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