Spatial organization of neuron–astrocyte interactions in the somatosensory cortex

Abstract

Microcircuits in the neocortex are functionally organized along layers and columns, which are the fundamental modules of cortical information processing. While the function of cortical microcircuits has focused on neuronal elements, much less is known about the functional organization of astrocytes and their bidirectional interaction with neurons. Here, we show that Cannabinoid type 1 receptor (CB1R)-mediated astrocyte activation by neuron-released endocannabinoids elevate astrocyte Ca²⁺ levels, stimulate ATP/adenosine release as gliotransmitters, and transiently depress synaptic transmission in layer 5 pyramidal neurons at relatively distant synapses (˃20 μm) from the stimulated neuron. This astrocyte-mediated heteroneuronal synaptic depression occurred between pyramidal neurons within a cortical column and was absent in neurons belonging to adjacent cortical columns. Moreover, this form of heteroneuronal synaptic depression occurs between neurons located in particular layers, following a specific connectivity pattern that depends on a layer-specific neuron-to-astrocyte signaling. These results unravel the existence of astrocyte-mediated nonsynaptic communication between cortical neurons and that this communication is column- and layer-specific, which adds further complexity to the intercellular signaling processes in the neocortex.

Introduction

The neocortex is the most complex structure of the mammalian brain involved in higher cognitive functions. The cellular organization of the cerebral cortex is well known since the work of Cajal and his disciple Lorente de Nó, who proposed that cortical neurons form functional modules that serve as the “elementary cortical unit of operation” (Lorente de Nó 1933; Lorente de No 1938). Cortical neurons are organized horizontally in 6 layers and vertically in columns (Mountcastle 1997). The columnar configuration of the neocortex has been a widely accepted concept with useful implications in explaining its functional organization (Markram 2008; DeFelipe 2012). A cortical column refers to cells in any vertical cluster that share the same tuning for any given receptive field attribute (Horton and Adams 2005). In the present study, we follow the concept of columnar organization of the cortex, although it is noteworthy that such idea remains debated (Horton and Adams 2005; da Costa 2010). A great amount of information has been provided regarding the neuronal elements involved in cortical circuits and their synaptic micro-organization (Harris and Shepherd 2015). However, the properties of nonneuronal cell types, such as astrocytes, and their functional interactions with neurons in this elementary module remain largely unexplored.

Astrocytes have emerged as key regulatory elements of synapses, responding with Ca²⁺ elevations to synaptically released neurotransmitters and releasing gliotransmitters that regulate synaptic transmission in different brain areas (Perea et al. 2009; Araque et al. 2014; Volterra et al. 2014). In the cortex, sensory stimuli or direct neuronal stimulation evoke astrocyte Ca²⁺ elevations (Wang et al. 2006; Schipke et al. 2008; Schummers et al. 2008; Benedetti et al. 2011; Takata et al. 2011; Min and Nevian 2012; Perez-Alvarez et al. 2014; Lines et al. 2020), which are topographically represented in the primary somatosensory cortex S1 (Ghosh et al. 2013) and spatially restricted to the cortical columns in the barrel cortex (Schipke et al. 2008). Cortical astrocyte Ca²⁺ elevations can, in turn, stimulate the release of gliotransmitters, such as glutamate or D-Serine, that can regulate synaptic transmission (Takata et al. 2011; Min and Nevian 2012) and that can be responsible for the observed astrocyte-mediated regulation of the cortical network function (Poskanzer and Yuste 2011, 2016; Lines et al. 2020, 2021). Moreover, synaptic regulation by astrocytes may be exerted at synapses relatively distant from the active synapses (Zhang et al. 2003; Pascual et al. 2005; Navarrete and Araque 2010). This phenomenon, termed lateral astrocyte synaptic regulation (Covelo and Araque 2016), resembles the classical heterosynaptic modulation but is mechanistically dissimilar because it involves astrocytes and may be crucial in brain circuits where spatial signaling greatly influences neural network function, such as the neocortical columns. However, the spatial properties of astrocyte–neuron interaction and the consequent synaptic regulation in defined cortical columns and layers remain unidentified.

The endocannabinoid system plays a key role in multiple brain functions through the modulation of synaptic transmission. In the neocortex, endocannabinoids (eCBs) have been found to regulate synaptic transmission and plasticity through activation of presynaptic type-1 cannabinoid receptors (CB1Rs) (Sjöström et al. 2003; Fortin and Levine 2007; for reviews see Alger 2002; Castillo et al. 2012; Augustin and Lovinger 2018). However, eCB signaling also has been proposed to mediate astrocyte–neuron communication in different brain regions, including the neocortex. In particular, endogenous activation of astroglial type-1 cannabinoid receptors (CB₁Rs) regulates hippocampal and neocortical synaptic transmission and plasticity (Navarrete and Araque 2010; Han et al. 2012; Min and Nevian 2012; Gómez-Gonzalo et al. 2015). Here, in order to decipher the regulatory role of astrocytes in the synaptic physiology of cortical columns, we took advantage of this eCB signaling to physiologically stimulate cortical astrocytes. We show that eCBs released from layer 5 (L5) pyramidal neurons induce Ca²⁺ elevations in astrocytes and transiently depressed synaptic transmission in adjacent pyramidal neurons. This form of heteroneuronal synaptic depression required astrocytic cannabinoid receptor type (CB₁R) activation and was mediated by presynaptic type 1 adenosine receptors (A1Rs). Astrocyte-mediated heteroneuronal synaptic depression was present between pyramidal neurons within a cortical column and was absent in neurons belonging to adjacent cortical columns. Moreover, this form of heteroneuronal synaptic depression occurred between neurons located in particular layers, following a specific connectivity pattern that depends on a layer-specific neuron-to-astrocyte signaling. These results reveal the existence of astrocyte-mediated nonsynaptic communication between cortical neurons, which is column- and layer-specific and which adds further complexity to the intercellular signaling processes in the cortex.

Methods

Ethics statement

All of the procedures for handling and sacrificing animals were approved by the University of Minnesota Institutional Animal Care and Use Committee (IACUC) in compliance with the National Institutes of Health guidelines for the care and use of laboratory animals.

Animals

Mice were housed under 12/12-h light/dark cycle, up to 5 animals per cage, at temperatures between 68 and 74°F at 30–70% humidity with freely available food and water. The following animals (males and females) were used for the present study C57BL/6 J, IP₃R₂^−/− (generously donated by Dr J Chen), and CB₁R^fl/fl (Marsicano et al. 2003; Li et al. 2005). Adult (≥8 weeks) mice were used.

Somatosensory cortex slice preparation

Mice were euthanized by decapitation and brains were rapidly removed and placed in ice-cold artificial cerebrospinal fluid (ACSF). Three-hundred and fifty-micrometer coronal brain slices containing the somatosensory cortex were prepared via a Leica VT1200 vibratome in a 4°C ACSF solution. Following cutting, slices were allowed to recover in ACSF containing (in mM): NaCl 124, KCl 2.69, KH₂PO₄ 1.25, MgSO₄ 2, NaHCO₃ 26, CaCl₂ 2 and glucose 10, gassed with 95% O₂/5% CO₂ (pH = 7.3) at 31°C for 30 min followed by 30 min at 20–22°C before recording. After a 1-h recovery period, slices were kept at 20–22°C for the rest of the recording day. Slices were then transferred to an immersion recording chamber and superfused at 2 mL/min with gassed ACSF and the temperature of the bath solution was kept at 34°C with a temperature controller TC-324B (Warner Instruments Co.). Cells were visualized using infrared-differential interference contrast optics (Nikon Eclipse E600FN, Tokyo, Japan) and 40x water immersion lens. L2/3, L4, and L5 from the forelimb and hindlimb somatosensory cortex and the barrel subfields were identified with a 10x objective.

Electrophysiology

Neurons were selected based on their location, morphology, and firing pattern. Simultaneous dual electrophysiological recordings from layers 2/3, 4, and 5 pyramidal neurons were made using the whole-cell-patch-clamp technique. When filled with an internal solution containing (in mM): KGluconate 135, KCl 10, HEPES 10, MgCl₂ 1, ATP-Na₂ 2 (pH = 7.3) patch electrodes exhibited a resistance of 3–10 MΩ. All recordings were performed using PC-ONE amplifiers (Dagan Instruments, Minneapolis, MN). Fast and slow whole-cell capacitances were neutralized, and series resistance was compensated (~70%), and the membrane potential was held at −70 mV. Intrinsic electrophysiological properties were monitored at the beginning and the end of the experiments. Series and input resistances were monitored throughout the experiment using a −5 mV pulse. Recordings were considered stable when the series and input resistances, resting membrane, and stimulus artifact duration did not change >20%. Furthermore, I–V curves and firing patterns at the beginning and the end of the experiments were similar. Recordings that did not meet these criteria were discarded. Signals were fed to a Pentium-based PC through a DigiData 1322A interface board. Signals were filtered at 1 kHz and acquired at a 10-kHz sampling rate using a DigiData 1322A data acquisition system and pCLAMP 10.3 software (Molecular Devices, San Jose, CA). Distance of the somas of the paired recorded neurons within a layer varied from 70 to 270 μm. In paired recordings across layers 2/3, 4, and 5, neurons were selected following the same vertical axis. In the barrel cortex, the distance of the somas of the paired recorded L4 neurons within the same (intracolumnar) or in adjacent columns (intercolumnar) was similar (70–270 μm). When intra- or intercolumns pair recordings were performed, cells located in septa were excluded from the experiment. All the experiments were performed in the primary somatosensory cortex and, specifically, in the barrel cortex for those reported in Fig. 4.

Fig. 4.

Astrocyte-mediated heterosynaptic depression is column specific. A) Representative infrared differential interference contrast images of the barrel field in the barrel cortex showing intracolumn (left) and intercolumn (middle) pair of neurons patched and the stimulation electrode. False-color biocytin loading barrel cortex intercolumnar pair of neurons image (right). B) Left: EPSCs amplitude versus time before (basal) and after ND in the intracolumn heteroneuron. Right: Relative changes in EPSC amplitude in the intracolumn heteroneuron. C) Left: EPSCs amplitude versus time before (basal) and after ND in the intercolumn heteroneuron. Right: Relative changes in EPSC amplitude in the intercolumn heteroneuron. Two-tailed Student’s paired t-test. Data are expressed as mean ± SEM, ^**P < 0.01.

Synaptic stimulation

Theta capillaries (2–5 μm tip) filled with ACSF were used for bipolar stimulation. The electrodes were connected to a stimulator S-900 through an isolation unit and placed in L2/3. When indicated, the stimulation electrode was placed in L4. Paired pulses of 1-ms duration and 50-ms interval were continuously delivered at 0.33 Hz. Excitatory postsynaptic currents (EPSCs) were isolated using picrotoxin (50 μM) and CGP5462 (1 μM) to block GABA_AR and GABA_BR, respectively. EPSC amplitude was determined as the peak current amplitude (2–20 ms after stimulus) minus the mean baseline current (10–30 ms before stimulus). The paired-pulse ratio (PPR) was estimated as PPR = (2nd EPSC/1st EPSC).

To induce eCB release, pyramidal neurons were depolarized from −70 to 0 mV for 5 s (ND; Ohno-Shosaku et al. 2001; Wilson and Nicoll 2001). Synaptic parameters were determined from 60 stimuli before (basal) and following ND. Baseline mean EPSC amplitude was obtained by averaging mean values obtained within 3 min of baseline recordings and mean EPSC amplitudes were normalized to baseline. The ND was applied 2.5 s after the last basal delivered pulse, and no pulses were presented during the ND. Immediately after the ND was finished, the 0.33-Hz pulse protocol was restarted. To illustrate the time course of ND-induced effects, synaptic parameters were grouped in 15 s bins. Three consecutive responses to ND were averaged. For every synaptic recording, the presence of homoneuronal or heteroneuronal depression was assessed in individual synaptic recordings if the EPSC amplitude decreased >2 times the standard deviation of the baseline EPSC amplitude during the first 45 s after the ND. To characterize pharmacologically the phenomena, we first identified specific synapses undergoing either homoneuronal or heteroneuronal depression in control conditions and then analyzed the effects of subsequent superfusion of pharmacological agents (CPT 5 μΜ and AM251 2 μΜ) on the synapses previously identified. In all cases, the effects of pharmacological agents were tested after 10-min bath perfusion and at <40 min after entering whole-cell mode in the stimulating neuron.

Ex vivo 2-photon Ca²⁺ fluorescence imaging and electrophysiology

Two-photon microscopy imaging was performed using a Leica SP5 multi-photon microscope (Leica Microsystems, USA) controlled by the Leica LAS software and adapted to perform electrophysiological recordings. C57BL/6 J, IP₃R₂^−/−, and aCB₁R mice injected into S1 with AAV5-GfaABC1d-GCaMP6f and AAV8-GFAP-mCherry were used (for aCB₁R^−/− instead AAV8-GFAP-mCherry we used AAV8-GFAP-mCherry-Cre). All Ca²⁺ experiments, except those in which synaptic transmission was recorded, were performed in the presence of TTX (1 μM) and a cocktail of neurotransmitter receptor antagonists containing: CNQX (20 μΜ), AP5 (50 μΜ), MPEP (50 μΜ), LY367385 (100 μΜ), picrotoxin (50 μΜ), CGP5462 (1 μΜ), atropine (50 μΜ), CPT (5 μΜ), flupenthixol (30 μΜ), and suramin (100 μΜ).

Videos were obtained at 512 × 512 resolution with a sampling interval of 1 s. Red and green fluorescence was obtained in parallel to match red mCherry-stained astrocyte structure with green GCaMP6f astrocyte Ca²⁺. A custom MATLAB program (Calsee: https://www.araquelab.com/code/) was used to quantify fluorescence level measurements in astrocytes. Ca²⁺ variations recorded at the soma and processes of the cells were estimated as changes of the fluorescence signal over baseline (ΔF/F₀), and cells were considered to show a Ca²⁺ event when the ΔF/F₀ increase was at least 2 times the standard deviation of the baseline.

The astrocyte Ca²⁺ signal was quantified from the Ca²⁺ event probability, which was calculated from the number of Ca²⁺ elevations grouped in 5-s bins recorded from 8 to 50 astrocytes per field of view (layer 2/3, 4, or 5 of S1). The time of occurrence was considered at the onset of the Ca²⁺ event. For each astrocyte analyzed, values of 0 and 1 were assigned for bins showing either no response or a Ca²⁺ event, respectively, and the Ca²⁺ event probability was obtained by dividing the number of astrocytes showing an event at each time bin by the total number of monitored astrocytes (Navarrete and Araque 2010). All the astrocytes that showed a Ca²⁺ event during the experiment were used for the analysis. The Ca²⁺ event probability was calculated in each slice, and for statistical analysis, the sample size corresponded to the number of slices as different slices were considered as independent variables. To examine the difference in Ca²⁺ event probability in distinct conditions, the basal Ca²⁺ event probability (mean of the 30 s before a stimulus) was averaged and compared with the average Ca²⁺ event probability (5 s after a stimulus). For ND experiments, each layer was recorded 1 min before and after the ND. Three consecutive responses to ND were averaged in each layer. For WIN application, a micropipette was filled with 300-μΜ WIN solution and placed 100–150 μm away from the tissue (layer 5), and a pressure pulse at 1 bar (PMI-100 DAGAN, Minneapolis, MN) was applied for 5 s. The absence of mechanical movement of the tissue was confirmed in every case. For acute application of CNO, a micropipette was filled with 1-mM CNO solution and placed 100–150 μm away from the recording neuron, and a pressure pulse was applied for 5 s. The absence of mechanical movement of the tissue was confirmed in every case. Stimulus effects on EPSCs were statistically tested comparing the normalized EPSCs recorded 1 min before and 30 s after the stimulus to assess changes in EPSC amplitude and PPR. Astrocytic Ca²⁺ events were recorded at the same time. The changes on the Ca²⁺ event probability after CNO application were statistically tested comparing the basal Ca²⁺ event probability 1 min before and 5 s after the stimulus.

The effects of pharmacological agents (CPT 5 μΜ and AM251 2 μΜ) were tested after 10 min bath perfusion in the same region and same astrocytes recorded in control conditions. In the cases when Ca²⁺ imaging and electrophysiology were performed at same time, the effects of pharmacological agents were tested at <40 min after entering whole-cell mode in the stimulating or recorded neuron.

A‌AV viral surgeries

Animals were anesthetized using a ketamine (10 mg/mL) xylazine (1 mg/mL) mixture and placed on a heating pad to maintain body temperature and faux tears were applied to the cornea. Animals (8 weeks of age) were placed in a stereotaxic apparatus and an incision was made down the midline of the scalp to expose the skull. A hole was drilled over the forelimb and hindlimb somatosensory cortex (S1: −0.4_a–p, 1.9_m–l), and a Hamilton syringe was lowered to (in mm from bregma: −0.7_d–v) and viruses were injected bilaterally at 100 nL/min (Paxinos and Franklin 2012). Mice were then sutured and left to heal for 2–3 weeks.

AAV5-pZac2.1-gfaABC1d-cyto-GCaMP6f (Addgene), AAV8-GFAP-hM3D(Gq)-mCherry (UMN vector core), AAV8-GFAP-mCherry (UMN vector core), and AAV8-GFAP-mCherry-Cre (UMN vector core) viral constructs were used. For CNO experiments, C57BL/6 J mice were injected with AAV8-GFAP-hM3D(Gq)-mCherry virus. In control conditions, a virus of AAV8-GFAP-mCherry was injected instead. For CB₁R^fl/fl mice experiments, AAV8-GFAP-mCherry-Cre was injected to delete CB₁R from astrocytes (aCB₁R^−/−). AAV8-GFAP-mCherry was used as a control (aCB₁R). Virus titers were between 10¹⁰ and 10¹² genomic copies per ml for all batches of virus used in the study and were injected bilaterally at 500 nL. When more than one virus was used, they were mixed and injected at 500 nL.

Immunohistochemistry

The animals were anesthetized with Avertin (2,2,2 tribromoethanol, 240 mg/kg, i.p.) and intracardially perfused with ice cold phosphate buffered saline (PBS) and subsequently with 4% paraformaldehyde (PFA) in 0.1 M PBS (pH 7.4). The brain was removed, and 100-μm coronal sections were made using a Leica VT1000S vibratome. Vibratome sections were incubated for 1 h in blocking buffer (0.1% Triton X-100, 10% Donkey or Goat serum in PBS) at room temperature. The primary antibodies were diluted in the blocking solution and the sections were incubated for 2 days at 4 °C. The following primary antibodies were used: Rabbit anti-GFAP (Sigma, 1:500) Mouse anti-NeuN (Millipore, 1:500). The slices were then washed 3 times for 15 min each in PBS. The secondary antibodies were diluted in the secondary antibody buffer (0.1% Triton X-100, 5% Donkey or Goat serum in PBS) and incubated for 2 days at room temperature. The following secondary antibodies were used: 488 goat anti-rabbit (Invitrogen, 1:1,000), 405 goat anti-mouse (Invitrogen, 1:500). The sections were then washed 3 times with 1xPBS for 10 min each and mounted using Vectashield Mounting media (Vector laboratories). The slides were imaged using a Leica SP5 multiphoton confocal microscope and Olympus FluoView FV1000.

The cellular specificity of Cre viral vectors was tested by immunohistochemical analysis of randomly selected areas of the S1. Out of the 784 cells expressing mCherry from the AAV8-GFAP-mCherry-Cre viral vector, 86.7% were astrocytes (identified by GFAP) and 11.3% were neurons (identified by NeuN) (Fig. S2A and B).

Biocytin-stained neurons

Pair of neurons were recorded with patch pipettes and filled with internal solution containing 0.5% biocytin. Slices were fixed in 4% PFA in 0.1 PBS (pH 7.4) at 4 °C. Slices were washed 3 times in 1xPBS (10 min each). To visualize biocytin slices were incubated with Alexa488-Streptavidin (RRID: AB 2315383; 1:500) for 48 h at 4 °C, slices were then washed for 3 times with 1xPBS (10 min each) and mounted with Vectashield mounting media (Vector laboratories). All mounted slices were imaged using a Leica SP5 multi-photon microscope. Also, pair of neurons were filled with biocytin through whole-cell recording, the slices were fixed using 4% paraformaldehyde. Then, the slices were washed with PBS (100-mM sodium phosphate, pH 7.2). Endogenous peroxidases were then quenched by incubation with 1% H2O2. The slices were subsequently rinsed in PBS. Slices were conjugated with avidin-biotinylated horseradish peroxidase following the manufacturer’s instructions (ABC-Elite, Vector stains). Slices were then washed, and subsequently, biocytin-stained neurons were visualized under a reaction with 0.5-mg/mL DAB and 0.01% H2O2. When the neuronal processes were visible, the reaction was stopped by washing with PBS.

Drugs and chemicals

4-[3-[2-(Trifluoromethyl)-9H-thioxanthen-9-ylidene]propyl]-1-piperazineethanol dihydrochloride (flupenthixol dihydrochloride), [S-(R^*,R^*)]-[3-[[1-(3,4-Dichlorophenyl)ethyl]amino]-2-hydroxypropyl](cyclohexylmethyl) phosphinic acid (CGP 54626 hydrochloride), 8,8′-[Carbonylbis[imino-3,1-phenylenecarbonylimino(4-methyl-3,1-phenylene)carbonylimino]]bis-1,3,5-naphthalenetrisulfonic acid hexasodium salt (suramin hexasodium salt), N-(Piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3- carboxamide (AM251), D-(−)-2-Amino-5-phosphonopentanoic acid (D-AP5), 6-Cyano-7-nitroquinoxaline-2,3-dione disodium (CNQX disodium salt), (S)-(+)-a-Amino-4-carboxy-2-methylbenzeneacetic acid (LY367385), and 2-Methyl-6-(phenylethynyl)pyridine hydrochloride (MPEP hydrochloride), Octahydro-12-(hydroxymethyl)-2-imino-5,9:7,10a-dimethano-10aH-[1,3]dioxocino[6,5-d] pyrimidine-4,7,10,11,12-pentol (Tetrodotoxin: TTX) were purchased from Tocris Bioscience. Endo-(±)-α-(Hydroxymethyl)benzeneacetic acid 8-methyl-8-azabicyclo[3.2.1] oct-3-yl ester (atropine) and 8-Cyclopentyl-1,3-dimethylxanthine (CPT) were from Sigma. Picrotoxin from Indofine Chemical Company (Hillsborough, NJ). All other drugs were purchased from Sigma.

Statistical analysis

Number of neurons was used as a sample size for electrophysiology comparisons and number of slices for Ca²⁺ signal comparisons. At least, 2 mice per experimental group were used. Data are expressed as mean ± standard error of the mean (SEM). Data normality was tested using a Shapiro–Wilk test. Results were compared using a two-tailed Student’s t-test (Paired, before-after stimulus-treatment; Unpaired between groups). A full report of the statistics used in every case is detailed in Table S1. Statistical differences were established with P < 0.05 (^*), P < 0.01 (^**) and P < 0.001 (^***).

Results

Endocannabinoid signaling induces homoneuronal and heteroneuronal synaptic depression in S1

To investigate the spatial regulation of synaptic transmission in the primary somatosensory cortex, we performed double patch-recordings of layer 5 (L5) pyramidal neurons and monitored excitatory postsynaptic currents (EPSCs) evoked by electrical stimulation of layer 2/3 (L2/3). We then stimulated 1 neuron by a depolarizing pulse and recorded synaptic currents in both the “stimulated” neuron (homoneuronal synapses) and the adjacent (70–270 μm apart) “nonstimulated” neuron (heteroneuronal synapses) (Fig. 1A and B). Stimulation of single L5 pyramidal neurons induced a transient synaptic depression in 36 out of 93 (38.7%) homoneuronal synapses. Furthermore, in simultaneously recorded heteroneuronal synapses, this neuronal depolarization (ND) also induced a transient depression of synaptic transmission in 20 out of 72 (27.8%) heteroneuronal synapses (Fig. 1C–E). Both homoneuronal and heteroneuronal synaptic depressions could be reliably induced by repeated stimulations (Fig. S1a and b) and were associated with changes in the PPR, which are consistent with presynaptic mechanisms (Fig. S1c and d). No correlation has been found between homoneuronal and heteroneuronal depression indicating that are independent phenomena (Fig. S1f). Heteroneuronal synaptic depression was found to be nonreciprocal between cells, i.e. only in one case out of 20 (5%) recorded pairs of neurons ND in 1 neuron evoked heteroneuronal depression in the other neuron and underwent heteroneuronal depression by depolarizing the other neuron.

Fig. 1.

Endocannabinoid signaling induces homoneuronal and heteroneuronal synaptic depression in S1. A) Biocytin loading S1 cortex L5 pyramidal neurons image. B) Schematic drawing depicting double patch-recordings from L5 pyramidal neurons and the stimulating electrode in L2/3. C) Averaged EPSCs (n = 20 stimuli) before (control) and after ND in wild-type mice. D) EPSCs amplitude versus time before (basal) and after ND in control (green or blue) and in the presence of AM251 (2 μM; open gray) in the homoneuron (left) and heteroneuron (right) from layer 5. E) Relative changes in EPSC amplitude in control and with AM251 (2 μM). Two-tailed Student’s paired t-test. F) Representative infrared differential interference contrast image of the experimental configuration with the stimulation pipette in layer 4 and the homoneuronal and heteroneuronal neurons located in layer 4. G) EPSCs amplitude versus time before (basal) and after ND in the homoneuron (left, green) and heteroneuron (middle, blue) in the experimental conditions represented in panel f. Right: Relative changes in EPSC amplitude. H) Representative infrared differential interference contrast image of the experimental configuration with the stimulation pipette in L2/3 and the homoneuronal and heteroneuronal neurons located in L2/3. I) EPSCs amplitude versus time before (basal) and after ND in the homoneuron (left, green) and heteroneuron (middle, blue) in the experimental conditions represented in panel h. Right: Relative changes in EPSC amplitude. Two-tailed Student’s paired t-test. Data are expressed as mean ± SEM, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

ND is known to trigger the release of eCBs (Ohno-Shosaku et al. 2001; Wilson and Nicoll 2001) that can directly affect relatively close synapses (~20 μm) (Wilson and Nicoll 2001; Chevaleyre and Castillo, 2003; Piomelli 2003; Chevaleyre and Castillo, 2004), a phenomenon called depolarization-induced suppression of excitation (DSE; Wilson and Nicoll 2001; Chevaleyre and Castillo, 2003; Piomelli 2003; Chevaleyre and Castillo, 2004; Diana and Marty 2004; Castillo et al. 2012), and indirectly regulate more distant synapses through stimulation of astrocytes (Navarrete and Araque 2010; Min and Nevian 2012; Gómez-Gonzalo et al. 2015), a phenomenon called astrocyte-mediated lateral regulation of synaptic transmission. Consistent with eCB-mediated synaptic regulation, homoneuronal and heteroneuronal synaptic depressions observed under control conditions were abolished following bath perfusion with the cannabinoid receptor type 1 (CB₁R) antagonist AM251 (2 μM; n = 12 and 6), indicating that both phenomena were mediated by CB₁R activation (Fig. 1D and E).

We then tested whether these phenomena were present in other cortical layers by performing paired recordings of neurons in L2/3 and L4. L2/3 or L4 ND induced both homoneuronal (14 out of 38 and 8 of 32 pairs; 36.8% and 25%, respectively) and heteroneuronal (14 out of 39 and 8 of 27 pairs; 35.9% and 29.6%, respectively) depression (Fig. 1F–I) with a similar DSE, which suggest similar mechanisms in the different layers of the S1 cortex.

Heteroneuronal, but not homoneuronal, synaptic depression requires endocannabinoid signaling in astrocytes

We then investigated the role of astrocyte CB₁Rs on the eCB-induced homoneuronal and heteroneuronal synaptic depression. We selectively deleted CB₁R expression in cortical astrocytes by expressing Cre-recombinase under the control of the astroglial promoter GFAP, using local injection of AAV8-GFAP-mCherry-Cre in S1 of CB₁R^flox/flox mice (Fig. 2A). These mice are herein termed aCB₁R^−/− mice, and their controls, termed aCB₁R mice, were CB₁R^flox/flox mice injected with AAV8-GFAP-mCherry (i.e. lacking Cre). We tested the impact of astroglial deletion of CB₁Rs on the ND-evoked homoneuronal and heteoneuronal synaptic depression in L5. Accordingly, the homoneuronal depression was not affected in mice lacking CB₁Rs in astrocytes (12 out of 35 cells; 34.3%) (Fig. 2B and C). In contrast, the heteroneuronal synaptic depression was absent in aCB₁R^−/− mice (0 out of 27 cells; 0%) (Fig. 2B and C). These results indicate that eCB-induced heteroneuronal, but not homoneuronal, synaptic depression involves CB₁R signaling in astrocytes.

Fig. 2.

Heteroneuronal, but not homoneuronal, synaptic depression requires endocannabinoid signaling in astrocytes. A) Viral vector injected into the S1 of CB₁R^fl/fl mice and fluorescence image showing mCherry-Cre expression in the S1 cortex (top), and immunohistochemistry images showing co-localization between mCherry-cre and GFAP (bottom). B) EPSCs amplitude versus time before (basal) and after ND in CB₁R mice injected with AAV8-GFAP-mCherry (aCB₁R; green or blue) or with AAV8-GFAP-mCherry-Cre (aCB₁R^−/−; purple) in the homoneuron (left) and heteroneuron (right) from L5. C) Relative changes in EPSC amplitude in aCB₁R and aCB₁R^−/− mice in the homoneuron (left) and heteroneuron (right). Two-tailed Student’s paired t-test. D) Viral vector injected into the S1 of CB₁R^fl/fl mice, fluorescence image showing GCaMP6f expression in the S1 and pseudocolor images showing the fluorescence intensities of GCaMP6f-expressing astrocytes before and after WIN (300 μΜ) application in L5. E) Ca²⁺ event probability over time (left) and Ca²⁺ event probability before (basal) and after WIN application in aCB₁R (black) and aCB₁R^−/− (purple) mice (right). Blue shadow indicates 5-s WIN application. Two-tailed Student’s paired t-test (before and after) and two-tailed Student’s unpaired t-test (between groups). F) Heat maps showing Ca²⁺-based fluorescence levels and raster plots showing Ca²⁺ events (i.e. when the ΔF/F0 increased at least 2 times the standard deviation of the baseline) recorded from all ROIs including astrocyte somas and processes in aCB1R (left) and aCB1R^−/− (right) mice before and after WIN stimulation. Blue shadow indicates 5-s WIN application. Data are expressed as mean ± SEM, ^**P < 0.01, ^***P < 0.001.

To assess the efficacy of the approach, we monitored the CB₁R-mediated astrocyte Ca²⁺ responses to the CB₁R agonist WIN 55,212-2 (300 μΜ) using 2-photon microscopy and the genetically encoded Ca²⁺ indicator GCaMP6f selectively expressed in astrocytes via injection of AAV5-gfaABC1D-cyto-GCaMP6f in S1 (Fig. 2D). While the astrocyte Ca²⁺ activity, quantified from the Ca²⁺ event probability, was increased by local application of WIN 55,212-2 in control aCB₁R mice (n = 190 astrocytes from 13 slices), the WIN-evoked responses were significantly reduced in aCB₁R^−/− mice (n = 261 astrocytes from 14 slices; Fig. 2D–F), confirming the suitability of the viral approach to delete CB₁R signaling in astrocytes.

Heteroneuronal synaptic depression requires astrocyte Ca²⁺ signaling and activation of presynaptic A1 receptors

Ca²⁺ elevations in astrocytes evoked by different neurotransmitters, including eCBs (Navarrete and Araque 2008, 2010; Martin-Fernandez et al. 2017), are known to stimulate the release of gliotransmitters that regulate synaptic function (e.g. Navarrete and Araque 2008, 2010; Min and Nevian 2012; Martin-Fernandez et al. 2017; Corkrum et al. 2020). Hence, we investigated whether the homoneuronal and heteroneuronal synaptic depressions depended on the astrocytic Ca²⁺ signal. We depolarized L5 pyramidal neurons using the approach that elicits homoneuronal and heteroneuronal synaptic depression as indicated above and monitored the astrocyte Ca²⁺ activity using GCaMP6f selectively expressed in astrocytes (Fig. 2D). To isolate eCB-induced effects, experiments were performed in the presence of a cocktail of antagonists of glutamatergic, GABAergic, purinergic, and cholinergic receptors (see Methods). Under these conditions, ND elevated astrocyte Ca²⁺ fluctuations (n = 142 astrocytes of 6 slices), an effect that was abolished in the presence of AM251, indicating that these responses were mediated by CB₁R activation (n = 122 astrocytes of 6 slices; Fig. 3A and B).

Fig. 3.

Heteroneuronal synaptic depression requires astrocyte Ca²⁺ signaling and activation of presynaptic A1 receptors. A) Left: Pseudocolor images showing the fluorescence intensities of GCaMP6f-expressing astrocytes in L5 of the S1 cortex before and after L5 ND in wild-type (top) and IP₃R₂^−/− mice (bottom). Right: Representative Ca²⁺ traces of astrocytes (arrow indicates ND). B) Left: L5 astrocytes Ca²⁺ event probability over time before (basal) and after L5 ND in wild-type (red) and IP₃R₂^−/− (purple) mice. Right: Relative changes in Ca²⁺ event probability in wild-type and IP₃R₂^−/− mice in control and with AM251 (2 μΜ). All experimental conditions were performed in TTX (1 μM) and in a cocktail of neurotransmitter receptor antagonists (see Material and methods). Two-tailed Student’s paired t-test. C) Top: EPSCs amplitude versus time before (basal) and after ND in wild-type mice in control (green or blue), in presence of CPT (5 μΜ) (open gray) and in IP₃R₂^−/− mice (purple) in the homoneuron (left) and heteroneuron (right) from layer 5. Bottom: Relative changes in EPSC amplitude in wild-type mice in control, in presence of CPT (5 μM) and in IP₃R₂^−/− mice. Two-tailed Student’s paired t-test. D) Top: Relative changes in EPSC amplitude before (basal) and after L2/3 ND in control and with CPT (5 μM) in the homoneuron (green) and heteroneuron (blue) from layer 2/3. Bottom: Relative changes in EPSC amplitude before (basal) and after L4 ND in control and with CPT (5 μM) in the homoneuron (green) and heteroneuron (blue) from layer 4. Two-tailed Student’s paired t-test. E) Left: Viral vectors injected into the S1 of wild-type mice and immunohistochemistry images showing the expression of NeuN (blue), mCherry (red) and GFAP (green) in the somatosensory cortex slices of a DREADDs injected mouse. Note the selective expression of hM3D–mCherry in astrocytes. Right: Scheme of the experimental approach and representative EPSC traces before (basal) and after CNO (1 mM) application in L5. F) Left: EPSCs amplitude versus time before (basal) and after CNO application in AAV8-GFAP-Gq-DREADD-mCherry injected mice in control (black, close) and in presence of CPT (gray, open), and in AAV8-GFAP-mCherry injected mice (black, open). Blue shadow indicates 5 s CNO application. Right: Relative changes in EPSC amplitude in DREADDs injected mice in control and in presence of CPT, and in mCherry injected mice. Two-tailed Student’s paired t-test. G) Left: Pseudocolor images showing the fluorescence intensities of GCaMP6f-expressing astrocytes before and after CNO application in L5. Top right: Ca²⁺ event probability over time of L5 astrocytes before (basal) and after CNO application in AAV8-GFAP-Gq-DREADD-mCherry injected mice in control (black, close) and in presence of CPT (gray, open), and in AAV8-GFAP-mCherry injected mice (black, open). Blue shadow indicates 5 s CNO application. Bottom right: Relative changes in Ca²⁺ event probability in DREADDs injected mice in control and in presence of CPT, and in mCherry injected mice. Two-tailed Student’s paired t-test. H) Schematic summary depicting the signaling pathways involved in eCBs-induced heteroneuronal synaptic depression. Data are expressed as mean ± SEM, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Moreover, ND-evoked astrocyte Ca²⁺ elevations were largely absent in inositol-1,4,5-trisphosphate (IP3)-receptor type 2-deficient mice (IP₃R₂^−/− mice), in which G protein-mediated Ca²⁺ signal is selectively impaired in astrocytes (Petravicz et al. 2008; Di Castro et al. 2011; Navarrete et al. 2012; Gómez-Gonzalo et al. 2015) (n = 164 astrocytes of 11 slices; Fig. 3A and B). In these mice, the homoneuronal depression was preserved (13 out of 30 cases; 43.3%), but the heteroneuronal depression was absent (0 out of 17 cases; 0%) (Fig. 3C). Collectively, these results indicate that the astrocyte Ca²⁺ signal is required for the heteroneuronal, but not the homoneuronal, synaptic depression.

We then investigated the gliotransmitter responsible for the heteroneuronal depression. ATP and its metabolic product adenosine are known to be released by astrocytes (Araque et al. 2014) and to regulate synaptic transmission in several brain areas (Zhang et al. 2003; Pascual et al. 2005; Serrano et al. 2006; Panatier et al. 2011; Martin-Fernandez et al. 2017; Corkrum et al. 2020). Therefore, we hypothesized that eCB-induced astrocyte Ca²⁺ elevations would stimulate the release of ATP/adenosine that acting on neuronal type 1 adenosine receptors (A1Rs) would depress synaptic transmission. To test this idea, we depolarized L5 pyramidal neurons and monitored the homo- and heteroneuronal synaptic depression before and after bath application of the A1R antagonist CPT (5 μM). While the homoneuronal depression was unaffected (n = 8), the heteroneuronal depression was abolished in the presence of CPT (n = 8; Fig. 3C). Consistent with results observed in L5, heteroneuronal depressions in L4 and L2/3 were also abolished by the A1R antagonist CPT (Fig. 3D).

To further test the astrocyte involvement, we investigated if activation of G-protein-mediated signaling in astrocytes depresses excitatory transmission in the S1 cortex by directly and selectively activating astrocytes using designer receptors exclusively activated by designed drugs (DREADDs). Astrocytes in the S1 cortex were targeted with AAV8-GFAP-Gq-DREADD-mCherry and AAV5-gfaABC1d-GCaMP6f (Fig. 3E). Activation of Gq-DREADDs with the synthetic agonist clozapine-N-oxide (CNO, 1 mM) delivered from a micropipette by pressure pulses (5 s) (Fig. 3E) induced Ca²⁺ elevations in astrocytes (n = 253 astrocytes from 14 slices) and depressed synaptic transmission in L5 pyramidal neurons (n = 11) (Fig. 3F and G), which was associated with an increase in PPR indicating a presynaptic mechanism (n = 11; Fig. S1E). Moreover, in the presence of CPT, CNO also induced Ca²⁺ elevations in astrocytes (n = 87 astrocytes from 7 slices, Fig. 3G) but failed to affect synaptic transmission (n = 7, Fig. 3F). In slices from mice that were injected with control AAV8-GFAP-mCherry virus (i.e. lacking DREADDs), CNO application failed to affect both synaptic transmission (n = 11; Fig. 3F) and astrocyte Ca²⁺ dynamics (n = 92 astrocytes from 8 slices; Fig. 3G). These results suggest that astrocyte Ca²⁺ elevations are sufficient to regulate cortical synaptic transmission.

Taking together, these results indicate that neuron-released eCBs induce homoneuronal depression by directly acting on neuronal CB₁Rs. Concomitantly, eCBs activate CB₁Rs in astrocytes, elevate their intracellular Ca²⁺, and stimulate the release of ATP/Adenosine, which acts on presynaptic A1Rs triggering the heteroneuronal depression (Fig. 3H).

Astrocyte-mediated heteroneuronal depression is column-specific

The functional properties of the somatosensory cortex rely on their modular organization, comprising subcircuits of layer connectivity within cortical columns (Mountcastle 1997; Markram 2008; Harris and Shepherd 2015). Thus, we investigated whether the astrocyte-mediate heteroneuronal depression was also spatially restricted to a single column. We analyzed this phenomenon in the somatosensory barrel cortex, where cortical columns can be easily identified (Fig. 4A). We performed paired whole-cell recordings of L4 neurons located at a similar distance (70–270 μm) but either within the same or in adjacent columns (Fig. 4A). Depolarization of a single L4 neuron to stimulate eCBs release induced heteroneuronal depression in the paired recorded neuron located in the same column (10 out of 29 paired recordings; 35%, Fig. 4B). Consistent with the mechanistic interpretation described above, this phenomenon was blocked after CPT perfusion (Fig. S1H). In contrast to this intracolumn regulation, the heteroneuronal regulation was absent in neurons located at a similar distance but in an adjacent cortical column (0 out 15 paired recordings; 0%, Fig. 4C). In both cases, intra- and intercolumn recordings displayed homoneuronal synaptic depression (16 out 48 cells; 33%, Fig. S1G).

Overall, these results indicate that the eCB-induced astrocyte-mediated heteroneuronal synaptic regulation is column-specific, i.e. it is not a wide unspecific phenomenon but a synaptic regulatory signaling that specifically occurs between cells located within a cortical column.

Astrocyte-mediated heterosynaptic depression is layer-specific

Cortical information processing depends not only on the columnar organization but also on the functional interaction across different layers (Harris and Shepherd 2015). Therefore, we examined the functional organization of heteroneuronal and homoneuronal synaptic depression across different cortical layers, i.e. between neurons located in L2/3, L4, and L5 of the primary S1 (Fig. 5). We performed paired recordings of vertically located (i.e. intracolumnar) neurons in these layers, depolarized 1 neuron to stimulate eCB release, and monitored EPSCs in the other neuron located in another layer.

Fig. 5.

Astrocyte-mediated heterosynaptic depression is layer-specific. A) Representative infrared differential interference contrast image of the experimental configuration with the stimulation pipette in layer 2/3 and a pair of neurons patched one in L4 and the other in L2/3 of the S1 cortex. B) Left: Heteroneuronal EPSCs amplitude versus time in a pair of neurons patched one in L4 and the other in L2/3 before (basal) and after ND of L4 (blue) or L2/3 neuron (open gray). Right: Relative changes in EPSC amplitude in L2/3 (when L4 neuron is stimulated) and L4 (when L2/3 neuron is stimulated) neuron. Two-tailed Student’s paired t-test. C) Representative infrared differential interference contrast image of the experimental configuration with the stimulation pipette in layer 2/3 and a pair of neurons patched one in L5 and the other in L2/3 of the S1 cortex. D) Left: Heteroneuronal EPSCs amplitude versus time in a pair of neurons patched one in L5 and the other in L2/3 before (basal) and after ND of L5 (blue) or L2/3 neuron (open gray). Right: Relative changes in EPSC amplitude in L2/3 (when L5 neuron is stimulated) and L5 (when L2/3 neuron is stimulated) neuron. Two-tailed Student’s paired t-test. E) Representative infrared differential interference contrast image of the experimental configuration with the stimulation pipette in layer 2/3 and a pair of neurons patched one in L4 and the other in L5 of the S1 cortex. F) Left: Heteroneuronal EPSCs amplitude versus time in a pair of neurons patched one in L4 and the other in L5 before (basal) and after ND of L4 (blue) or L5 neuron (open gray). Right: Relative changes in EPSC amplitude in L5 (when L4 neuron is stimulated) and L4 (when L5 neuron is stimulated) neuron. Two-tailed Student’s paired t-test. G) Schematic summary depicting the astrocyte-mediated heterosynaptic regulation pathways into the same layer (intralayer) and between layers (interlayer). Data are expressed as mean ± SEM, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

While depolarizing a single L2/3 neuron did not affect synaptic transmission in L4 neurons (n = 11) (Fig. 5A and B), stimulation of a single L4 neuron induced heteroneuronal synaptic depression in L2/3 neurons (5 out of 10 cases; 50%, Fig. 5A and B). Likewise, stimulation of a single L2/3 neuron did not alter synaptic transmission in L5 neurons (n = 8; Fig. 5C and D), but stimulation of L5 neurons induced heteroneuronal depression in L2/3 neurons (4 out of 10 cases; 40%, Fig. 5C and D). Finally, stimulation of L4 neurons depressed neurotransmission in L5 pyramidal neurons (4 out of 14 cases; 28.1%, Fig. 5E and F), but L5 neuron stimulation did not impact synaptic transmission in L4 neurons (n = 16; Fig. 5E and F). In summary, astrocyte-mediated heteroneuronal depression occurs between neurons located in different layers, but following a specific pattern and not necessarily reciprocally. For example, synapses in L2/3 neurons can be regulated by neurons located in L4 or L5, but not vice versa; and L4 neurons can regulate neurons in L2/3 and L5 but are not regulated by them.

Altogether, these results indicate that eCB-induced astrocyte-mediated heteroneuronal synaptic regulation is not an unspecific phenomenon, rather it is layer-specific, selectively occurring among neurons following a layer-specific pattern (Fig. 5G).

Astrocytic Ca²⁺ responses to eCBs are not homogeneous across cortical layers

We then tested whether the astrocyte Ca²⁺ signals evoked by ND were layer-specific by examining the astrocyte Ca²⁺ signals in different layers in response to eCBs released by depolarization of neurons (Fig. 6). To ensure that the astrocyte activation was due to eCBs, we performed the experiments in the presence of TTX (1 μΜ) and the cocktail of neurotransmitter receptor antagonists (see Methods). ND of L2/3, L4, or L5 neurons elevated intracellular Ca²⁺ in astrocytes located within the same layer (n = 110 astrocytes from 5 slices; n = 110 astrocytes from 4 slices; n = 142 from 6 slices, respectively). These responses were abolished by AM251, confirming to be the result of eCB signaling (n = 89 astrocytes from 5 slices; n = 79 astrocytes from 4 slices; n = 122 from 6 slices, respectively, Fig. 6B–D).

Fig. 6.

Astrocytic Ca²⁺ responses to eCBs are not homogeneous across cortical layers. A) Representative infrared differential interference contrast image and pseudocolor image representing fluorescence intensities of GCaMP6f-expressing astrocytes in the different layers of the primary somatosensory cortex. B–D) left: Ca²⁺ event probability over time before (basal) and after ND (from −70 to 0 mV) in control and in presence of AM251 (2 μM) when patched and recorded in L2/3 (B, orange), L4 (C, purple), or L5 (D, red). Right: Relative changes in Ca²⁺ event probability in control and in presence of AM251 (2 μM) when patched and recorded in L2/3 (B, orange), L4 (C, purple), or L5 (D, red). Two-tailed Student’s paired t-test. E) Left: Ca²⁺ event probability over time of astrocytes of layer 2/3 (orange), 4 (purple), and 5 (red) before (basal) and after L2/3 neuron depolarization. Right: Relative changes in Ca²⁺ event probability of astrocytes of layer 2/3 (orange), 4 (purple), and 5 (red). Two-tailed Student’s paired t-test. F) Left: Ca²⁺ event probability over time of astrocytes of layer 2/3 (orange), 4 (purple), and 5 (red) before (basal) and after L4 neuron depolarization. Right: Relative changes in Ca²⁺ event probability of astrocytes of layer 2/3 (orange), 4 (purple), and 5 (red). Two-tailed Student’s paired t-test. G) Left: Ca²⁺ event probability over time of astrocytes of layer 2/3 (orange), 4 (purple), and 5 (red) before (basal) and after L5 neuron depolarization. Right: Relative changes in Ca²⁺ event probability of astrocytes of layer 2/3 (orange), 4 (purple), and 5 (red). Two-tailed Student’s paired t-test. All experimental conditions were performed in TTX (1 μM) and in a cocktail of neurotransmitter receptor antagonists (see Material and methods). H) Schematic summary depicting the Ca²⁺ responses of astrocytes located in the same (intralayer) or different (interlayer) layers to the endogenous mobilized eCBs from neurons located in the same or different layers. Data are expressed as mean ± SEM, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

We then examined the astrocyte responses to neuron-released eCBs across cortical layers. Stimulation of L2/3 neurons increased the astrocyte Ca²⁺ event probability in L2/3 (n = 185 astrocytes from 10 slices) but failed to increase Ca²⁺ signaling in L4 (n = 62 astrocytes from 5 slices) or L5 (n = 129 astrocytes from 6 slices) astrocytes (Fig. 6E). Likewise, stimulation of L4 neurons elevated Ca²⁺ in L4 (n = 163 astrocytes from 8 slices), L5 (n = 87 astrocytes from 7 slices), and L2/3 (n = 71 astrocytes from 6 slices) astrocytes (Fig. 6F). Finally, stimulation of L5 neurons produced an increase in the Ca²⁺ event probability in L5 (n = 267 astrocytes from 13 slices) and L2/3 (n = 202 astrocytes from 8 slices) astrocytes but not in L4 (n = 144 astrocytes from 8 slices) astrocytes (Fig. 6G). These results indicate that astrocyte Ca²⁺ increases mediated by eCBs signaling obeys a layer-specific pattern in agreement with the astrocyte-mediated heterosynaptic regulation produced by ND (Fig. 6H).

In summary, eCB-mediated neuron-to-astrocyte signaling is a form of communication that occurs between cells that can be located in different layers, but follows a specific connectivity pattern. Like the heteroneuronal depression, this specific pattern does not necessarily involve reciprocal signaling between layers. Notably, the specific neuron-to-astrocyte connectivity pattern mirrors the heteroneuronal depression pattern, suggesting that the selective neuron-astrocyte signaling between layers is responsible for the astrocyte-mediated nonsynaptic communication between neurons in different cortical layers.

Discussion

Present results show that astrocytes regulate cortical synaptic function in a layer- and column-specific manner and that the functional interaction between cortical astrocytes and synapses is highly spatially organized. We show that the activation of astrocytes by endogenous stimuli—eCBs physiologically released from cortical neurons—induced astrocyte Ca²⁺ elevations and transiently depressed synaptic transmission in neurons located in the same and distinct cortical layers. This heteroneuronal synaptic depression requires astrocytic CB₁R activation, is mediated by activation of presynaptic A1 receptors, and can be mimicked by astrocyte-specific chemogenetic stimulation. Additionally, our results also show that neuron-released eCBs can depress synaptic transmission by directly activating CB₁Rs in homoneuronal synapses, a phenomenon known as DSE (Wilson and Nicoll 2001; Chevaleyre and Castillo, 2003; Piomelli 2003; Chevaleyre and Castillo, 2004; Diana and Marty 2004; Castillo et al. 2012).

The neocortex is highly organized in layers and columns with precisely neuronal connectivity. Our results indicate that eCB-mediated astrocyte-neuron signaling is also exquisitely organized. First, the astrocyte-mediated heteroneuronal depression was found to be column-specific because it only occurred between neurons located within the same column and not between neurons located at similar distances but in adjacent columns (Fig. 4). Second, astrocyte-mediated heteroneuronal regulation occurred between neurons located in different layers, but according to a specific connectivity pattern (Fig. 5). Third, the eCB-mediated neuron-to-astrocyte signaling was also layer-specific because astrocytic Ca²⁺ responses to eCBs released by neurons in different cortical layers were not homogeneous across the cortical layers; rather neuron-to-astrocyte signaling occurred according to particular signaling patterns (Fig. 6). Nevertheless, the mechanisms underlying this specific pattern remain unknown. Astrocytes in the somatosensory cortex have been reported to exhibit layer-specific properties regarding morphology, structural interaction with synapses, molecular expression (Lanjakornsiripan et al. 2018; Bayraktar et al. 2020), and Ca²⁺ dynamics (Takata and Hirase 2008). In addition, CB1R expression is not equally expressed throughout all layers of the somatosensory cortex (Egertová et al. 2003; Deshmukh et al. 2007). These differences may account for the layer-specific pattern of the heteroneuronal depression. Further studies are required to fully determine the underlying mechanism.

Several previous studies have shown that astrocytes stimulated by eCBs lead to regulation of synaptic transmission in diverse brain areas, including the hippocampus, amygdala, and neocortex (Navarrete and Araque 2010; Han et al. 2012; Min and Nevian 2012; Gómez-Gonzalo et al. 2015). In the neocortex, synapses onto layer 2/3 neurons undergo spike-timing long-term depression (LTD) mediated by glutamate released from astrocytes (Min and Nevian 2012). In contrast, we found (Figs. 1–3) that eCB-induced astrocyte activation transiently depresses synapses through ATP/adenosine release as gliotransmitters. Different neuronal stimulating paradigms used in these studies may account for such discrepancies. As a matter of fact, astrocytes are competent to release distinct gliotransmitters depending on the pattern of neuronal stimulation as demonstrated in the hippocampus, where astrocytes can release glutamate upon low-frequency stimulation of neighboring interneurons or glutamate and ATP/adenosine upon high-frequency stimulation (Covelo and Araque 2018).

The eCB-induced astrocyte-mediated heteroneuronal depression was found to be restricted within a single cortical column, supporting the idea of a highly organized signaling between astrocytes and neurons at a modular level. These results agree with previous reports showing that astrocyte Ca²⁺ signal is spatially restricted in astrocytes located within the columns of the barrel cortex (Houades et al. 2008; Schipke et al. 2008; Eilam et al. 2016). The column-specific astrocyte-mediated synaptic regulation also indicates that astrocyte-neuron networks are functionally organized obeying the columnar organization of the neuronal connectivity pattern. While present results are focused on the primary somatosensory cortex, it seems likely that it also occurs in other cortical areas, provided that the columnar organization of the cortex is present (Horton and Adams 2005; Markram 2008; da Costa 2010; DeFelipe 2012). However, further studies, out of the scope of the present work, are required to test whether this is the case.

In conclusion, the present data indicate that astrocytes modulate cortical synaptic transmission in a column and layer-specific manner, obeying the structural and functional organization of the cortex, which suggests that they are an integral part of the cortical modules. Moreover, astrocytes, by providing layer-specific signaling pathways of nonsynaptic communication between neurons, add further complexity to the signaling mechanisms underlying cortical network function. This finely controlled that astrocyte–synapse interaction is particularly significant in the neocortex, where the spatial integration of synaptic signals is highly relevant for cortical information processing.

Funding

Postdoctoral fellowship from Basque Government, Spain, to AMB; National Institutes of Health (NIH-MH R01MH119355 to AA and PK; NIH-NINDS R01NS097312-to AA; and NIH-NIDA R01DA048822 to AA); Department of Defense (W911NF2110328) to AA; FEDER and ISCIII (AES2018-PI18/00513); Basque Government (PIBA19-0059 to SM); ARSEP Foundation to SM and GM.

Conflict of interest statement: None declared.

Authors’ contributions

AMB performed experiments and analyzed data. LB and PK performed immunohistochemistry. PK, AA, GM, CM, SM, and AMB conceived the study and wrote the manuscript.