Control of a Hippocampal Recurrent Excitatory Circuit by Cannabinoid Receptor-interacting Protein Gap43 Irene Maroto Complutense University - CIBERNED https://orcid.org/0000-0002-4156-166X Carlos Costas-Insua Complutense University - CIBERNED https://orcid.org/0000-0003-3446-6899 Coralie Berthoux Albert Einstein College of Medicine Estefania Moreno Andrea Ruiz-Calvo Complutense University - CIBERNED Carlos Montero-Fernández Centro de Investigación Biomédica en Red de Enfermedades Respiratorias Andrea Macías-Camero Complutense University - CIBERNED Ricardo Martín Universidad Complutense de Madrid Nuria García-Font Universidad Complutense de Madrid José Sánchez-Prieto Complutense University of Madrid Giovanni Marsicano Inserm https://orcid.org/0000-0003-3804-1951 Luigi Bellocchio INSERM Enric Canela https://orcid.org/0000-0003-4992-7440 Vicent Casado University of Barcelona https://orcid.org/0000-0002-1764-3825 Ismael Galve-Roperh IRYCIS- Complutense University https://orcid.org/0000-0003-3501-2434 David Fernandez de Sevilla Autonomous University of Madrid https://orcid.org/0000-0001-6344-0773 Ignacio Rodríguez-Crespo Complutense University - CIBERNED Pablo Castillo Albert Einstein College of Medicine https://orcid.org/0000-0002-9834-1801 Manuel Guzmán (  mguzman@quim.ucm.es ) Complutense University - CIBERNED Article Keywords: Cannabinoid, CB1 receptor, GAP43, neurotransmission, hippocampus Posted Date: October 18th, 2022 DOI: https://doi.org/10.21203/rs.3.rs-2128033/v1 License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License 1 CONTROL OF A HIPPOCAMPAL RECURRENT EXCITATORY CIRCUIT BY 2 CANNABINOID RECEPTOR-INTERACTING PROTEIN GAP43 3 4 5 6 7 8 9 10 Irene B. Maroto1,2,3, Carlos Costas-Insua1,2,3, Coralie Berthoux4, Estefanía Moreno5, Andrea Ruiz-Calvo1,2,3, Carlos Montero-Fernández1, Andrea Macías-Camero1, Ricardo Martín1,6, Nuria García-Font1,6, José Sánchez-Prieto1,6, Giovanni Marsicano7, Luigi Bellocchio7, Enric I. Canela5, Vicent Casadó5, Ismael Galve-Roperh1,2,3, David Fernández de Sevilla8, Ignacio Rodríguez-Crespo1,2,3, Pablo E. Castillo4,9, Manuel Guzmán1,2,3,* 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 1Department of Biochemistry and Molecular Biology, Instituto Universitario de Investigación Neuroquímica (IUIN), Complutense University, Madrid 28040, Spain 2Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED), Instituto de Salud Carlos III, Madrid 28029, Spain 3Instituto Ramón y Cajal de Investigación Sanitaria (IRYCIS), Madrid 28034, Spain 4Dominick P. Purpura Department of Neuroscience, Albert Einstein College of Medicine, Bronx, NY 10461, USA 5Department of Biochemistry and Molecular Biomedicine, Faculty of Biology and Institute of Biomedicine of the University of Barcelona, University of Barcelona, Barcelona 08028, Spain 6Instituto de Investigación Sanitaria del Hospital Clínico San Carlos (IdISSC), Madrid 28040, Spain 7Institut National de la Santé et de la Recherche Médicale (INSERM) and University of Bordeaux, NeuroCentre Magendie, Physiopathologie de la Plasticité Neuronale, U1215, Bordeaux 33077, France 8Department of Anatomy, Histology and Neuroscience, School of Medicine, Autónoma University, Madrid 28029, Spain 9Department of Psychiatry and Behavioral Sciences, Albert Einstein College of Medicine, Bronx, NY 10461, USA Running title: Control of cannabinoid signaling by GAP43 Keywords: Cannabinoid; CB1 receptor; GAP43; neurotransmission; hippocampus *To whom correspondence should be addressed: Manuel Guzmán, PhD Department of Biochemistry and Molecular Biology Instituto Universitario de Investigación Neuroquímica (IUIN) Complutense University, Madrid 28040, Spain Email: mguzman@quim.ucm.es Title: 12 words; 103 characters (with spaces) Total number of words: Abstract, 150; Introduction, 529; Results, 2,469; Discussion, 1,005; Methods, 2,311 References: 70 Figures: 6; Supplementary figures: 4 Tables: 0; Supplementary tables: 1 1 50 ABSTRACT 51 52 The type-1 cannabinoid receptor (CB1R) is widely expressed in both excitatory and 53 inhibitory nerve terminals, and its activation, by suppressing neurotransmitter release, 54 modulates neural circuits and brain function. While the interaction of CB1R with various 55 intracellular proteins is thought to alter receptor signaling, the identity and role of these 56 proteins are poorly understood. Using a high-throughput proteomic analysis 57 complemented with an array of in vitro and in vivo approaches in the mouse brain, we 58 report that the C-terminal, intracellular domain of CB1R interacts specifically with 59 growth-associated protein of 43 kDa (GAP43). The CB1R-GAP43 interaction occurs 60 selectively at mossy cell axon boutons, which establish excitatory synapses with 61 dentate granule cells in the hippocampus. This interaction impairs CB1R-mediated 62 suppression of mossy cell to granule cell transmission, thereby inhibiting cannabinoid- 63 mediated anti-convulsant activity in mice. Thus, GAP43 acts as a synapse type-specific 64 regulatory partner of CB1R that hampers CB1R-mediated effects on hippocampal circuit 65 function. 66 2 67 INTRODUCTION 68 69 The endocannabinoid system comprises cannabinoid receptors, their lipid ligands (the 70 so-called endocannabinoids), and the enzymatic machinery required for 71 endocannabinoid synthesis, deactivation, and bioconversion1,2. The endocannabinoids 72 2-arachidonoylglycerol (2-AG) and anandamide, as well as the exogenous cannabinoid 73 Δ9-tetrahydrocannabinol (THC), the main psychoactive component of cannabis, bind to 74 and activate type-1 and type-2 cannabinoid receptors (CB1R and CB2R, respectively), 75 which are evolutionarily-conserved members of the G protein-coupled receptor (GPCR) 76 superfamily1,2. CB1R is one of the most abundant GPCRs in the mammalian central 77 nervous system (CNS), and its activation mediates retrograde suppression of 78 neurotransmitter release in a short-term and long-term manner3,4. Thus, CB1R 79 regulates a plethora of body functions, including learning and memory, emotions, 80 feeding and energy metabolism, pain response, and motor behavior5,6. Despite the vast 81 number of reports on CB1R-modulated neurobiological processes, studies addressing 82 the precise molecular mechanisms and signaling partners of CB1R at the synapse level 83 remain scarce. CB1R triggers a wide range of downstream cascades that regulate 84 synaptic function and neuronal activity in a markedly context-dependent manner7. We 85 and others have previously proposed that the interaction of CB1R with various 86 cytoplasmatic proteins2,8,9, as well as plasma-membrane GPCRs10,11, may fine-tune 87 CB1R signaling in vivo. However, the precise functional relevance of these CB1R 88 protein-protein interactions in the brain has not been fully elucidated yet. 89 90 CB1Rs are present in both excitatory and inhibitory nerve terminals and their activation 91 can modify the excitatory/inhibitory balance. Studies conducted on conditional CB1R 3 92 knockout and genetic-rescue mice have revealed that CB1Rs located on excitatory 93 presynaptic boutons, despite their moderate levels of expression compared to 94 GABAergic terminals3,12,13, act as a synaptic circuit breaker that is crucial for the control 95 of brain excitability14,15. Thus, activation of glutamatergic-neuron CB1Rs mediates key 96 (endo)cannabinoid-evoked processes such as hyperphagia16, anxiolysis17, 97 neuroprotection18, and anti-convulsion19. Particularly high levels of glutamatergic- 98 neuron CB1Rs occur in the axon boutons of hilar mossy cells (MCs) of the dentate 99 gyrus (DG)13,19,20, a region that critically processes information from the entorhinal 100 cortex (EC) to the hippocampal formation. MC boutons are located in the inner 101 molecular layer (IML) impinging on proximal dendrites of granule cells (GCs), the 102 foremost excitatory neurons in the DG. In turn, GCs project back to MCs, thereby 103 establishing an associative GC-MC-GC excitatory circuit that gates information transfer 104 from the EC to CA3, and is involved not only in processing various forms of memory 105 but also in driving hyperexcitability-evoked epileptic seizures21,22. Unlike CB1Rs located 106 on EC-projecting axon boutons, whose activation can even potentiate excitatory 107 synaptic transmission23, activation of CB1Rs located on MCs suppresses synaptic 108 transmission and prevents a long-term potentiation of MC-GC synaptic transmission 109 and GC output24,25, thus supporting an anti-convulsant action19. 110 111 Here, using a high-throughput proteomic approach complemented with a wide array of 112 in vitro and in vivo assays, we unveil that CB1R interacts specifically with growth- 113 associated protein of 43 kDa (GAP43; aka neuromodulin), a major presynaptic protein 114 that is involved in neurite outgrowth, axonal regeneration, and synaptic plasticity26. 115 Moreover, we show that the CB1R-GAP43 interaction is enhanced by GAP43 116 phosphorylation, occurs selectively in MC axons impinging on GC dendrites, hampers 4 117 CB1R-mediated depression at MC-GC synapses, and impairs cannabinoid-evoked anti- 118 convulsant activity. Thus, our findings identify GAP43 as a novel CB1R-interacting 119 protein that regulates receptor function in a synapse-specific manner. 120 5 121 RESULTS 122 123 Identification of GAP43 as a CB1R-interacting protein 124 We initially identified GAP43 as a potential CB1R-interacting protein in a high- 125 throughput screening conducted by affinity chromatography and subsequent proteomic 126 analysis. As the large C-terminal domain (CTD) encompasses the bulk of the 127 cytoplasmic domain of CB1R, we used recombinant hCB1R-CTD (amino acids 408-472) 128 as bait. A whole-brain homogenate was passed through a lectin-hCB1R-CTD 129 Sepharose 4B column, and after washing and elution with lactose, the resulting 130 proteins were digested and subjected to tandem MS/MS (Fig 1A). We obtained a list of 131 ~50 potential CB1R-interacting candidate proteins (Suppl Table 1). While some of the 132 hits, such as plasma membrane Ca2+ ATPases, G-protein α subunits (specifically, 133 Gαi1), Na+ and Cl--dependent GABA transporters, Hsp70, and MAPK family members 134 coincided with those found in similar high-throughput studies27,28, our list also included 135 the pleiotropic protein GAP43/neuromodulin. A Gene ontology (GO) enrichment-based 136 cluster analysis of the list of proteins was performed using the STRING Database and 137 MCL Clustering. We identified two enriched functional GO terms, both including CB1R 138 and GAP43: GO.0008037-Cell recognition (with matching proteins CNR1, CRTAC1, 139 GAP43 and MFGE8), and GO.0008038-Neuronal recognition (with matching proteins 140 CNR1, CRTAC1 and GAP43) (p = 0.0384 for each GO term). Based on its presynaptic 141 localization and anatomical distribution in the CNS (see below), which raised the 142 possibility of a functional interaction with CB1R, we focused our further analyses on 143 GAP43. 144 6 145 As a first approach to validate a potential GAP43-CB1R interaction, we used an in vitro 146 fluorescence polarization-based, protein-protein binding assay (Fig 1B). We tested a 147 fixed concentration of purified, 5-IAF-labeled hCB1R-CTD and increasing 148 concentrations of purified, unlabeled hGAP43. A saturating polarization curve with Kd = 149 38.2 ± 10.8 µM (n = 3 experiments) was obtained, which supports a direct, specific, 150 and high-affinity interaction between GAP43 and CB1R-CTD. To assess the GAP43- 151 CB1R interaction in neural tissue, we performed co-immunoprecipitation assays in 152 primary mouse hippocampal neurons (Fig 1C, upper panel) as well as in mouse whole- 153 hippocampus extracts (Fig 1C, lower panel). These experiments indicated that 154 endogenous GAP43 and CB1R interact in the mouse brain in vivo. As both GAP43 and 155 CB1R are predominantly located on presynaptic boutons3,26, we tested whether they 156 interact at this precise subcellular site by using synaptosomal preparations isolated 157 from mouse hippocampus. Immunostaining of synaptosomes revealed that GAP43 and 158 CB1R were present in 19.7 ± 3.1 % (GAP43) or 20.1 ± 4.3 % (CB1R) of total 159 synaptophysin 1-positive buttons, and that 8.2 ± 1.8 % of total synaptophysin 1-positive 160 buttons were double-positive for GAP43 and CB1R (n = 5 synaptosomal preparations; 161 Fig 1D), thus pointing to a restricted location of the complexes within the whole 162 hippocampus. Moreover, to evaluate a potential protein-protein interaction in 163 synaptosomes, we performed in situ proximity ligation assay (PLA) experiments. 164 Hippocampal synaptosomal preparations from WT mice showed an overt GAP43-CB1R 165 PLA-positive signal, which was notably reduced in synaptosomes from CB1R-deficient 166 (Cnr1-/-) mice (Fig 1E). Taken together, these observations support a physical 167 interaction between GAP43 and CB1R at a selective pool of mouse hippocampal 168 presynaptic boutons. 169 7 170 Phosphorylation of GAP43 at S41 facilitates its interaction with CB1R 171 Previous studies had shown that phosphorylation of GAP43 at S41 is critical for its 172 biological activity26,29,30. We thus designed two mutant versions of GAP43 (harboring a 173 phospho-resistant S41A or phospho-mimetic S41D point mutation, respectively) to 174 modulate the activation state of the protein (Fig 2A). First, PLA was conducted in 175 HEK293T cells co-transfected with a myc-tagged CB1R plus the different forms of 176 GAP43, namely, GFP-GAP43(WT), GFP-GAP43(S41D) or GFP-GAP43(S41A). 177 GAP43(WT)-CB1R and GAP43(S41D)-CB1R complexes were readily detectable and 178 quantified as PLA-positive puncta in GFP-positive cells, while remarkably lower 179 complex levels were found in cells transfected with GFP-GAP43(S41A) (Fig 2B). 180 Second, cells were co-transfected with HA-CB1R and each of the GAP43 mutants. 181 Upon HA-CB1R immunoprecipitation using anti-HA antibody and blotting with an anti- 182 pan-GAP43 antibody, GAP43(S41D) was the predominant co-immunoprecipitated form 183 of the protein (Fig 2C). Third, BRET experiments conducted with an Rluc-tagged 184 version of CB1R showed a positive and saturating BRET signal for CB1R-RLuc plus 185 GFP-GAP43(WT) or GFP-GAP43(S41D), while the pair CB1R-RLuc/GFP- 186 GAP43(S41A) gave a basically linear, non‐specific BRET signal (Fig 2D). 187 188 We next asked whether GAP43 binding affects CB1R activity. To this end, we 189 performed dynamic mass redistribution (DMR) assays to quantify changes in the 190 overall signaling triggered by agonist-evoked receptor activation (Fig 2E). We and 191 others have previously used this approach to evaluate CB1R signaling in response to 192 various manipulations9–11. When HEK293T cells expressing CB1R were treated with the 193 CB1R agonist WIN-55,212-2 (100 nM), we found that both GAP43(WT) and 194 GAP43(S41D) blunted receptor action, and this inhibitory effect was not evident with 8 195 GAP43(S41A). Taken together, these findings indicate that GAP43 and CB1R interact 196 specifically in vitro, and that this interaction requires GAP43 phosphorylation and 197 inhibits CB1R. 198 199 GAP43 interacts with CB1R in mossy cell axon boutons of the dentate gyrus 200 To map the GAP43-CB1R interaction in the mouse brain, we first performed 201 immunofluorescence co-localization assays. Consistent with previous studies13,19,20,31,32, 202 we found that both GAP43 and CB1R were densely expressed in the IML of the DG 203 (Suppl Fig 1A). To evaluate the neurochemical identity of the immunolabeled 204 presynaptic boutons, we used conditional knockout mice in which the CB1R-encoding 205 gene (Cnr1) had been selectively deleted from forebrain GABAergic neurons (hereafter 206 GABA-Cnr1-/- mice)19 or dorsal telencephalic glutamatergic neurons (hereafter Glu- 207 Cnr1-/- mice)19. We found abundant double-positive puncta for GAP43 and (conceivably 208 glutamatergic-neuron) CB1R in the IML of GABA-Cnr1-/- mice, while colocalization 209 between GAP43 and (conceivably GABAergic-neuron) CB1R was essentially absent in 210 Glu-Cnr1-/- animals (Suppl Fig 1B). The glutamatergic boutons of the IML likely 211 correspond to MC axons impinging on proximal dendrites of GCs21. Thus, triple-positive 212 puncta for (conceivably glutamatergic-neuron) CB1R, GAP43, and calretinin (a well- 213 known MC marker21) were visible in high-magnification micrographs of the IML of 214 GABA-Cnr1-/- mice (Suppl Fig 1C). 215 216 We subsequently conducted PLA experiments in the IML of GABA-Cnr1-/- and Glu- 217 Cnr1-/- mice to seek for CB1R-GAP43 complexes. Consistent with our immunostaining 218 data, an overt PLA signal, visualized as positive puncta, was found in Cnr1fl/fl and 219 GABA-Cnr1-/- mice, notably diminishing in Glu-Cnr1-/- and full Cnr1-/- animals (Fig 3A). 9 220 To unequivocally ascribe CB1R-GAP43 complexes to glutamatergic boutons, we used 221 a Cre-mediated, lineage-specific CB1R genetic rescue strategy from a Cnr1-null 222 background (hereafter Stop-Cnr1 mice; Fig 3B)33,34. Thus, we rescued CB1R 223 expression selectively in dorsal telencephalic glutamatergic neurons (hereafter Glu- 224 Cnr1-RS mice) or forebrain GABAergic neurons (hereafter GABA-Cnr1-RS mice). As a 225 control, a systemic CB1R expression-rescue was conducted (hereafter Cnr1-RS mice). 226 The PLA signal of CB1R-GAP43 complexes was markedly restored in Cnr1-RS and 227 Glu-Cnr1-RS mice. In contrast, no significant rescue of complex expression was 228 observed in GABA-Cnr1-RS or Stop-Cnr1 animals. Taken together, these findings 229 support that the GAP43-CB1R interaction occurs at glutamatergic neurons in the IML, 230 presumably on MC axon boutons. 231 232 Phosphorylated GAP43 inhibits CB1R function at mossy cell to granule cell 233 synapses 234 CB1Rs at the MC-GC synapse mediate a form of short-term plasticity known as 235 depolarization-induced suppression of excitatory transmission (DSE), and tonic 236 suppression of glutamate release19,24,25. We therefore examined whether these CB1R- 237 mediated effects could be affected by GAP43 interaction. To this end, we generated 238 AAV1/2 vectors encoding phospho-mimetic GAP43 [AAV1/2-CBA-GAP43(S41D)-CFP] 239 and phospho-resistant GAP43 [AAV1/2-CBA-GAP43(S41A)-CFP] fused to the 240 fluorescent reporter CFP. An empty vector (AAV1/2-CBA-CFP) was used as control. 241 These viral vectors were injected unilaterally into the hilus (where MC somata are 242 located) of 3-4-week-old WT mice, and electrophysiological recordings were then 243 performed in the contralateral DG to activate commissural MC axons expressing the 244 vectors35. We confirmed the presence of CFP-positive fibers selectively in the IML of 10 245 the contralateral DG (Fig 4A). Whole-cell patch-clamp recordings were performed from 246 GCs, 3 weeks after viral injection, and excitatory postsynaptic currents (EPSCs) in GCs 247 were evoked by electrical stimulation in the IML. First, we observed that both paired- 248 pulse ratio [Fig 4B PPR, Control: 1.16 ± 0.06, n = 9 cells; GAP43(S41A): 1.15 ± 0.05, n 249 = 8 cells; GAP43(S41D): 0.92 ± 0.04, n = 12 cells; F(2, 26) = 8.146; Control vs 250 GAP43(S41D) and GAP43(S41A) vs GAP43(S41D), p < 0.01 by one-way ANOVA] and 251 coefficient of variation [Fig 4B CV, Control: 0.29 ± 0.02; GAP43(S41A): 0.30 ± 0.02; 252 GAP43(S41D): 0.20 ± 0.01; F(2, 26) = 13.77; Control vs GAP43(S41D) and 253 GAP43(S41A) vs GAP43(S41D), p < 0.01 by one-way ANOVA] were decreased in 254 GAP43(S41D) compared to control vector or GAP43(S41A)-injected mice, suggesting 255 that phosphorylation of GAP43 increases glutamate release probability at MC-GC 256 synapses. This observation would be consistent with a reduction in CB1R tonic 257 activity25 at GAP43(S41D)-expressing MC-GC synapses. In addition, the magnitude of 258 DSE was reduced in GAP43(S41D)-injected mice compared to control vector-injected 259 or GAP43(S41A)-injected mice [Fig 4C, Control: 69.3 ± 1.8 % of baseline, n = 8 cells; 260 GAP43(S41A): 69.9 ± 4.4 % of baseline, n = 9 cells; GAP43(S41D): 91.0 ± 2.8 % of 261 baseline, n = 10 cells; F(2, 24) = 15.48; Control vs GAP43(S41D) and GAP43(S41A) vs 262 GAP43(S41D), p < 0.01 by one-way ANOVA]. 263 264 To directly assess CB1R function, we tested the effect of the agonist WIN-55,212-2 on 265 extracellular MC-GC synaptic responses (i.e., extracellular field excitatory postsynaptic 266 potentials, or fEPSPs) recorded in the IML. While WIN-55,212-2 (5 μM for 25 min) 267 decreased MC-GC fEPSP amplitude in control vector-injected or GAP43(S41A)- 268 injected mice, this effect was attenuated in GAP43(S41D)-injected mice [Fig 4D, 269 Control: 76.9 ± 3.1 % of baseline, n = 8 slices; GAP43(S41A): 77.1 ± 3.0 % of baseline, 11 270 n = 8 slices; GAP43(S41D): 87.3 ± 2.5 % of baseline, n = 8 slices; F(2, 21) = 5.695; 271 Control vs GAP43(S41D) and GAP43(S41A) vs GAP43(S41D), p < 0.05 by one-way 272 ANOVA]. In contrast, WIN-55,212-2-mediated reduction of inhibitory synaptic 273 responses (i.e., extracellular field inhibitory postsynaptic potentials, or fIPSPs) recorded 274 in the IML was unaltered under any condition [Fig 4E, Control: 84.8± 2.3 % of baseline, 275 n = 5 slices; GAP43(S41A): 88.8 ± 2.3 of baseline, n = 5 slices; GAP43(S41D): 84.5 ± 276 2.6 % of baseline, n = 7 slices; F(2,14) = 0.7654; p = 0.4836 by one-way ANOVA]. Taken 277 together, these observations strongly suggest that phosphorylated GAP43 inhibits 278 CB1R function at MC-GC synapses. 279 280 GAP43 genetic deletion from mossy cells enhances CB1R synaptic function 281 To further characterize the effect of endogenous GAP43 on CB1R function at the MC- 282 GC synapse, we generated Gap43fl/fl mice (Suppl Fig 2A and 2B, steps i and ii; Suppl 283 Fig 2C) and selectively knocked-out Gap43 from MCs. Briefly, a mix of AAV5-CaMKII- 284 Cre-mCherry and AAV-DG-FLEX-ChIEF-TdTomato was injected in the hilus of 285 Gap43fl/fl mice. This manipulation allowed us to selectively and optically stimulate 286 commissural MC axons that lack GAP43 and express a fast version of 287 channelrhodopsin (ChIEF). WT mice injected with the same mix of viral vectors were 288 used as control (Fig 5A). Cre recombinase activity was confirmed by the reduction of 289 GAP43 labeling at infected MC axon boutons in IML-containing contralateral 290 hippocampal sections (Fig 5B). Whole-cell patch-clamp recordings were performed 291 from GCs 4 weeks post-injection, and EPSCs (o-EPSCs) were evoked by optically 292 stimulating MC axons in the IML of the contralateral DG. GAP43-lacking MC-GC 293 synapses displayed normal PPR (Fig 5C, WT: 1.09 ± 0.07, n = 7 cells; Gap43-/-: 1.14 ± 294 0.09, n = 9 cells; t(14) = 0.4248; p = 0.6775 by unpaired Student’s t-test), CV (Fig 5C, 12 295 WT: 0.27 ± 0.04; Gap43-/-: 0.33 ± 0.04; t(14) = 0.9890; p = 0.3395 by unpaired Student’s 296 t-test), and DSE (Fig 5D; WT: 62.0 ± 7.1 % of baseline, n = 7 cells; Gap43-/-: 57.4 ± 4.4 297 % of baseline, n = 12 cells; t(17) = 0.5926; p = 0.5612 by unpaired Student’s t-test). 298 However, WIN-55,212-2-mediated suppression of MC-GC synaptic transmission was 299 faster in GAP43-deficient compared to WT MC-GC synapses (Fig 5E, WT: 81.2 ± 9.0 300 % of baseline, n = 5 cells at time = 15 min; Gap43-/-: 40.3 ± 10.5 % of baseline, n = 5 301 cells at time = 15 min; t(8) = 2.941; p < 0.05 by unpaired Student’s t-test), supporting 302 that loss of endogenous GAP43 transiently enhances CB1R function at MC-GC 303 synapses. 304 305 Enhanced anti-convulsant response to THC in Glu-Gap43-/- mice 306 Next, we aimed to unveil the behavioral relevance of the GAP43-CB1R interaction. To 307 delete endogenous GAP43 from restricted neuronal subpopulations we generated 308 conditional knockout mouse lines in which the GAP43-encoding gene was selectively 309 inactivated in dorsal telencephalic glutamatergic neurons or forebrain GABAergic 310 neurons (hereafter Glu-Gap43-/- and GABA-Gap43-/- mice, respectively; Suppl Fig 2A 311 and 2B, step iii). Both lines (especially Glu-Gap43-/-) showed reduced levels of GAP43 312 protein in the whole hippocampus by Western blot analysis (Suppl Fig 2D). This 313 reduction was particularly evident in glutamatergic presynaptic boutons of the IML, as 314 evidenced by immunofluorescence microscopy (Suppl Fig 2D). Glu-Gap43-/- and 315 GABA-Gap43-/- mice did not exhibit any noticeable dysmorphology or alteration in 316 survival, growth, and fertility. Both lines, at 8 weeks of age, had normal overall brain 317 morphology (Suppl Fig 3A), body weight (Suppl Fig 3B), body temperature (Suppl Fig 318 3C), gait pattern (Suppl Fig 3D), pain response (Suppl Fig 3E), and anxiety-like 319 behavior (Suppl Fig 3F). In line with the previously described memory alterations 13 320 evoked by hippocampal GAP43 inactivation26,36,37, we found a deficit of long-term novel 321 object recognition memory in Glu-Gap43-/- mice that was not evident in GABA-Gap43-/- 322 animals (Suppl Fig 3G). Of note, hippocampal CB1R levels and number of MCs (as 323 identified by calretinin staining) were not affected in any of the two mouse lines (Suppl 324 Fig 3H). 325 326 Hippocampal MC-GC circuits22,38, GAP4339,40 and glutamatergic-neuron CB1Rs19 have 327 all been implicated in the control of epileptic seizures. Specifically, activation of CB1Rs 328 with THC and other cannabinoids, under specific therapeutic windows, protects mice41– 329 43 330 GAP43 complexes in this process by using a combined pharmacological and genetic 331 approach. We first tested two doses of the pro-convulsant drug kainic acid (KA; 20 or 332 30 mg/kg; 1 single i.p. injection), a well-established model of temporal lobe epilepsy46, 333 to induce excitotoxic epileptic seizures in WT mice. Administration of THC (10 mg/kg; 1 334 single i.p. injection) 15 min before KA transiently reduced the mild behavioral seizures 335 induced by 20 mg/kg KA (Suppl Fig 4A) but was unable to counteract the more severe 336 pro-epileptic phenotype induced by 30 mg/kg KA (Suppl Fig 4B). Hence, we selected 337 the dose of 30 mg/kg KA to be tested in Glu-Gap43-/- and GABA-Gap43-/- mice (Fig 338 6A), aiming to a protective response to THC in the (conceivably “CB1R-disinhibited”) 339 Glu-Gap43-/- animals. As expected, no difference in KA-induced seizures over time was 340 evident between vehicle and THC-treated control Gap43fl/fl mice (Fig 6B, left panels). In 341 contrast, THC reduced seizure progression in Glu-Gap43-/- mice, while it had no 342 protective effect in GABA-Gap43-/- animals (Fig 6B, right panels). Seizure severity was 343 accordingly reduced in THC-treated Glu-Gap43-/- mice (Fig 6C, left panel) but not in 344 THC-treated GABA-Gap43-/- animals (Fig. 6C, right panel). Collectively, these and other mammals44,45 against seizures. We therefore evaluated the role of CB1R- 14 345 observations are consistent with our neuroanatomical (Fig 3, Suppl Fig S3) and 346 electrophysiological data (Figs 4 and 5), and indicate that, upon selective GAP43 347 inactivation in glutamatergic neurons, a “disinhibited” CB1R reduces the KA-induced 348 pro-convulsant phenotype of mice. 349 15 350 DISCUSSION 351 352 Here, we discovered that GAP43, a protein known for decades for its functions in 353 axonal plasticity, is a CB1R-binding partner. Moreover, this is the first study that unveils 354 the anatomical and functional specificity of a CB1R-interactor complex at the synapse 355 level, thereby substantiating the unique identity of every CB1R subpopulation in the 356 brain. We (present study) and others27,28 have used MS proteomic-based approaches 357 to define the CB1R interactome. We were not able to detect various cytoplasmic 358 proteins that have been previously reported to interact with CB1R-CTD, for example 359 CRIP1a, GASP1, SGIP1, FAN and WAVE1 complex, likely owing to differences in the 360 experimental conditions and/or because they were minor components compared to 361 other cellular proteins in our starting brain sample. Both GAP43 and CB1R are mainly 362 sorted to presynaptic boutons and anchored in lipid rafts14,26,30,47, which conceivably 363 makes them prone to interact. Studies on other potential CB1R interactors reported to 364 date27,48–50, including the seminal studies on CRIP1a51,52, have been mostly conducted 365 with transfected cell lines in culture. Here, we added several molecular techniques to 366 provide a detailed characterization of the interaction of CB1R with WT and mutant 367 forms of GAP43 in various in vitro and in vivo systems. Phosphorylation of GAP43 by 368 protein kinase C at a unique site (S41) is known to constitute the most relevant 369 modification of GAP43 biological activity26. In the present study we provide evidence 370 supporting that the CB1R-GAP43 interaction is dependent on the phosphorylation 371 status of GAP43-S41, in line with previous studies on the interaction of GAP43 with 372 proteins as SNAP25, syntaxin, F-actin and rabaptin-553–55. Moreover, by mapping CB1R 373 and GAP43 expression, we found that both proteins reside in close proximity in the DG, 374 showing a high abundance at MC axon boutons of the IML and an absence of 16 375 expression in GCs. This distribution pattern defines a dynamic, highly restricted, 376 synapse type-selective occurrence of CB1R-GAP43 complexes in the mouse 377 hippocampus. 378 379 Previous reports had also used GAP43-S41 phospho-mimetics to investigate the 380 biological role of GAP43. Of note, compared to WT counterparts, mutant mice with a 381 generalized expression of GAP43(S41D) showed enhanced levels of Hebbian LTP as 382 induced in the perforant path (PP) of the DG in vivo36, as well as at SC-CA1 synapses, 383 together with an enhanced PPR56. In contrast, the expression of GAP43(WT), 384 GAP43(S41A) or a form of GAP43 with a deletion of the entire effector domain did not 385 affect the level of LTP36,56. On the other hand, LTD was not affected by GAP43(S41D) 386 overexpression56, and the induction of LTD in WT animals led to a reduction in GAP43 387 phosphorylation status57. Distinctively, instead of a generalized transgenic model, our 388 approach with AAV-mediated delivery ensures the specific transgene expression at 389 presynaptic boutons of MC axons. Our findings show an inhibitory effect on WIN- 390 55,212-2-mediated depression of neurotransmission and on DSE upon GAP43(S41D) 391 but not GAP43(S41A) expression. These effects of GAP43(S41D) appear remarkably 392 robust despite the coexistence of the GAP43 mutant form with the endogenous protein. 393 The observation that the phospho-resistant GAP43(S41A) construct behaves just like a 394 control empty vector may indicate that, in vivo, the CB1R-GAP43 interaction does not 395 stand as a constant interaction but rather occurs under specific biological triggers (e.g., 396 those relying on high-activity regimes associated to protein kinase C activation). 397 Therefore, the CB1R-GAP43 interaction would most likely reflect a transient functional 398 state. The high CB1R expression at MC axon boutons supports a powerful negative 399 control of MC-GC synaptic transmission and LTP25,35. Remarkably, LTP at PP-GC 17 400 synapses is associated with increased GAP43 mRNA expression in MCs58. Thus, 401 activity-dependent upregulation of GAP43 phosphorylation and levels, and the resultant 402 inhibition of CB1R signaling, could be a mechanism by which information transfer is 403 fine-tuned in the DG, possibly contributing to DG-dependent learning. To complement 404 the GAP43-overexpression approach, we combined Cre-mediated promotor-driven 405 recombination and optogenetics to allow a precise spatiotemporal modulation of MC 406 axon boutons lacking GAP43. We found that the selective deletion of endogenous 407 GAP43 in MCs significantly delayed the onset of the CB1R-mediated inhibition of MC- 408 GC synaptic transmission, but had no significant effect on DSE. While this mild 409 synaptic phenotype could result from compensatory changes (e.g., by other CB1R- 410 modulating proteins) and/or incomplete GAP43 deletion (e.g., due to deficient Cre 411 recombination), the reduction in PPR and CV upon GAP43(S41D) expression, which 412 likely reflects an increase in glutamate release25, strongly suggests that a gain of 413 GAP43 function hinders CB1R signaling at the MC-GC synapse. 414 415 GAP43 has been previously implicated in epileptogenesis40. GAP43 expression levels 416 increase at MC axon boutons of the IML upon seizure induction prior to mossy fiber 417 sprouting39,59,60. Moreover, MCs regulate GC activity directly through innervation or 418 indirectly through modulation of GABAergic interneurons21,22. MC-evoked excitation of 419 GCs is normally weak, but becomes dramatically strengthened under pathological 420 conditions such as pro-convulsant insults22. On the other hand, CB1Rs can dampen 421 overactive neural circuits. For example, glutamatergic-neuron CB1R in the DG is crucial 422 for attenuating the KA-induced epileptic phenotype in mice, presumably by stabilizing 423 the recurrent GC-MC-GC circuitry19. A specific down-regulation of CB1R protein and 424 mRNA on glutamatergic but not GABAergic axon terminals has been reported in 18 425 epileptic human hippocampal tissue61. In addition, somatic transfer of the CB1R- 426 encoding gene to hippocampal glutamatergic neurons was sufficient to protect mice 427 against acute seizures and neuronal damage62. CB1R agonists can act as anti- 428 convulsants in various animal models of hyperexcitability and epilepsy, presumably by 429 decreasing glutamatergic transmission42,43,63,64. Here, we used a dose of THC lower 430 than the ED50 values of previous studies, which had reported a mild anti-convulsant 431 activity of the drug in various seizure models41–43. Remarkably, only by deleting GAP43 432 selectively from glutamatergic neurons, including MCs, we were able to achieve an 433 unambiguous THC-mediated anti-convulsant effect. It has been reported that THC may 434 also induce pro-convulsant actions at high doses, likely via CB1R-dependent inhibition 435 of GABAergic neurons15,65, thus highlighting the critical importance of the neuron 436 population-specificity of CB1R action. Understanding this specificity of multimodal 437 cannabinoid signaling may be crucial to gain further insight into the unwanted effects of 438 cannabis abuse, and to design personalized interventions aimed to enhance or 439 depress CB1R activity in selective pathological situations. 440 19 441 METHODS 442 443 Animals 444 Experimental procedures used were performed in accordance with the guidelines and 445 approval of the Animal Welfare Committees of Universidad Complutense de Madrid 446 and Comunidad de Madrid, the directives of the Spanish Government and the 447 European Commission, as well as the guidelines of NIH and Albert Einstein College of 448 Medicine Institutional Animal Care and Use Committee. Animal housing, handling, and 449 assignment to the different experimental groups was conducted as described 450 previously9. Adequate measures were taken to minimize pain and discomfort of the 451 animals. Both males and females were used in this study. We used C57Bl/6J mice 452 (Charles River), Cnr1floxed/floxed (Cnr1fl/fl) mice, Cnr1floxed/floxed;CMV-Cre (Cnr1-/-) mice, 453 conditional Cnr1floxed/floxed;Nex1-Cre (Glu-Cnr1-/-) mice, and conditional Cnr1floxed/floxed;Dlx5/6-Cre 454 (GABA-Cnr1-/-) mice19,66,67; as well as Stop-Cnr1 mice, Stop-Cnr1EIIa-Cre (Cnr1-RS) mice, 455 conditional Stop-Cnr1Nex1-Cre (Glu-Cnr1-RS) mice, and conditional Stop-Cnr1Dlx5/6-Cre 456 (GABA-Cnr1-RS) mice, to allow CB1R gene-expression rescue from a CB1R-null 457 background33,34. As for GAP43 animal models, we purchased B6Dnk;B6Brd;B6N-Tyrc- 458 Brd Gap43tm1a(EUCOMM)Wtsi/WtsiBiat mice from the EMMA Mouse Repository. This mouse 459 line contains a LacZ cassette and a Neo cassette with a Stop codon flanked by Frt 460 sites, followed by the Gap43 exon 2 flanked by LoxP sites. These mice were crossed 461 with mice carrying a constitutive expression of Flp recombinase under the control of the 462 constitutive promoter ACTB (The Jackson Laboratory; kindly provided by Dr. Rui 463 Benedito, CNIC, Madrid, Spain), thus allowing a rescued, “conditional-ready” floxed 464 allele (Gap43fl/fl). Subsequent crossing with Nex1-Cre or Dlx5/6-Cre-expressing mice19 465 yielded the corresponding conditional knockout mouse lines (Glu-Gap43-/- and GABA20 466 Gap43-/-, respectively). All the generated GAP43 conditional knockout mice were 467 backcrossed for at least 8 generations before use. 468 469 Affinity-based proteomics 470 A whole sheep brain was homogenized by mechanical disaggregation in RIPA buffer 471 (50 mM Tris-HCl, 150 mM NaCl, 1 % v/v Triton X-100, 0.1 % w/v deoxycholic acid, 0.1 472 % w/v SDS, pH 7.35). The soluble fraction of the homogenate was loaded onto a 473 Sepharose 4B column after extensively washing the column with 100 mM Tris-HCl, 200 474 mM NaCl, pH 7.0. The eluted soluble fraction was then collected and loaded onto the 475 Sepharose 4B column saturated with purified hCB1R-CTD bound to lectin. Purified 476 lectin-empty plasmid was expressed in an additional Sepharose 4B column as a 477 control. After washing with RIPA and TBS buffer (50 mM Tris-HCl, 150 mM NaCl, pH 478 7.0), the bound fraction was eluted with 200 mM lactose. Proteins were then subjected 479 to nLC/MS-MS proteomics analysis. Briefly, the samples were loaded on a 12 % 480 acrylamide gel and a denaturing electrophoresis was carried out. Once samples had 481 reached the resolving gel, electrophoresis was stopped, and the gel was died with 482 Coomassie Colloidal Blue overnight. After fading, the region of the gel containing the 483 sample was cut just above the recombinant protein, this piece being divided into 484 smaller fractions that were thereafter digested with trypsin. The resulting peptide 485 fragments were retained in an Acclaim PepMap 100 precolumn (Thermo Scientific, 486 Waltham, MA, USA) and then eluted in an Acclaim PepMap 100 C18 column, 25-cm 487 long, 75 µm-internal diameter, and 3-µm particle size (Thermo Scientific, Waltham, MA, 488 USA). The peptides were separated in a gradient for 110 min (90 min 0 – 35 % of 489 Buffer B; 10 min 45 – 95 % Buffer B; 9 min 95 % Buffer B; and 1 minute 10 % Buffer B; 490 Buffer B supplemented with 0.1 % formic acid in acetonitrile) at 250 nL/min in a 21 491 nanoEasy nLC 1000 (Proxeon) coupled to an ionic source with nanoelectrospray 492 (Thermo Scientific, Waltham, MA, USA) for electrospray ionization. Mass spectra were 493 acquired in an LTQ–Orbitrap Velos mass spectrometer (Thermo Scientific, Waltham, 494 MA, USA) working in positive mode, which measures the mass-to-charge ratio (m/z) of 495 ionized particles and detects the relative number of ions at each m/z ratio. Mass 496 spectra corresponding to a full screening (m/z 400 to 2000) were obtained with a 497 resolution of 500,000 to 60,000 (m/z = 200), and the 15 most intense ions from each 498 screening were selected for fragmentation by cleavage at peptide bonds via collision 499 with a gaseous matrix by dissociation induced by collision (CID) with the energy of 500 collision normalized to 35%. Ions with unique charge or no charge were discarded. A 501 dynamic exclusion of 45-s duration was conducted. The masses of the fragments were 502 then determined by ion trap to define the amino acid sequence of the peptides. 503 Spectrum files were challenged to data bases and Uniprot of sheep (Ovis aries) (23112 504 sequences) and other mammals (Mammalian) (1184488 sequences) by using the 505 software Proteome Discoverer (version 1.4.1.14, Thermo Scientific, Waltham, MA, 506 USA) and the searching tool Sequest. In the searching criteria, carbamide methylation, 507 cysteine nitrosylation and methionine oxidation were established as dynamic 508 modifications. Tolerance to precursor selection and product ions was fixed at 10 ppm 509 and 0.5 Da, respectively. Peptide identification was validated by the Percolator 510 algorithm using q ≤ 0,01 (q-value being the p-value additionally adjusted to the False 511 Discovery Rate). 512 513 Acute hippocampal slice preparation 514 Viral vectors were stereotactically injected into the hilus of 3–4-week-old C57Bl/6J 515 mice. Briefly, mice were anesthetized with isoflurane (up to 5 % for induction and 1–2 22 516 % for maintenance) and either AAV1/2-CBA-CFP, AAV1/2-CBA-GAP43(S41A)-CFP or 517 AAV1/2-CBA-GAP43(S41D)-CFP were injected into the hilus (1 µL at a flow rate of 0.1 518 µL/min) using the following coordinates (mm to bregma): antero-posterior +2.18, lateral 519 ±1.5, dorso-ventral -2.2 from dura. Coordinates were adjusted according to the bregma 520 to lambda distance for each mouse. At 3 weeks post-injection, acute hippocampal 521 slices were prepared. Animals were anesthetized with isoflurane, and brains were 522 removed and rapidly transferred into ice-cold cutting solution containing (in mM): 110 523 choline chloride, 25 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 0.5 CaCl2, 7 MgCl2, 25 D- 524 glucose, 11.6 sodium L-ascorbate, and 3.1 sodium pyruvate. Hippocampi were isolated 525 and sliced (300 µm-thick) using a VT1200S microslicer (Leica Microsystems Co.). 526 Slices were then transferred and incubated for 30 min in a chamber placed at 33–34°C 527 and with artificial cerebrospinal fluid (ACSF) solution containing (in mM): 124 NaCl, 26 528 NaHCO3, 10 D-glucose, 2.5 KCl, 1 NaH2PO4, 2.5 CaCl2, and 1.3 MgSO4. Slices were 529 kept at room temperature for at least 45 min prior to experiments. All solutions were 530 maintained at 95 % O2/5 % CO2 (pH 7.4). 531 532 Electrophysiology 533 All recordings were performed at 28 ± 1 °C in a submersion-type recording chamber 534 535 perfused at ∼2 mL/min with ACSF supplemented with GABAA receptor antagonist 536 the presence of D-APV (50 μM) and NBQX (10 μM). Whole-cell patch-clamp 537 recordings were made from GCs voltage clamped at –60 mV using a patch-type pipette 538 electrode (∼3–4 MΩ) containing (in mM): 131 Cs-gluconate, 8 NaCl, 1 CaCl2, 10 539 EGTA, 10 D-glucose, and 10 HEPES, pH 7.2 (285–290 mOsm). Series resistance 540 (∼8–28 MΩ) was monitored throughout all experiments with a –5 mV, 80 ms voltage picrotoxin (100 µM), except for inhibitory synaptic transmission which was monitored in 23 541 step, and cells that exhibited a change in series resistance (> 20 %) were excluded 542 from analysis. Extracellular field recordings (fEPSPs and fIPSCs) were performed 543 using a patch-type pipette filled with 1 M NaCl and placed in the IML. All experiments 544 were performed in an interleaved fashion –i.e., control experiments were performed 545 every test experiment on the same day. 546 547 The stimulating patch-type pipette was filled with ACSF and placed in the IML (< 50 µm 548 from the border of the GC body layer) to activate MC axons. To elicit synaptic 549 responses, paired, monopolar square-wave voltage pulses (100–200 µs pulse width, 550 4–25 V) were delivered through a stimulus isolator (Isoflex, AMPI) connected to a 551 broken tip (10–20 µm) stimulating patch-type pipette filled with ACSF. Stimulus 552 intensity was adjusted to get comparable magnitude synaptic responses across 553 experiments (e.g., 50–100 pA EPSCs at Vh = –60 mV or 5 mV fEPSPs). Stimulation 554 was achieved by delivering paired pulses 100 ms apart. PPR was defined as the ratio 555 of the amplitude of the second EPSC, to the amplitude of the first EPSC. CV was 556 calculated as the standard deviation of EPSC amplitude divided by the mean of EPSC 557 amplitude. Both PPR and CV were measured during 10-min baseline. In drug-delivery 558 experiments, stimulation was triggered every 20 s, and every 10 s in DSE experiments. 559 DSE was induced with 5-s depolarizing voltage step (from –60 mV to 0 mV) to trigger 560 endocannabinoid release from GC. The magnitude of DSE was determined as the 561 percentage change between the mean amplitude of 3 consecutive EPSCs preceding 562 DSE and the mean amplitude of 3 consecutive EPSCs following depolarization. For 563 WIN-55,212-2-mediated depression, the magnitude of depression was calculated by 564 comparing the 10-min baseline responses with the 10-min responses at the end of 24 565 WIN-55,212-2 application (specifically, from 5 min before washout to 5 min after 566 washout). Representative traces were obtained by averaging 15 individual traces. 567 568 Electrophysiological data were acquired at 5 kHz, filtered at 2.4 kHz, and analyzed 569 using custom-made software for IgorPro 7.01 (Wavemetrics, Inc.). Picrotoxin, WIN- 570 55,212-2, AM251, D-APV and NBQX were purchased from Tocris-Cookson Inc. 571 (Ellisville, MO, USA) and dissolved in water or DMSO. Total DMSO in the bath solution 572 was maintained at 0.1 % in all experiments. 573 574 Optogenetics 575 Gap43fl/fl mice were anesthetized with isoflurane (up to 5% for induction and 1%–2% 576 for maintenance). A mix of AAV5-CaMKII-Cre-mCherry and AAV-DG-FLEX-ChIEF- 577 TdTomato viruses (ratio 1:2) was stereotactically injected into the hilus, using the same 578 coordinates as mentioned above. Acute hippocampal slices were prepared as 579 described above 4 weeks after viral injection, and whole-cell patch clamp recordings 580 were performed in the contralateral DG. Pulses of blue light (0.5–2 ms) were provided 581 by using a 473-nm LED (Thorlabs, Inc.) through the microscope objective (40x, 0.8 582 NA), and centered in the IML. 583 584 Behavioral tests 585 Adult (ca. 3-month-old) Glu-Gap43−/− and GABA-Gap43−/− mice, as well as their 586 respective Gap43fl/fl control littermates, were used for behavioral tests. Animals were 587 assigned randomly to the different treatment groups, habituated to the experimental 588 room, and handled one week before testing. All the behavioral tests were video- 589 recorded for subsequent blind analysis by a different trained observer, using Smart3.0 25 590 Software (Panlab, Barcelona, Spain). Body weight and temperature was measured, the 591 latter with a thermo-coupled flexible probe (Panlab, Madrid, Spain) located in the 592 rectum for 10 s. Analgesia was evaluated by the hot-plate paradigm. The test consisted 593 of placing a mouse on an enclosed hot plate (Columbus Instruments, Columbus, OH, 594 USA) and measuring the latency to lick one of the paws. Walking patterns were 595 monitored with ink-painted paws (blue, fore; red, hind) on 70-cm long paper sheets. 596 Anxiety-like behaviors were assessed in the elevated plus-maze test. The maze 597 consisted of a cross-shaped plastic device with two 30 cm-long, 5 cm-wide opposite 598 open arms, and two 30 cm-long, 5 cm-wide, 16 cm-high opposite closed arms, 599 connected by a central structure (5x5 cm) and elevated 50 cm from the floor. First, a 5- 600 min observation session was performed, in which each mouse was placed in the 601 central neutral zone facing one of the open arms. The cumulative time spent in the 602 open and closed arms was then recorded. One arm entry was considered when the 603 animal had placed at least both forelimbs in the arm. Data are represented as the 604 number of visits to the open arms. For the novel-object recognition task, mice were 605 habituated for 9 min in an L-maze68. Twenty-four hours later, mice were first exposed to 606 two identical objects located at the edges of the maze arms, and allowed exploration 607 for 9 min during the training session. After an inter-trial interval of 24 h, mice were 608 placed back again for a 9-min test session in which one of the familiar objects had 609 been replaced by a novel object. Object exploration was defined as the mouse 610 directing its nose to the object (> 2 cm) and being involved in active exploration. Mice 611 did not show preference for any object before trials. A discrimination index (Id) was 612 calculated to measure recognition memory as (tnew object - t(familiar object)) / (tnew object + t(familiar 613 object)), 614 times below 15 s were excluded from the analyses. being the denominator the total exploration time. Mice showing total exploration 26 615 For induction of acute excitotoxic seizures, KA (Sigma) was dissolved in isotonic saline, 616 pH 7.4, and injected i.p. Seizures were induced by injecting 20 mg/kg or 30 mg/kg of 617 KA in a volume of 10 mL/kg body weight19,67 where indicated, and i.p. injection with 618 vehicle (1 % v/v DMSO in 1:18 v/v Tween-80 / saline solution) or 10 mg/kg THC (THC 619 Pharm) was performed 15 min prior to KA injection. Right after, mice were placed in 620 clear plastic cages and constantly recorded and monitored for 2 h for seizure behavior. 621 They were scored every 5 min according to the following modified Racine scale69: 622 immobility (stage 1); forelimb and/or tail extension, rigid posture (stage 2); repetitive 623 movements and head bobbing (stage 3); rearing and falling (stage 4); continuous 624 rearing and falling, jumping, and/or wild running (stage 5); generalized tonic-clonic 625 seizures (stage 6); and death (stage 7)19. Seizure severity was determined as 626 previously described70 by the formula Seizure severity = Σ(all scores of a given mouse) 627 / time of experiment. 628 629 Other methods 630 The rest of the experimental procedures used in this study are extensively described in 631 Supplementary Methods. That section provides precise details on gene constructs, 632 protein expression and purification, fluorescence polarization, cell culture and 633 transfection, Western blotting and co-immunoprecipitation, synaptosomes preparation, 634 immunofluorescence, PLA, BRET, DMR assays, and recombinant AAV production. 635 636 Statistics 637 Data are presented as mean ± SEM, and the number of experiments is indicated in 638 every case. Statistical analysis was performed with GraphPad Prism v8.0.1 (GraphPad 639 Software, La Jolla, CA, USA). All variables were first tested for normality (Kolmogorov– 27 640 Smirnov’s and Shapiro-Wilk´s test) and homocedasticity (Levene´s test) before 641 analysis. When variables satisfied these conditions, one-way or two-way ANOVA with 642 Tukey`s or Sidak`s post hoc test, or paired or unpaired Student’s t test, was used as 643 appropriate and indicated in every case. We considered p-values < 0.05 as statistically 644 significant. Data acquisition and analysis was conducted in a blinded manner respect to 645 the experimental conditions. 646 28 647 ACKNOWLEDGMENTS 648 This work was supported by the Spanish Ministerio de Ciencia e Innovación 649 (MICINN/FEDER; grants RTI2018-095311-B-I00 and PID2021-125118OB-I00 to M.G., 650 SAF-2017-87629-R to E.I.C. and V.C., PID2020-113938RB-I00 to E.M. and V.C., BFU 651 2017-83292-R to J.S.-P., and PID2020-119358GB-I00 to D.F. de S.) and the NIH 652 (grants R01 MH116673, R01 MH125772, R01 NS115543, and R01 NS113600 to 653 P.E.C.). L.B. and G.M. were supported by INSERM. I.B.M. and C.C.-I. were supported 654 by contracts from the Spanish Ministerio de Universidades (Formación de Profesorado 655 Universitario Program, references FPU15/01833 and FPU16/02593, respectively). We 656 are indebted to Rodrigo Barderas-Machado, Alba Hermoso-López and Lucía Rivera- 657 Endrinal for expert technical assistance. 658 659 AUTHOR CONTRIBUTIONS 660 I.B.M., C.C.-I., C.B., E.M., G.M., L.B., E.I.C., V.C., I.G.-R., D.F. de S., I.R.-C., P.E.C., 661 and M.G. designed research. I.B.M., C.C.-I., C.B., E.M., A.R.-C., C.M.-F., A.M.-C., 662 R.M., and N.G.-F. performed research. I.B.M., C.C.-I., C.B., E.M., A.R.-C., R.M., J.S.- 663 P., D.F. de S., I.R.-C., P.E.C., and M.G. analyzed data. I.B.M. and M.G. drafted the 664 paper. I.B.M., C.C.-I., C.B., P.E.C., and M.G. wrote the final version of the paper. All 665 the authors revised the final version of the paper. As senior members of each 666 collaboration group on the study, J.S.-P., L.B., V.C., D.F. de S., P.E.C., and M.G. take 667 responsibility for their respective group's contribution, including preservation of the 668 original data on which the paper is based; verification that the figures and conclusions 669 accurately reflect the data collected, and that manipulations to images are in 670 accordance with Nature journal guidelines; and minimization of obstacles to sharing 671 materials, data, and algorithms through appropriate planning. 29 672 COMPETING INTERESTS 673 The authors declare no competing interests. 674 30 675 REFERENCES 676 1. Neurosci. 4, 873–84 (2003). 677 678 Piomelli, D. The molecular logic of endocannabinoid signalling. Nat. Rev. 2. Pertwee, R. G. et al. International Union of Basic and Clinical Pharmacology. 679 LXXIX. Cannabinoid receptors and their ligands: Beyond CB1 and CB2. 680 Pharmacol. Rev. 62, 588–631 (2010). 681 3. Kano, M., Ohno-Shosaku, T., Hashimotodani, Y., Uchigashima, M. & Watanabe, 682 M. Endocannabinoid-mediated control of synaptic transmission. Physiol. Rev. 683 89, 309–380 (2009). 684 4. Endocannabinoid signaling and synaptic function. Neuron 76, 70–81 (2012). 685 686 5. Mechoulam, R. & Parker, L. A. The endocannabinoid system and the brain. Annu. Rev. Psychol. 64, 21–47 (2013). 687 688 Castillo, P. E., Younts, T. J., Chávez, A. E. & Hashimotodani, Y. 6. Lutz, B., Marsicano, G., Maldonado, R. & Hillard, C. J. The endocannabinoid 689 system in guarding against fear, anxiety and stress. Nat. Rev. Neurosci. 16, 690 705–718 (2015). 691 7. Busquets-Garcia, A., Bains, J. & Marsicano, G. CB1 receptor signaling in the 692 brain: extracting specificity from ubiquity. Neuropsychopharmacology 43, 4–20 693 (2018). 694 8. Guggenhuber, S. et al. Cannabinoid receptor-interacting protein Crip1a 695 modulates CB1 receptor signaling in mouse hippocampus. Brain Struct. Funct. 696 221, 2061–2074 (2016). 697 9. Costas-Insua, C. et al. Identification of BiP as a CB1 receptor-interacting protein 698 that fine-tunes cannabinoid signaling in the mouse brain. J. Neurosci. 41, 7924– 699 7941 (2021). 31 700 10. Viñals, X. et al. Cognitive impairment induced by delta-9-tetrahydrocannabinol 701 occurs through heteromers between cannabinoid CB1 and serotonin 5-HT2A 702 receptors. PLoS Biol. 13, e1002194 (2015). 703 11. Moreno, E. et al. Singular location and signaling profile of adenosine A2A- 704 cannabinoid CB1 receptor heteromers in the dorsal striatum. 705 Neuropsychopharmacology 43, 964–977 (2018). 706 12. Domenici, M. R. et al. Cannabinoid receptor type 1 located on presynaptic 707 terminals of principal neurons in the forebrain controls glutamatergic synaptic 708 transmission. J. Neurosci. 26, 5794–5799 (2006). 709 13. glutamatergic synapses. J. Neurosci. 26, 5628–37 (2006). 710 711 14. 15. 16. Bellocchio, L. et al. Bimodal control of stimulated food intake by the endocannabinoid system. Nat. Neurosci. 13, 281–283 (2010). 716 717 Soltesz, I. et al. Weeding out bad waves: towards selective cannabinoid circuit control in epilepsy. Nat. Rev. Neurosci. 16, 264–277 (2015). 714 715 Katona, I. & Freund, T. F. Endocannabinoid signaling as a synaptic circuit breaker in neurological disease. Nat. Med. 14, 923–930 (2008). 712 713 Katona, I. et al. Molecular composition of the endocannabinoid system at 17. Rey, A. A., Purrio, M., Viveros, M.-P. & Lutz, B. Biphasic effects of cannabinoids 718 in anxiety responses: CB1 and GABAB receptors in the balance of GABAergic 719 and glutamatergic neurotransmission. Neuropsychopharmacology 37, 2624– 720 2634 (2012). 721 18. neuroprotective activity. Proc. Natl. Acad. Sci. USA 111, 8257–8262 (2014). 722 723 724 Chiarlone, A. et al. A restricted population of CB1 cannabinoid receptors with 19. Monory, K. et al. The endocannabinoid system controls key epileptogenic circuits in the hippocampus. Neuron 51, 455–466 (2006). 32 725 20. Kawamura, Y. The CB1 cannabinoid receptor is the major cannabinoid receptor 726 at excitatory presynaptic sites in the hippocampus and cerebellum. J. Neurosci. 727 26, 2991–3001 (2006). 728 21. Neurosci. 17, 562–575 (2016). 729 730 22. Botterill, J. J. et al. An excitatory and epileptogenic effect of dentate gyrus mossy cells in a mouse model of epilepsy. Cell Rep. 29, 2875-2889.e6 (2019). 731 732 Scharfman, H. E. The enigmatic mossy cell of the dentate gyrus. Nat. Rev. 23. Wang, W. et al. Atypical endocannabinoid signaling initiates a new form of 733 memory-related plasticity at a cortical input to hippocampus. Cereb. Cortex 28, 734 2253–2266 (2018). 735 24. Chiu, C. Q. & Castillo, P. E. Input-specific plasticity at excitatory synapses 736 mediated by endocannabinoids in the dentate gyrus. Neuropharmacology 54, 737 68–78 (2008). 738 25. Jensen, K. R., Berthoux, C., Nasrallah, K. & Castillo, P. E. Multiple cannabinoid 739 signaling cascades powerfully suppress recurrent excitation in the hippocampus. 740 Proc. Natl. Acad. Sci. USA 118, e2017590118 (2021). 741 26. Holahan, M. R. A shift from a pivotal to supporting role for the growth-associated 742 protein (GAP-43) in the coordination of axonal structural and functional plasticity. 743 Front. Cell. Neurosci. 11, 1–19 (2017). 744 27. Njoo, C., Agarwal, N., Lutz, B. & Kuner, R. The cannabinoid receptor CB1 745 interacts with the WAVE1 complex and plays a role in actin dynamics and 746 structural plasticity in neurons. PLoS Biol. 13, 1–36 (2015). 747 28. Mattheus, T., Kukla, K., Zimmermann, T., Tenzer, S. & Lutz, B. Cell type-specific 748 tandem affinity purification of the mouse hippocampal CB1 receptor-associated 749 proteome. J. Proteome Res. 15, 3585–3601 (2016). 33 750 29. Leu, B., Koch, E. & Schmidt, J. T. GAP43 phosphorylation is critical for growth 751 and branching of retinotectal arbors in zebrafish. Dev. Neurobiol. 70, 897–911 752 (2010). 753 30. Gauthier-Kemper, A. et al. Interplay between phosphorylation and palmitoylation 754 mediates plasma membrane targeting and sorting of GAP43. Mol. Biol. Cell 25, 755 3284–3299 (2014). 756 31. Benowitz, L. I., Apostolides, P. J., Perrone-Bizzozero, N., Finklestein, S. P. & 757 Zwiers, H. Anatomical distribution of the growth-associated protein GAP-43/B-50 758 in the adult rat brain. J. Neurosci. 8, 339–352 (1988). 759 32. Uchigashima, M. et al. Molecular and morphological configuration for 2- 760 arachidonoylglycerol-mediated retrograde signaling at mossy cell-granule cell 761 synapses in the dentate gyrus. J. Neurosci. 31, 7700–7714 (2011). 762 33. Ruehle, S. et al. Cannabinoid CB1 receptor in dorsal telencephalic glutamatergic 763 neurons: distinctive sufficiency for hippocampus-dependent and amygdala- 764 dependent synaptic and behavioral functions. J. Neurosci. 33, 10264–10277 765 (2013). 766 34. Remmers, F. et al. Addressing sufficiency of the CB1 receptor for 767 endocannabinoid-mediated functions through conditional genetic rescue in 768 forebrain GABAergic neurons. Brain Struct. Funct. 222, 3431–3452 (2017). 769 35. Hashimotodani, Y. et al. LTP at hilar mossy cell-dentate granule cell synapses 770 modulates dentate gyrus output by increasing excitation/inhibition balance. 771 Neuron 95, 928-943.e3 (2017). 772 36. Routtenberg, A., Cantallops, I., Zaffuto, S., Serrano, P. & Namgung, U. 773 Enhanced learning after genetic overexpression of a brain growth protein. Proc. 774 Natl. Acad. Sci. USA 97, 7657–7662 (2000). 34 775 37. impaired in heterozygous GAP-43 knockout mice. Hippocampus 15, 1–7 (2005). 776 777 38. Bui, A. D. et al. Dentate gyrus mossy cells control spontaneous convulsive seizures and spatial memory. Science 359, 787–790 (2018). 778 779 Rekart, J. L., Meiri, K. & Routtenberg, A. Hippocampal-dependent memory is 39. McNamara, R. K. & Routtenberg, A. NMDA receptor blockade prevents kainate 780 induction of protein F1/GAP-43 mRNA in hippocampal granule cells and 781 subsequent mossy fiber sprouting in the rat. Mol. Brain Res. 33, 22–28 (1995). 782 40. Nemes, A. D. et al. Growth associated protein 43 (GAP-43) as a novel target for 783 the diagnosis, treatment and prevention of epileptogenesis. Sci. Rep. 7, 1–13 784 (2017). 785 41. Karler, R. & Turkanis, S. A. Subacute cannabinoid treatment: anticonvulsant 786 activity and withdrawal excitability in mice. Br. J. Pharmacol. 68, 479–484 787 (1980). 788 42. Wallace, M. J., Wiley, J. L., Martin, B. R. & DeLorenzo, R. J. Assessment of the 789 role of CB1 receptors in cannabinoid anticonvulsant effects. Eur. J. Pharmacol. 790 428, 51–57 (2001). 791 43. Wallace, M. J., Blair, R. E., Falenski, K. W., Martin, B. R. & DeLorenzo, R. J. The 792 endogenous cannabinoid system regulates seizure frequency and duration in a 793 model of temporal lobe epilepsy. J. Pharmacol. Exp. Ther. 307, 129–137 (2003). 794 44. Wada, J. A., Wake, A., Sato, M. & Corcoran, M. E. Antiepileptic and prophylactic 795 effects of tetrahydrocannabinols in amygdaloid kindled cats. Epilepsia 16, 503– 796 510 (1975). 797 45. Colasanti, B. K., Lindamood, C. & Craig, C. R. Effects of marihuana 798 cannabinoids on seizure activity in cobalt-epileptic rats. Pharmacol. Biochem. 799 Behav. 16, 573–578 (1982). 35 800 46. Neurosci. Biobehav. Rev. 37, 2887–2899 (2013). 801 802 Lévesque, M. & Avoli, M. The kainic acid model of temporal lobe epilepsy. 47. Fletcher-Jones, A. et al. The C-terminal helix 9 motif in rat cannabinoid receptor 803 type 1 regulates axonal trafficking and surface expression. eLife 8, e44252 804 (2019). 805 48. Sánchez, C. et al. The CB1 cannabinoid receptor of astrocytes is coupled to 806 sphingomyelin hydrolysis through the adaptor protein Fan. Mol. Pharmacol. 59, 807 955–959 (2001). 808 49. Martini, L., Thompson, D., Kharazia, V. & Whistler, J. L. Differential regulation of 809 behavioral tolerance to WIN55,212-2 by GASP1. Neuropsychopharmacology 35, 810 1363–1373 (2010). 811 50. Hájková, A. et al. SGIP1 alters internalization and modulates signaling of 812 activated cannabinoid receptor 1 in a biased manner. Neuropharmacology 107, 813 201–214 (2016). 814 51. Niehaus, J. L. et al. CB1 cannabinoid receptor activity is modulated by the 815 cannabinoid receptor interacting protein CRIP1a. Mol. Pharmacol. 72, 1557– 816 1566 (2007). 817 52. Booth, W. T., Walker, N. B., Lowther, W. T. & Howlett, A. C. Cannabinoid 818 receptor interacting protein 1a (CRIP1a): function and structure. Molecules 24, 819 3672 (2019). 820 53. Haruta, T., Takami, N., Ohmura, M., Misumi, Y. & Ikehara, Y. Ca2+-dependent 821 interaction of the growth-associated protein GAP-43 with the synaptic core 822 complex. Biochem. J. 325, 455–463 (1997). 823 824 54. He, Q., Dent, E. W. & Meiri, K. F. Modulation of actin filament behavior by GAP43 (neuromodulin) is dependent on the phosphorylation status of serine 41, the 36 protein kinase C Site. J. Neurosci. 17, 3515–3524 (1997). 825 826 55. rabaptin-5 and participates in endocytosis. J. Neurosci. 18, 7757–7767 (1998). 827 828 Neve, R. L. et al. The neuronal growth-associated protein GAP-43 interacts with 56. Hulo, S., Alberi, S., Laux, T., Muller, D. & Caroni, P. A point mutant of GAP-43 829 induces enhanced short-term and long-term hippocampal plasticity. Eur. J. 830 Neurosci. 15, 1976–1982 (2002). 831 57. Ramakers, G. M. J., McNamara, R. K., Lenox, R. H. & De Graan, P. N. E. 832 Differential changes in the phosphorylation of the protein kinase C substrates 833 myristoylated alanine-rich C kinase substrate and growth-associated protein- 834 43/B-50 following Schaffer collateral long-term potentiation and long-term 835 depression. J. Neurochem. 73, 2175–83 (1999). 836 58. Namgung, U., Matsuyama, S. & Routtenberg, A. Long-term potentiation 837 activates the GAP-43 promoter: Selective participation of hippocampal mossy 838 cells. Proc. Natl. Acad. Sci. USA 94, 11675–11680 (1997). 839 59. Naffah-Mazzacoratti, M. G., Funke, M. G., Sanabria, E. R. G. & Cavalheiro, E. A. 840 Growth-associated phosphoprotein expression is increased in the supragranular 841 regions of the dentate gyrus following pilocarpine-induced seizures in rats. 842 Neuroscience 91, 485–492 (1999). 843 60. Meberg, P. J., Gall, C. M. & Routtenberg, A. Induction of F1/GAP-43 gene 844 expression in hippocampal granule cells after seizures. Brain Res. Mol. Brain 845 Res. 17, 295–9 (1993). 846 61. Ludanyi, A. et al. Downregulation of the CB1 cannabinoid receptor and related 847 molecular elements of the endocannabinoid system in epileptic human 848 hippocampus. J. Neurosci. 28, 2976–2990 (2008). 849 62. Guggenhuber, S., Monory, K., Lutz, B. & Klugmann, M. AAV vector-mediated 37 850 overexpression of CB1 cannabinoid receptor in pyramidal neurons of the 851 hippocampus protects against seizure-induced excitoxicity. PLoS One 5, e15707 852 (2010). 853 63. Blair, R. E. et al. Activation of the cannabinoid type-1 receptor mediates the 854 anticonvulsant properties of cannabinoids in the hippocampal neuronal culture 855 models of acquired epilepsy and status epilepticus. J. Pharmacol. Exp. Ther. 856 317, 1072–1078 (2006). 857 64. Shafaroodi, H. et al. The interaction of cannabinoids and opioids on 858 pentylenetetrazole-induced seizure threshold in mice. Neuropharmacology 47, 859 390–400 (2004). 860 65. Lutz, B. On-demand activation of the endocannabinoid system in the control of 861 neuronal excitability and epileptiform seizures. Biochem. Pharmacol. 68, 1691– 862 1698 (2004). 863 66. aversive memories. Nature 418, 530–534 (2002). 864 865 67. 68. 69. 872 Racine, R. J. Modification of seizure activity by electrical stimulation: II. Motor seizure. Electroencephalogr. Clin. Neurophysiol. 32, 281–294 (1972). 870 871 Oliveira da Cruz, J. F. et al. An alternative maze to assess novel object recognition in mice. Bio Protoc 10, e3651 (2020). 868 869 Marsicano, G. et al. CB1 cannabinoid receptors and on-demand defense against excitotoxicity. Science 302, 84–88 (2003). 866 867 Marsicano, G. et al. The endogenous cannabinoid system controls extinction of 70. Armas-Capote, N. et al. SGK1.1 reduces kainic acid-induced seizure severity and leads to rapid termination of seizures. Cereb. Cortex 30, 3184–3197 (2020). 873 38 A LectinhCB1RCTD Seph4B column Tryptic digestion Lactose elution Whole-brain homogenate B C Fluorescence polarization (mFP units) Intensity nLC/MS-MS spectrum Mass/charge (M/z) Mouse hippocampal neurons WT 80 Cnr1-/- GAP43 IgG CB1R 50 60 40 WCL (0.02%) GAP43-IP Mouse hippocampal tissue 20 0 GAP43 50 WT 0 100 200 300 [GAP43] (µM) Cnr1-/- GAP43 IgG 50 CB1R 50 GAP43 WCL (0.02%) GAP43-IP Syn1 GAP43 CB1R E WT mice PLA SIGNAL 10 µm Merged Colocalization with Syn1 (%) * 40 * 30 20 10 0 GAP43 CB1R GAP43/ only only CB1R Cnr1-/- mice 2 μm PLA signal / synaptosome (%) D * 150 100 50 0 WT Cnr1-/- Figure 1. GAP43 interacts with CB1R. A. Schematic workflow of the affinity purification and tandem MS/MS experiment conducted. A sheep whole-brain homogenate was loaded onto a lectin-hCB1R-CTD-bound Sepharose 4B column. After washing, elution with lactose, eluted-fraction separation by SDS-PAGE, and digestion with trypsin, peptides were subjected to nLC/MS-MS proteomic analysis. B. Fluorescence polarization (FP)-based protein–protein binding experiments using 5-IAF-labeled CB1R-CTD and increasing amounts of unlabeled GAP43. FP was expressed as milli-FP units. Each point is the mean of 3 different experiments. C. Co-immunoprecipitation experiments in (top) primary mouse hippocampal neurons or (bottom) mouse hippocampal tissue. Immunoprecipitation (IP) was conducted with anti-GAP43 antibody. Whole cell lysates (WCL) from WT and Cnr1-/- mice are shown, together with the IgG control immunoprecipitation. D. Representative confocal images of hippocampal synaptosomes of WT mice immunostained for synaptophysin 1 (Syn1), GAP43 and CB1R, and quantification of the percentage of Syn1+ synaptosomes that colocalize with either CB1R only, GAP43 only, or both CB1R and GAP43 (n = 5 mice; *p < 0.05 by two-tailed unpaired Student’s t-test). E. PLA for CB1R and GAP43 was performed in hippocampal synaptosomes from WT mice and Cnr1-/mice as control. Representative confocal images of CB1R-GAP43 complexes appearing as red signal, and quantification of the number of PLA-positive signal per synaptosome (n = 3 mice per genotype; *p < 0.05 by two-tailed unpaired Student’s t-test). A Phosphorylation site Membraneanchoring domain AATKIQASFRGHITRKKLKGEKKG 1-31 N-terminal domain 32-52 Effector domain Ser41 Ala41 (S41A) Ser41 Asp41 (S41D) hGAP43 67-238 C-terminal domain B GAP43(WT) GAP43(S41D) GAP43(S41A) 10 μm PLA dots / cell (%) * 300 200 100 0 8 Normalized OD GAP43(S41A) GAP43(S41D) GAP43(WT) GAP43(S41A) GAP43(WT) GAP43(S41D) PLA SIGNAL DAPI C GAP43 50 70 * * 6 4 2 0 HA -CB1R Tubulin 50 WCL (0.02%) HA – IP D E CB1R + control + WIN CB1R + GAP43(WT) + WIN CB1R + GAP43(S41D) + WIN CB1R + GAP43(S41A) + WIN CB1R-Rluc + GFP-GAP43(WT) CB1R-Rluc + GFP-GAP43(S41D) CB1R-Rluc + GFP-GAP43(S41A) 80 300 60 Response (pm) Signal (mBU) * 40 20 0 0 20 40 60 GFP / Rluc 200 100 0 0 1000 2000 Time (s) Figure 2. Phosphorylation of GAP43 at S41 facilitates its interaction with CB1R. A. Scheme of the mutant constructs aimed to modify GAP43 activation state. B. PLA for CB1R and GAP43 was performed with anti-c-myc and anti-GFP antibodies in HEK293T cells transfected with CB1R-myc plus GFP-GAP43(WT), GFP-GAP43(S41D) or GFP-GAP43(S41A). Left, Representative confocal microscopy images show CB1R-GAP43 complexes appearing as red dots. Cell nuclei were stained with DAPI (blue). Right, Quantification of PLA-positive dots per GFP-transfected cell (n = 6 experiments; *p < 0.05 by one-way ANOVA with Tukey’s multiple comparisons test). C. Left, Co-immunoprecipitation experiments in HEK293T cells cotransfected with HA-tagged CB1R and GAP43(WT), GAP43(S41D), GAP43(S41A). Whole cell lysates (WCL) are shown. Right, Quantification of optical density (OD) values of co-immunoprecipitated GAP43 relative to those of HA-CB1R are shown (n = 7 experiments; *p < 0.05 by one-way ANOVA with Tukey’s multiple comparisons test). D. BRET saturation experiments in HEK293T cells expressing CB1R-Rluc and increasing amounts of GFP-GAP43(WT), GFP-GAP43(S41D) or GFPGAP43(S41A). BRET is expressed as milli-BRET units (mBU) (n = 3 experiments). E. DMR assays in HEK293T cells transfected with CB1R plus GAP43(WT), GAP43(S41D), GAP43(S41A) or a control empty vector, and exposed to 100 nM WIN-55,212–2 (WIN). A representative experiment is shown (n = 3 experiments). A PLA SIGNAL DAPI 20 μm 40 ** ** 30 20 10 0 Cnr1-/- Glu-Cnr1-/- B * PLA dots / field Cnr1fl/fl GABA-Cnr1-/- 5 μm 5 μm Cnr1-RS Glu-Cnr1-RS 20 μm 50 PLA dots / field Stop-Cnr1 GABA-Cnr1-RS PLA SIGNAL DAPI ** ** ** ** 40 30 20 10 0 Figure 3. GAP43 interacts with CB1R in MC axon terminals of the DG. PLA experiments were performed in hippocampal sections from 3-month-old mice of different genotypes. GAP43-CB1R complexes are shown as PLApositive red dots. Nuclei are stained with DAPI (blue). A. Representative images of DG-IML sections from Cnr1fl/fl, GABA-Cnr1-/-, Glu-Cnr1-/- and full Cnr1-/- mice. Arrows point to some of the complexes. The inset magnifies a PLApositive dot. Quantification of PLA-positive dots per field is shown. B. Representative images of DG-IML sections from Stop-Cnr1, GABA-Cnr1-RS, Glu-Cnr1-RS and Cnr1-RS mice. Arrows point to some of the complexes. The inset magnifies a PLA-positive dot. Quantification of PLA-positive dots per field is shown. In panels A and B, n = 5 - 7 mice per group (*p < 0.05, **p < 0.01 by one-way ANOVA with Tukey’s multiple comparisons test). A AAV1/2 Recording WT mice Control IML GC GC GC GAP43(S41D) 1.4 ** ** 0.5 0.4 1.2 CV 1.0 0.8 0.3 0.2 0.6 0.4 0.1 0.2 Control 2 2 125 50 pA 1 10 ms 50 pA 1, 2 10 ms 10 ms 1 100 75 Control 8c/4m GAP43(S41A) 9c/5m GAP43(S41D) 10c/4m 2 50 0.0 0.0 GAP43(S41D) GAP43(S41A) 50 pA 1 10 ms ** ** 1.6 C 25 pA 10 ms 1.8 PPR IML 25 pA 10 ms GAP43(S41D)-CFP IML GAP43(S41A) 25 pA GAP43(S41A)-CFP EPSC amplitude (%) B CFP (Control) 0 1 2 3 Time (min) Control GAP43(S41A) 2 2 1 0.2 mV 1 10 ms 125 fEPSP amplitude (%) GAP43(S41D) 0.2 mV 1, 2 10 ms 0.2 mV 10 ms 2 Control 8s/5m GAP43(S41A) 8s/5m GAP43(S41D) 8s/5m 0 10 20 30 40 Time (min) 2 2 1 1 0.1 mV 1 0.1 mV 0.1 mV 10 ms 10 ms WIN 5 µM 1 100 2 75 50 50 GAP43(S41D) 2 125 1 75 GAP43(S41A) Control 10 ms WIN 5 µM 100 50 E fIPSP amplitude (%) D Control 5s/2m GAP43(S41A) 5s/2m GAP43(S41D) 7s/3m 0 10 20 30 40 50 Time (min) Figure 4. Phosphorylated GAP43 inhibits CB1R function at MC-GC synapses. A. Schematic diagram illustrating the injection of AAV1/2-CBA-CFP (control), AAV1/2-CBA-GAP43(S41A)-CFP or AAV1/2-CBA-GAP43(S41D)-CFP in the hilus of 3-week-old WT mice. Electrophysiological recordings were performed in the contralateral DG. Infrared differential interference contrast (left) and fluorescence (right) images showing CFP expression in the commissural MC axon terminals in the contralateral DG. Note the presence of CFP-positive fibers in the IML and its absence in the GC layer. B. Whole-cell patchclamp recordings were performed on GCs from injected mice. Representative traces (top) and quantification bar graph (bottom) for basal PPR and CV are shown (n = 8 - 12 cells; **p < 0.01 by one-way ANOVA with Tukey’s multiple comparisons test). C. DSE magnitude in control, GAP43(S41A) or GAP43(S41D)-injected mice. Representative traces (top) and timecourse summary plot (bottom) are shown [c = cells, m = mice; Control vs GAP43(S41D) and GAP43(S41A) vs GAP43(S41D) p < 0.001 by one-way ANOVA with Tukey’s multiple comparisons test]. D. fEPSPs recorded from control, GAP43(S41A) and GAP43(S41D)-injected mice upon WIN-55,212-2 bath application (WIN; 5 μM, 25 min). Representative fEPSP traces, before and after WIN application (top), and time-course summary plot (bottom) are shown [s = slices, m = mice; Control vs GAP43(S41D) and GAP43(S41A) vs GAP43(S41D) p < 0.05 by one-way ANOVA with Tukey’s multiple comparisons test]. E. fIPSPs recorded from control, GAP43(S41A) or GAP43(S41D)-injected mice upon WIN bath application (5 μM, 25 min). Representative fIPSP traces, before and after WIN application (top) and time-course summary plot (bottom) are shown (s = slices, m = mice; n. s. by one-way ANOVA with Tukey’s multiple comparisons test). Optical stimulation GC 25 ms 0.2 0.0 2 1 25 pA 1 o-EPSC amplitude (%) 0.4 * 1.5 1.0 0.5 0.0 100 75 25 2 0 1 2 Time (min) 150 Gap43-/- 25 pA 10 ms 2 1 25 pA 10 ms WIN (5 µM) 1 100 1 50 2 1 25 pA 10 ms WT (7c/4m) Gap43-/- (12c/7m) 125 WT Gap43-/- 10 ms 150 CV PPR 0.0 2 0.6 1.2 0.4 WT 25 ms 1.6 0.8 E D 25 pA 2.0 25 µm o-EPSC amplitude (%) Gap43-/- 25 pA GC 10 µm Gap43-/- WT or Gap43fl/fl mice WT Merged GC IML C IML 100µm 200µm IML WT AAV5-CaMKII-Cre mCherry AAV5-FLEX-ChIEF tdTomato mCherry GAP43 GAP43 immunoreactive area in contralateral IML B A 3 2 50 0 -50 WT (7c/4m) Gap43-/- (12c/7m) 10 20 Time (min) 30 40 Figure 5. GAP43 genetic deletion from MCs enhances CB1R function at MC-GC synapses. A. Left, Schematic diagram illustrating the injection of a mix of AAV5-CamKII-Cre-mCherry and AAV5-FLEX-ChIEF-tdTomato in the hilus of 3-week-old Gap43fl/fl and WT mice. Light pulses of 0.5 – 2.0 ms were used to evoke EPSCs driven by the ChIEF-expressing axons originating from commissural MCs. B. Reduced GAP43 immunoreactivity (green) in AAV-infected mCherry+ fibers of the contralateral IML from injected Gap43-/- mice compared to WT mice. Representative images and quantification are shown (n = 3 - 4 mice per group; *p < 0.05 by two-tailed unpaired Student’s t-test). C. Whole-cell patch-clamp recordings of GCs were performed in Gap43-/- and WT mice. Representative traces (top) and quantification bar graph for basal PPR (bottom) are shown (n = 7 - 9 cells; n. s. by two-tailed unpaired Student's t-test). D. Representative traces (top) and time-course summary plot (bottom) for DSE are shown (c = cells, m = mice; Gap43-/- vs WT n. s. by two-tailed unpaired Student's t-test). E. Representative traces (top) and time-course summary plot (bottom) for o-EPSC amplitude upon WIN-55,212-2 bath application (WIN; 5 μM, 20 min). (c = cells, m = mice; WT vs Gap43-/- at time = 15 min p < 0.01 by two-tailed unpaired Student's t-test). A Veh / THC 10 mg/kg i.p. Glu-Gap43fl/fl Glu-Gap43-/GABA-Gap43fl/fl GABA-Gap43-/- 120 min monitoring 5 4 3 2 0 30 7 90 Sacrifice Veh (n=10) THC (n=10) 5 4 3 2 0 120 0 30 60 90 120 Time (min) Veh (n=15) THC (n=14) 5 4 3 2 1 GABA-Gap43-/- 7 6 Racine score Racine score 60 Time (min) GABA-Gap43fl/fl 6 0 6 1 1 0 Glu-Gap43-/- 7 Veh (n=10) THC (n=9) Glu-Gap43fl/fl 6 Racine score 15 min Racine score 7 Seizure induction * B KA 30 mg/kg i.p. Veh (n=16) THC (n=16) 5 4 3 2 1 0 30 60 90 120 0 0 30 Time (min) C * 200 150 100 50 0 Veh THC Veh THC GluGap43fl/fl GluGap43-/- 120 250 Seizure severity (%) Seizure severity (%) 250 60 90 Time (min) 200 150 100 50 0 Veh THC Veh THC GABAGap43fl/fl GABAGap43-/- Figure 6. Enhanced anti-convulsant response to THC in Glu-Gap43-/- mice. A. Timeline of excitotoxic seizureinduction experiments. Vehicle or THC (10 mg/kg, i.p.; 1 injection) was administered to 3-month-old Glu-Gap43-/- or GABA-Gap43-/- mice and their corresponding Gap43fl/fl littermates. Kainic acid (KA; 30 mg/kg, i.p.; 1 injection) was administered 15 min later, and behavioral score was monitored for 120 min. B. Scoring of seizures by using a modified Racine scale at 5-min intervals (number of animals in parentheses; *p < 0.05 by two-way ANOVA with Sidak`s multiple comparisons). C. Integrated seizure severity expressed as normalized percentage from the respective Gap43fl/fl / Vehicle group (n = 9 - 16 mice per group; Glu-Gap43-/- / Vehicle vs Glu-Gap43-/- / THC, *p < 0.05 by two-way ANOVA with Tukey`s multiple comparisons test). Supplementary Files This is a list of supplementary les associated with this preprint. Click to download. MarotoSI.pdf