocked with PBS/Tween (PBS supplemented with 0.1% Tween 80 (Ferak Berlin GmbH)) containing 22 mg/ml skim milk powder (Merck) and 4.8 mg/ml Albumin Bovine Serum (Across Organics, Fisher Scientific) for 45 min. at room temperature. Blots were then incubated with primary antibodies rabbit polyclonal mCherry (GeneTex Cat# GTX128508, RRID:AB_2721247), mouse monoclonal Flag (Sigma‐Aldrich Cat# F1804, RRID:AB_262044) and the five SALM antibodies described above (all Synaptic Systems) diluted in PBS/Tween overnight at room temperature. Blots were washed three times with PBS/Tween and incubated with the following alkaline phosphate secondary antibodies for 1 h at room temperature: alkaline phosphatase AffiniPure goat anti‐mouse IgG (H+L) (Jackson ImmunoResearch) and alkaline phosphatase AffiniPure goat anti‐rabbit IgG (H+L) (Jackson ImmunoResearch). Blots were washed three times with PBS/Tween and incubated with AttoPhos (Promega) for 5 min. Signals

40

were detected using a FUJIFILM FLA‐5000 imaging system and ImageReader FLA5000 software (version 2.0).

Immunocytochemistry

Coverslips were fixed using 4% PFA (Electron Microscopy Sciences) for 30 min. Cells were washed three times with PBS and permeabilized for 5 min with 0.5% Triton X‐ 100 (Fisher Chemical). Aspecific binding sites were blocked by incubating cells with 0.1% Triton X‐100 (Fisher Chemical) and 2% normal goat serum (Gibco) for 30 min. Cells were incubated for 2 h with primary antibodies. The following primary antibodies were used: the five SALM rabbit polyclonal antibodies described above 1:100 (Synaptic Systems), chicken polyclonal MAP2 1:10,000 (Abcam Cat# ab75713, RRID:AB_1310432), guinea pig polyclonal VGluT1 1:5,000 (Millipore Cat# AB5905, RRID:AB_2301751), guinea pig polyclonal Homer 1:300 (Synaptic Systems Cat# 160 004, RRID:AB_10549720), mouse monoclonal Smi312 1:1,000 (BioLegend Cat# 837904, RRID:AB_2566782). After primary antibody incubation, cells were washed three times with PBS and incubated with secondary antibodies for 1 h. The following secondary antibodies were used: goat anti‐mouse Alexa 488 and 546, goat anti‐rabbit Alexa 488 and 546, goat anti‐guinea pig Alexa 488 and 546, and goat anti‐ chicken Alexa 647 (all Invitrogen). Coverslips were then washed three times with PBS and embedded in Mowiol (Calbiochem). All steps were performed at room temperature.

Confocal image acquisition and analysis

Stained cultures were imaged with a Zeiss LSM510 confocal microscope equipped with a 63× oil objective (N.A 1.4, plan apochromat) or a 40× oil objective (N.A 1.3, plan neoflux). Images were acquired using an AxioCam MRm (Zeiss) camera and Zeiss LSM510 software (version 4.2). All images were acquired at room temperature. Gain and amplifier offset were kept constant for different conditions for each experiment. The JACoP plugin in ImageJ (version 1.50a) was used to determine Manders overlap coefficients (Bolte and Cordelieres, 2006).

Electron microscopy

For immunogold labeling of SALM1, hippocampi of P75 mice were fixed using 4% PFA (Electron Microscopy Sciences) and 0.1% glutaraldehyde (Merck) in 0.1 M phosphate buffer (pH 7.4). Samples were embedded in increasing concentrations of gelatin at 37°C, infiltrated with 2.3 M sucrose at 4°C, and frozen in liquid nitrogen. A cryo‐

41

ultramicrotome (UC6, Leica) was used to cut tissue in 70‐nm‐thick sections. These were collected in 1% methyl‐cellulose and 1.2 M sucrose solution at −120°C. The sections were then transferred to formvar/carvon‐coated copper mesh grids, washed with PBS at 37°C, and treated with 0.1% glycine. Hippocampal sections were immunolabeled using a primary antibody against SALM1 (ProSci Cat#5067, RRID:AB_10906317) diluted 1:10 in PBS supplemented with 0.1% BSA. Antibodies were further labeled using Protein A‐10 nm gold (CMC, UMC Utrecht, Netherlands). Sections were counterstained on ice with 0.4% uranyl acetate in 1.8% methyl‐cellulose and were imaged on a Tecnai 12 BioTwin transmission electron microscope (FEI Company).

Co-immunoprecipitation

To investigate the potential interaction between SALM1 and CASK, HEK cells were transfected with full‐length CASK alone or together with V5‐tagged cytoplasmic SALM1, V5‐tagged cytoplasmic SALM1ΔPDZ, or empty vector V5‐pcDNA3.1. One day after transfection, medium was replaced for DMEM/F12 medium with l‐glutamine supplemented with 10% FCS, 1% NEAA, and 1% Pen/Strep (all Gibco). The next day, cells were scraped and collected in lysis buffer containing 50 mM Tris (pH7.5) (AppliChem), 1% Triton X‐100 (Fisher Chemicals), 1.5 mM MgCl2 (Sigma), 5 mM

EDTA (Applichem), 100 mM NaCl (Sigma), and 1× Protease inhibitor (Sigma) on ice. Lysates were spun down for 10 min. at 12,000 g at 4°C. Part of the supernatant was mixed with 5× SDS loading buffer and stored at −20°C to use later as input control. Protein A agarose beads were washed once with lysis buffer and combined with the remaining lysate. Samples were then tumbled for 1 h at 4°C followed by 1 min. centrifugation at 12,000 g. Supernatant was transferred to new tubes and incubated for 2 h at 4°C with the following antibodies: mouse monoclonal V5 (Abcam Cat# ab27671, RRID:AB_471093), and mouse monoclonal CASK (UC Davvis/NIH NeuroMab Facility Cat#73‐000, RRID:AB_10671954). Protein A agarose beads were pre‐blocked with blocking buffer existing of lysis buffer supplemented with 20% glycerol (VWR, BDH) and 0.2% chicken egg albumin (Sigma) for 1 h at 4°C. Beads were then washed once with lysis buffer and mixed with the lysate samples. After 1 h of incubation at 4°C, beads were washed five times alternately with low salt buffer consisting of 50 mM Tris (pH7.5) (AppliChem), 0.1% Triton X‐100 (Fisher Chemicals), 1.5 mM MgCl2 (Sigma), 5 mM EDTA (AppliChem), 100 mM NaCl (Sigma), and 1×

42

Protease inhibitor (Sigma) and with high salt buffer containing 50 mM Tris (pH7.5) (AppliChem), 0.1% Triton X‐100 (Fisher Chemicals), 1.5 mM MgCl2 (Sigma), 5 mM

EDTA (AppliChem), 200 mM NaCl (Sigma), and 1× Protease inhibitor (Sigma). Wash buffer was then fully removed using an insulin syringe (BD MicroFine), and SDS loading buffer was added to the beads. Samples were stored at −20°C until further use.

Statistical analysis

Statistical analysis was performed using SPSS (IBM SPSS Statistics 21). Kolmogorov– Smirnov and Shapiro–Wilk tests were used to test for normality of data distribution. Student’s t‐test was used to compare two groups when data were normally distributed. Bonferroni correction was used in case of multiple Student’s t‐tests within one data set.