Endocannabinoids (eCBs) play key roles in brain function, acting as modulatory signals in synaptic transmission and plasticity. They are recognized as retrograde messengers that mediate long-term synaptic depression (LTD), but their ability to induce long-term potentiation (LTP) is poorly known. We show that eCBs induce the long-term enhancement of transmitter release at single hippocampal synapses through stimulation of astrocytes when coincident with postsynaptic activity. This LTP requires the coordinated activity of the 3 elements of the tripartite synapse: 1) eCB-evoked astrocyte calcium signal that stimulates glutamate release; 2) postsynaptic nitric oxide production; and 3) activation of protein kinase C and presynaptic group I metabotropic glutamate receptors, whose location at presynaptic sites was confirmed by immunoelectron microscopy. Hence, while eCBs act as retrograde signals to depress homoneuronal synapses, they serve as lateral messengers to induce LTP in distant heteroneuronal synapses through stimulation of astrocytes. Therefore, eCBs can trigger LTP through stimulation of astrocyte–neuron signaling, revealing novel cellular mechanisms of eCB effects on synaptic plasticity.
The endocannabinoid (eCB) system comprises 2 cannabinoid receptors, type 1 (CB1) and type 2 (CB2), and the endogenous ligands, eCB. eCBs retrogradely modulate synaptic transmission widely throughout the central nervous system playing relevant neuromodulatory roles in multiple physiological processes (Freund et al. 2003; Chevaleyre et al. 2006). They are released from postsynaptic neurons, activate presynaptic CB1 cannabinoid receptors, and reduce transmitter release either transiently (endocannabinoid-mediated short-term depression; eCB-STD) or persistently (endocannabinoid-mediated long-term depression; eCB-LTD) (Kreitzer and Regehr 2001a; Alger 2002; Wilson and Nicoll 2002; Freund et al. 2003; Chevaleyre et al. 2006; Heifets and Castillo 2009), and are responsible for most of the behavioral effects of cannabinoids (Maldonado et al. 2006). Recent studies in goldfish (Cachope et al. 2007), lampreys spinal cord (Song et al. 2012), and rodent hippocampal slices (Navarrete and Araque 2010) have reported that CB1R activation can transiently enhance neurotransmission. Although the role of eCB signaling in synaptic plasticity is thought to exclusively induce long-term synaptic depression (LTD), its involvement in the long-term potentiation (LTP) of synaptic transmission is still poorly known (Xu et al. 2012).
Astrocytes have emerged as active elements directly involved in synaptic physiology. They respond with Ca2+ elevations to neurotransmitters released by neurons and modulate neuronal excitability and synaptic transmission by releasing gliotransmitters (Araque et al. 2001; Nedergaard et al. 2003; Volterra and Meldolesi 2005; Haydon and Carmignoto 2006; Perea et al. 2009; Parpura and Zorec 2010; Singh et al. 2014), hence being active elements of synaptic function at tripartite synapses (Araque et al. 1999; Perea et al. 2009). While eCBs released by neurons exert their effects at short distances [<20 µm (Wilson and Nicoll 2001; Piomelli 2003; Chevaleyre and Castillo 2004; Chevaleyre et al. 2006; Navarrete and Araque 2010)], the possible long-range effects of eCBs signaling on synaptic plasticity through astrocyte network stimulation remain largely unknown. The demonstration that hippocampal astrocytes express functional CB1Rs, which increase astrocyte Ca2+ levels and stimulate glutamate release (Navarrete and Araque 2008), suggests that astrocytes are directly involved in the eCB intercellular signaling as well as in its functional consequences on synaptic transmission. Indeed, a more recent study has shown that eCB-mediated neuron-to-astrocyte signaling lead to the short-term potentiation of synaptic transmission in hippocampal synapses (Navarrete and Araque 2010). Furthermore, the eCB-astrocyte signaling has been related with cortical LTD (Min and Nevian 2012) and the impairment of spatial working memory and hippocampal LTD induced by exogenous cannabinoids (Han et al. 2012). However, despite these studies revealing the role of eCB-astrocyte signaling on synaptic depression processes, the consequences and underlying molecular mechanisms of eCB-astrocyte signaling on long-lasting enhancement of synaptic activity remain largely unknown.
Our previous studies have demonstrated that calcium elevations in astrocytes elicited by Ca2+ uncaging or eCBs released from neurons induce a transient synaptic potentiation of hippocampal synaptic transmission (Perea and Araque 2007; Navarrete and Araque 2010), and that pairing astrocyte Ca2+ uncaging and mild postsynaptic depolarization can induce long-term changes in synaptic efficacy. Using electrophysiological and Ca2+ imaging techniques in mouse brain slices, we have investigated whether eCBs released from neurons, acting as endogenous signals that activate astrocytes (Navarrete and Araque 2008), could induce the LTP of synaptic transmission. We have also investigated the underlying signaling pathways and molecular mechanisms responsible for the LTP mediated by the coincidence of astrocytic and postsynaptic activities. Here, we show that the coincidence of eCB signaling and postsynaptic activity leads to the eCB-induced long-term potentiation (eLTP) at single synapses in neurons relatively distant from the eCBs source. This form of LTP results from the coincidence of signaling events through the concerted activity of the tripartite synapse: 1) eCB-evoked astrocyte Ca2+ signal that stimulates glutamate release; 2) postsynaptic nitric oxide (NO) production; and 3) activation of presynaptic metabotropic glutamate receptors (group I mGluRs) and protein kinase C (PKC). Therefore, eCBs induce lateral LTP of synaptic transmission in relatively distant synapses through stimulation of CB1Rs in astrocytes. These results contribute to the growing body of evidence highlighting the role of astrocytes in long-term synaptic modulation.
Materials and Methods
All the procedures for handling and sacrificing animals followed the European Commission guidelines (2010/63/EU).
Hippocampal Slice Preparation
Hippocampal slices were obtained from C57BL/6 mice (13–18 days old). In some cases, slices from CB1 receptor knockout mice and IP3R2 knockout mice, generously donated by Dr A. Zimmer and Dr J. Chen, respectively, were used (Zimmer et al. 1999; Li et al. 2005). Animals were anesthetized and decapitated. The brain was rapidly removed and placed in ice-cold artificial cerebrospinal fluid (ACSF). Slices (350–400 μm thick) were incubated (>30 min) at room temperature (21–24 °C) in ACSF containing (in mm): NaCl 124, KCl 2.69, KH2PO4 1.25, MgSO4 2, NaHCO3 26, CaCl2 2, and glucose 10, and was gassed with 95% O2/5% CO2 (pH = 7.3). Slices were transferred to an immersion recording chamber and superfused at 2 mL/min with gassed ACSF including 0.05 mm Picrotoxin and 5 µm (2S)-3-[[(1S)-1-(3,4-dichlorophenyl )ethyl]amino-2-hydroxypropyl](phenylmethyl)phosphonic acid hydrochloride (CGP 55845) to block GABAA and GABAB receptors, respectively. Cells were visualized under an Olympus BX50WI microscope (Olympus Optical, Tokyo, Japan).
Simultaneous electrophysiological whole-cell patch-clamp recordings from 2 CA1 pyramidal neurons (distance of the somas: 60–150 μm) were made. Patch electrodes had resistances of 3–10 MΩ when filled with the internal solution containing (in mm): K-gluconate 135, KCl 10, HEPES 10, MgCl2 1, ATP-Na2 2 (pH 7.3 adjusted with KOH; osmolality 280–290 mOsm/L). Recordings were obtained with PC-ONE amplifiers (Dagan Instruments, Minneapolis, MN, USA). Membrane potential was held at −70 mV. 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 potential, and stimulus artifact duration did not change >20%. Cells that did not meet these criteria were discarded. Signals were fed to a Pentium-based PC through a DigiData 1440A interface board. Signals were filtered at 1 kHz and acquired at 10-kHz sampling rate. The pCLAMP 10.2 (Axon instruments) software was used for stimulus generation, data display, acquisition, and storage. In some experiments, astrocytes were patched with 4–9 MΩ electrodes filled with an intracellular solution containing (in mm): KMeSO4 100, KCl 50, HEPES 10, ATP-Na2 2 (pH 7.3 adjusted with KOH; osmolality 280–290 mOsm/L). Since the synaptic potentiation induced in experiments performed at room temperature (21–24 °C) and at 30–32 °C was not significantly different (see Fig. 1E), experiments were performed at room temperature, unless stated otherwise.
Theta capillaries filled with ACSF were used for bipolar minimal stimulation. Electrodes were connected to a stimulator S-910 through an isolation unit and placed in the “stratum radiatum” to stimulate Schaffer collaterals (SCs). Single pulses (250-μs duration) or paired pulses (50-ms interval) were continuously delivered at 0.33 Hz. Excitatory postsynaptic currents (EPSCs) were simultaneously recorded from the eCB-releasing neuron (homoneuronal synapses) and from the postsynaptic neuron (heteroneuronal synapses). Stimulus intensity (0.1–10 mA) was adjusted to meet the conditions that putatively stimulate single or very few presynaptic fibers [cf. (Raastad 1995; Isaac et al. 1996; Perea and Araque 2007; Navarrete and Araque 2010)]. Recordings that did not meet these criteria and did not show amplitude stability of basal EPSCs were rejected. The synaptic parameters analyzed were: synaptic efficacy (mean peak amplitude of all responses including failures); synaptic potency (mean peak amplitude of the successes); probability of release (Pr, ratio between number of successes vs. total number of stimuli); and paired-pulse facilitation ratio (PPF = 2nd EPSC/1st EPSC). Synaptic plasticity changes were determined from 180 stimuli before (basal) and 60 min after stimulus (i.e., depolarization of the eCB-releasing neuron or astrocyte Ca2+ uncaging). To illustrate the time course, synaptic parameters were grouped in 3-min bins. EPSC amplitude was determined as the peak current amplitude (2–9 ms after stimulus) minus the mean baseline current (20–30 ms before stimulus). A response was considered a failure if the amplitude of the current was <3 times the standard deviation of the baseline current, and was verified by visual inspection.
All the experiments were performed in an unbiased form and a blind design whenever possible. The experimental paradigm used was based on single-cell stimulation and single-synapse recording, which presents the limitation that some specific recorded synapses may not be under the direct influence of the stimulated cell (pyramidal neuron or astrocyte; cf. Perea and Araque 2007; Navarrete and Araque 2010). Consistent with this idea, although a subset of the recorded single synapses (8 of 12; i.e., around 66.6%) showed eLTP, in the present study synapses were not selected according to their susceptibility to modulation, and data correspond to average values of all the recorded synapses in each condition. Consequently, while the average values underestimate the actual magnitude of the modulatory phenomena at a single synapse, they validate the relevance of the phenomenon on the overall synaptic activity.
The following “paired depolarizing protocol” was used: The termed postsynaptic neuron was depolarized to −30 mV for 3 min to mimic postsynaptic activity. After the first minute, the termed eCB-releasing neuron was depolarized to 0 mV for 5 s to stimulate eCB release. SC stimulation was interrupted during the paired depolarizing protocol, and resumed afterward. When indicated, theta burst stimulation (TBS) stimulation [4 trains at 5 Hz of 4 stimuli (at 40 Hz) delivered 10 times at 0.1 Hz] applied by an independent glass pipette to independent SC pathways was used to synaptically induce postsynaptic depolarization recorded in current-clamp mode, and 50 s after initiating the TBS paradigm the eCB-releasing neuron was depolarized to 0 mV for 5 s. The independence of the SC pathways stimulated by single synapse and conventional TBS stimuli was confirmed by stimulating both pathways simultaneously with time intervals of 50 ms and observing no cross-facilitation of the corresponding EPSCs.
The slow Ca2+-activated K+ current (sIAHP) was evoked by 200-ms depolarizing pulses (from −60 to 20 mV). The sIAHP magnitude was quantified from the area under the current trace, measured 60–100 ms after the end of the pulse (Martin et al. 2001).
Ca2+ levels in astrocytes located in the stratum radiatum of the CA1 region of the hippocampus were monitored by fluorescence microscopy using the Ca2+ indicator fluo-4 (Molecular Probes, Eugene, OR, USA). Ca2+ signal was monitored from cells located within a 150-µm wide region around the stimulated neuron and perpendicular to the “stratum pyramidale.” Slices were incubated with fluo-4 AM (2–5 μL of 2 mm dye were dropped over the hippocampus, attaining a final concentration of 2–10 μm and 0.01% of pluronic) for 20–30 min at room temperature. In these conditions, most of the cells loaded were astrocytes (Kang et al. 1998; Parri et al. 2001; Araque et al. 2002; Nett et al. 2002; Perea and Araque 2005) as confirmed in some cases by their electrophysiological properties. Astrocytes were imaged using a CCD camera (Retiga EX; Qimaging, Canada). Cells were illuminated during 200–500 ms with a xenon lamp at 490 nm using a monochromator Polychrome V (T.I.L.L. Photonics, Planegg, Germany), and images were acquired every 1 s. The monochromator Polychrome V and the CCD camera were controlled and synchronized by the IP Lab software (BD Biosciences, MD, USA) that was also used for quantitative epifluorescence measurements. Ca2+ variations, termed Ca2+ spikes, were recorded at the soma of the cells and estimated as changes of the fluorescence signal over baseline (ΔF/F0), and cells were considered to respond to the stimulation when ΔF/F0 increased 3 times the standard deviation of the baseline for at least 2 consecutive images and with a delay ≤15 s after the stimulation. Neuronal Ca2+ variations were recorded with patch pipettes filled with the internal solution containing 50 µm fluo-4 to monitor Ca2+ levels, and estimated by same criteria as for astrocyte Ca2+ signals.
Astrocyte Ca2+ responses were quantified analyzing the probability of occurrence of Ca2+ spikes; this is, detecting the onset of Ca2+ elevations (Ca2+ spikes) during the recording period and grouped in 5-s bins from 6 to 18 astrocytes in the field of view. Values of 0 and 1 were assigned for bins showing no response and Ca2+ spike, respectively, to obtain the Ca2+ spike probability index. To test the effects on Ca2+ spike probability (Pr) under different conditions, the respective mean basal (15 s before the paired depolarizing protocol) and maximum Ca2+ spike probability (15 s after the paired depolarizing protocol) from different slices were averaged and compared. Slices from at least 4 independent animals per condition were used.
Ca2+ Uncaging by UV-flash Photolysis
In photostimulation experiments, single astrocytes were recorded with patch pipettes filled with the internal solution containing 5 mm NP-EGTA (and 50 µm fluo-4 to monitor Ca2+ levels). Ca2+ uncaging was achieved by delivering train pulses (1-ms duration, 6–15 mW) of UV light (340–380 nm) at 2 Hz during 10 s to the soma and processes of the recorded astrocyte (optical window of 15–25 µm diameter), inducing reliable Ca2+ elevations, using a flash photolysis system (Rapp OptoElectronic, Hamburg, Germany). Control experiments were performed by delivering UV photostimulating astrocytes not loaded with NP-EGTA. No astrocyte Ca2+ signal or synaptic transmission parameter changes were observed (cf. Perea and Araque 2007; Navarrete et al. 2012).
Immunohistochemical reactions for electron microscopy were performed using the pre-embedding immunogold method (Lujan et al. 1996). C57BL/6 mice (18 days old) were anesthetized and transcardially perfused with ice-cold fixative containing 4% paraformaldehyde, 0.05% glutaraldehyde, and 15% (v/v) saturated picric acid made up in 0.1 M phosphate buffer (PB, pH 7.4). After perfusion, brains were removed, washed thoroughly in 0.1 M PB, and cut on a Vibratome (Leica V1000) obtaining coronal 60-μm-thick sections. Free-floating sections were incubated in 10% (v/v) NGS diluted in Tris-buffered saline solution. Sections were then incubated with pan-mGluR1 antibodies at a final protein concentration of 3–5 μg/mL diluted in Tris-buffered saline solution containing 1% (v/v) NGS. After several washes in Tris-buffered saline solution, sections were incubated with goat anti-mouse IgG coupled to 1.4-nm gold (Nanoprobes, Inc., Stony Brook, NY, USA). Sections were postfixed in 1% (v/v) glutaraldehyde and washed in double distilled water, followed by silver enhancement of the gold particles with a HQ Silver kit (Nanoprobes, Inc.). Sections were then treated with osmium tetraoxide (1% in 0.1 M PB), block-stained with uranyl acetate, dehydrated in graded series of ethanol, and flat-embedded on glass slides in Durcupan (Fluka) resin. Regions of interest were cut at 70–90 nm on an ultramicrotome (Reichert Ultracut E, Leica, Austria) and collected single slot pioloform-coated copper grids. Ultrastructural analyses were performed in a Jeol-1010 electron microscope (Jeol, Tokio, Japan). The rabbit polyclonal antibody mGlu1 recognize an extracellular epitope that is present in all isoforms of mGluR1. The specificity of this antibody has been described previously (Lujan et al. 1996), and further characterization on the presynaptic labeling is presented here (Fig. 6G) using mGluR1b rescue mice (generously donated by Dr Atsu Aiba), in which the mGluR1b isoform is rescue in Purkinje cells, but is basically an mGluR1 KO mice in the rest of the brain (Kishimoto et al. 2002).
To establish the relative frequency of mGluR1 immunoreactivity in the stratum radiatum of the CA1 region, we used 60-µm coronal slices processed for pre-embedding immunogold immunohistochemistry. The procedures were similar to those used previously (Lujan et al. 1996). Briefly, for each of 3 young adult animals, 3 samples of tissue were obtained for preparation of embedding blocks, totaling n = 9 blocks. To minimize false negatives, electron microscopic serial ultrathin sections were cut close to the surface of each block, as immunoreactivity decreased with depth. We estimated the quality of immunolabeling by always selecting areas with optimal gold labeling at approximately the same distance from the cutting surface, which was defined within 5–10 µm from the surface. Randomly selected areas were then photographed from the selected ultrathin sections and printed with a final magnification of ×50 000. Quantification of immunogold labeling was carried out in reference areas totaling ∼1000 µm2. Immunoparticles identified in each reference area and present in different subcellular compartments (dendritic spines, dendritic shafts, and axon terminals) were counted. Data were expressed as percentage of immunoparticles in each subcellular compartment. We also counted the number of excitatory synapses labeled and unlabeled for mGluR1, both at postsynaptic and presynaptic sites.
Drugs and Chemicals
N-(Piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (AM-251); 2-methyl-6-(phenylethynyl)pyridine hydrochloride (MPEP), (S)-(+)-α-amino-4-carboxy-2-methylbenzeneacetic acid (LY367385), CGP 55845, and bisindolylmaleimide 1 (BIS I) were purchased from Tocris Cookson (Bristol, UK); fluo-4 AM and o-nitrophenyl EGTA, tetrapotassium salt (NP-EGTA) from Molecular Probes; and (S)-3,5-Dihydroxyphenylglycine (DHPG) from Ascent Scientific (Bristol, UK). All other drugs were from Sigma.
Data are expressed as mean ± standard error of the mean (SEM). Normality test (Shapiro–Wilk test) was applied to the data before running statistical tests. Based on that, data were analyzed using parametric (Student's t-test, α = 0.05) and nonparametric tests (Wilcoxon signed-rank test) as appropriate. Statistical differences were established with P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).
Endocannabinoid Signaling Induces LTP of Transmitter Release in Heteroneuronal Synapses
To investigate whether eCB signaling could induce LTP of synaptic plasticity through astrocyte network activation, we recorded pairs of hippocampal CA1 pyramidal neurons (>60 µm apart), recorded astrocyte Ca2+ levels, and monitored SC-evoked synaptic transmission at single synapses (using the minimal stimulation technique that activates single or very few presynaptic fibers; see Materials and Methods; Fig. 1A,B). We used the following paired depolarizing protocol: We depolarized one neuron (termed eCB-releasing neuron, and whose synapses are called homoneuronal synapses) to 0 mV for 5 s to stimulate eCB release (Kreitzer and Regehr 2001a; Wilson and Nicoll 2001; Ohno-Shosaku et al. 2002; Chevaleyre and Castillo 2004; Navarrete and Araque 2008, 2010) and monitored synaptic activity in the other neuron (termed postsynaptic neuron, and whose synapses are called heteroneuronal synapses), which was mildly depolarized to −30 mV to mimic postsynaptic activity (cf. Perea and Araque 2007). We focused our study on heteroneuronal synapses, because homoneuronal synapses onto the eCB-releasing neuron are known to undergo CB1R-mediated synaptic depression (see Fig. 3A,B; cf. Kreitzer and Regehr 2001a; Ohno-Shosaku et al. 2001; Chevaleyre et al. 2006; Navarrete and Araque 2010). The paired depolarizing protocol induced a long-lasting enhancement of synaptic transmission in 8 of 12 heteroneuronal synapses. The synaptic efficacy increased from −4.8 ± 0.6 to −7.8 ± 1.7 pA (n = 12; P = 0.016) which resulted from the increase of the probability of release (Pr; from 0.50 ± 0.05 to 0.69 ± 0.06; n = 12; P = 0.021) without changes in the synaptic potency (from −11.1 ± 0.9 to −11.7 ± 1.5 pA; n = 12; P = 0.719) (Fig. 1D,E). These changes, which were similarly observed at room temperature and more physiological temperatures (30–32 °C), were associated with a decrease in the PPF (from 1.52 ± 0.11 to 1.27 ± 0.16; n = 8; P = 0.015; Fig. 1C), indicating a presynaptic mechanism of action, which is also consistent with changes in Pr but not in synaptic potency.
These long-term synaptic changes were absent when either the mild depolarization of the postsynaptic neuron or the depolarization of the eCB-releasing neuron were delivered alone (n = 6 and 8, respectively; Fig. 1E; cf. Perea and Araque 2007; Navarrete and Araque 2010) and they were not prevented in the presence of the NMDAR antagonist AP5 (50 µm) (n = 10; Fig. 1E), indicating that NMDARs were not involved in this form of LTP. They were also absent when the paired depolarizing protocol was applied in the presence of the CB1R antagonist AM251 (2 μm) or in transgenic mice lacking CB1Rs (n = 5 and 6, respectively) (Fig. 1D,E), indicating that these changes were mediated by the coincidence of eCBs and postsynaptic signaling. Taken together, these results indicate that eCBs released from the eCB-releasing neuron can trigger the long-term potentiation (eLTP) of transmitter release in heteroneuronal synapses of relatively distant neurons when it coincides with postsynaptic activity.
To test whether eLTP could be triggered by a more physiological stimulus, instead of artificially depolarizing the postsynaptic neuron to −30 mV, we stimulated independent SC pathways using a theta burst paradigm (4 trains at 5 Hz of 4 stimuli at 40 Hz were delivered 10 times at 0.1 Hz). TBS, delivered at intensities that depolarized the postsynaptic neuron below the action potential threshold, did not evoke synaptic potentiation per se, but it induced lateral eLTP when paired with the depolarization of the eCB-releasing neuron (n = 6; Fig. 2B–D). These results indicate that lateral eLTP can be evoked under more physiological conditions.
Astrocyte Signaling: eLTP Requires Astrocyte Ca2+ Elevations
Because eCBs can modulate neurotransmission through stimulation of CB1R-mediated astrocyte Ca2+ elevations (Navarrete and Araque 2010; Min and Nevian 2012), we tested the involvement of the astrocyte Ca2+ signal in the generation of eLTP. We used hippocampal slices from transgenic mice lacking the inositol-1,4,5-trisphosphate(IP3)-receptor type 2 (IP3R2−/− mice) (Li et al. 2005), which is the primary functional IP3 receptor expressed by astrocytes that mediates G protein-mediated intracellular Ca2+ mobilization (Petravicz et al. 2008; Di Castro et al. 2011; Navarrete et al. 2012). First, we assessed whether the eCB release evoked by neuronal depolarization was affected in these mice by monitoring the depolarization-induced suppression of excitation (DSE) in homoneuronal synapses, a well-known phenomenon mediated by presynaptic CB1Rs (Kreitzer and Regehr 2001a; Ohno-Shosaku et al. 2002; Chevaleyre et al. 2006; Navarrete and Araque 2010). The amplitude of the DSE was not significantly different in wild-type and IP3R2−/− mice (n = 6 and 6, respectively, Fig. 3A,B), indicating that the eCB release machinery and neuronal CB1R activation is preserved in these knockout animals. In contrast, while the neuronal depolarization elevated astrocyte Ca2+ in slices from wild-type mice (47 of 74 astrocytes from 8 slices) (cf. Navarrete and Araque 2008, 2010), it failed to elevate astrocyte Ca2+ in IP3R2−/− mice (only 1 of 55 astrocytes from 8 slices displayed Ca2+ elevations) (Fig. 3C–E). Accordingly, the eLTP observed in slices from wild-type mice was largely suppressed in IP3R2−/− mice (n = 8; Fig. 3F–H). Therefore, these results indicate that eLTP requires astrocyte Ca2+ elevations.
To further confirm the involvement of astrocyte Ca2+ signal on eLTP, we stimulated astrocytes downstream the eCB signaling, by loading astrocytes with the Ca2+ cage NP-EGTA (5 mm), which upon UV light stimulation reliably elevated astrocyte Ca2+ in slices from both wild-type and IP3R2−/− mice (Fig. 4A,B). Pairing the mild postsynaptic depolarization (to −30 mV) and the astrocyte Ca2+ uncaging resulted in the LTP of transmitter release in slices from wild-type and IP3R2−/− mice (n = 6 and 6, respectively, Fig. 4C,D; cf. Perea and Araque 2007). These results indicate that eCB effects can be mimicked by astrocyte Ca2+ uncaging and that the signaling pathways downstream the astrocyte Ca2+ signal involved in this form of LTP are unaffected in IP3R2−/− mice. Therefore, when coincident with the postsynaptic activity, the astrocyte Ca2+ signal is necessary to induce LTP.
Postsynaptic Signaling: eLTP Requires NO Production in the Postsynaptic Neuron
Because eLTP is expressed presynaptically as an enhancement of transmitter release probability and requires the postsynaptic depolarization, we aimed to identify the retrograde signal that being released from the postsynaptic cell impact the presynaptic function. A likely candidate is NO because it is an intercellular messenger that may act retrogradely and is involved in synaptic plasticity (Schuman and Madison 1991, 1994; Hawkins et al. 1998). When slices were incubated with the NO scavenger PTIO (100 μm; Fig. 5A), the eCB-induced astrocyte Ca2+ signal was unaffected (n = 98 astrocytes from n = 7 slices: Fig. 5B,E), but the eLTP was absent (n = 6: Fig. 5C,D), which indicates the involvement of NO downstream the astrocyte calcium signal. Because NO could be produced by both pyramidal neurons and astrocytes, to elucidate its source we included 100 μm L-NAME (an antagonist of the NO synthase that prevents NO production) in the recording pipette of the postsynaptic neuron and astrocytes to specifically prevent NO production in the postsynaptic cell and astrocyte local network, respectively (Fig. 5A,D). When L-NAME was included in the postsynaptic cell (L-NAME_N) the astrocyte Ca2+ signal was unaffected (n = 58 astrocytes from n = 5 slices; Fig. 5B,E), but the paired depolarizing protocol did not induce long-term synaptic changes (n = 5; Fig. 5C,D). However, eLTP induced by pairing protocol was unaffected by previous loading of L-NAME into the astrocyte network (L-NAME_A; synaptic efficacy, Pr, and synaptic potency were 174.16 ± 32.1%, 157.16 ± 20.5%, and 108.9 ± 11.9%, from baseline; n = 11; P = 0.043, 0.018, and 0.47, respectively; Fig. 5D), indicating that NO synthesis from astrocytes does not contribute significantly to this form of synaptic plasticity. Taken together, these results indicate that in addition to the astrocyte Ca2+ signal, NO production from postsynaptic cell during neuronal activity is required to generate eLTP.
To test whether NO was sufficient to induce eLTP without the involvement of additional postsynaptic signals, we directly supplied NO in the absence of postsynaptic depolarization, by pressure delivery (180 s) of the NO donor diethylamine NONOate (100 μm) from a micropipette (Fig. 5A). While NO donor application alone did not change synaptic transmission (synaptic efficacy, Pr, and synaptic potency were 97.7 ± 9.6%, 102.9 ± 7.0%, and 99.8 ± 11.5%, from baseline; n = 8; P = 0.581, 0.989, and 0.689, respectively), pairing the NO delivery with the depolarization of the eCB-releasing neuron resulted in the generation of LTP in heteroneuronal synapses (n = 9; Fig. 5C,D). Taken together, these results confirm that NO produced by the postsynaptic neuron is the retrograde messenger involved in the eLTP.
We further investigated whether the co-release of NO and eCB by a single neuron could induce eLTP of synapses into that neuron (homosynapses). We then imposed the paired depolarizing protocol to the same neuron, that is the mild depolarization to −30 mV and to 0 mV for 5 s; but no changes in synaptic transmission were found in these neurons (n = 9; Fig. 5D). These data indicate that eLTP was independent of the spatial coincidence of eCB and NO signaling at the same synapses but it did depend on the temporal coincidence of both signals through astrocyte network activity.
Presynaptic Signaling: eLTP is Mediated by Activation of mGluRs and PKC in Presynaptic Terminals
After the identification of the retrograde messenger originated from the postsynaptic neuron, we sought the presynaptic molecular mechanisms responsible for the eLTP. Because eLTP is mediated by eCB-induced stimulation of astrocytes, which release glutamate that regulates synaptic transmission through activation of mGluRs (Perea and Araque 2007; Navarrete and Araque 2010), we first investigated whether these receptors were involved in the eLTP. While in control conditions, the eLTP was induced by the paired depolarizing protocol (see Fig. 1), it was prevented in parallel experiments when slices were incubated with the antagonists of group I mGluRs MPEP (50 μm) and LY367385 (100 μm; mGluR5 and mGluR1 antagonists, respectively) (n = 8, synaptic efficacy change from basal values: 95.9 ± 15.9%; P = 0.592), without affecting the eCB-induced astrocyte Ca2+ signal (Ca2+ spike probability change from 0.04 ± 0.01 to 0.37 ± 0.08; n = 53 astrocytes from n = 5 slices, P = 0.010: cf. Navarrete and Araque 2010). In a subset of experiments without postsynaptic depolarization, DEA NONOate was applied in the presence of MPEP (50 µm) and LY367385 (100 µm) and no significant changes in synaptic efficacy were observed (111.7 ± 20.0% from basal values; n = 11; P = 0.588), suggesting that NO was not sufficient to induce eLTP, which also required mGluR1/5 receptor activation (Fig. 5F–H). We next investigated the subtype of mGluR involved by perfusing separately the group I mGluR antagonists (Fig. 6). While LY367385 abolished the eLTP (Fig. 6A–C) without modifying the astrocyte Ca2+ signal (Fig. 6D,E), MPEP did not prevent eLTP (Fig. 6B,C), indicating that eLTP was specifically mediated by activation of mGluR1.
The changes in Pr and PPF indicate that eLTP is mediated by activation of presynaptic mGluRs. To test the possible contribution of postsynaptic mGluRs to the eLTP, we filled the postsynaptic neuron with Guanosine 5′-[β-thio]diphosphate (GDPβS, 2mM) to prevent mGluR-mediated signaling, as confirmed by the absence of calcium elevations evoked by local application of DHPG (2 mm; specific agonist of group I mGluRs) or acetylcholine (1 mm) (18 of 18 control neurons and 3 of 20 GDPβS-filled neurons showed calcium elevations) (Fig. 7A–C). In these conditions, paired stimulation still induced the potentiation of synaptic transmission (Fig. 7D–F). Additionally, in the presence of CB1 receptor antagonist AM251 and postsynaptic GDPβS, local application of DHPG, paired with postsynaptic depolarization to −30 mV was sufficient to induce eLTP (Fig. 7G–I). Although mGluR1/5 are largely located postsynaptically, present results indicate that eLTP was mediated by presynaptic rather than postsynaptic activation of these receptors.
However, no anatomical evidence exists for the presence of mGluRs at SC. We therefore performed immunoelectron microscopy studies using a pan-mGluR1 antibody, which recognizes an extracellular epitope common to all mGluR1 isoforms, and the pre-embedding immunogold method (Lujan et al. 1996), which allows the unambiguous localization at the plasma membrane of those subcellular compartments expressing the receptor. In a representative sample of asymmetrical synapses with CA1 pyramidal cells in the stratum radiatum (302 synapses from 3 animals), mGluR1 immunoparticles were found not only postsynaptically (57 synapses), but also at presynaptic sites (27 synapses) (Fig. 6F,G). Indeed, mGluR1 immunoparticles were found at both sites in a subset of synapses (11 of 27 synapses). These results provide anatomical evidence of the presence of mGluR1 in SC, confirming synaptic physiology data that indicate that eLTP results from activation of presynaptic mGluRs.
We then investigated the molecular target responsible for the long-term expression of the enhancement of synaptic transmitter release evoked by mGluR activation. Stimulation of mGluRs activates PKC, which is related with synaptic plasticity (for a review see Anwyl (2009)). Therefore, we tested different PKC inhibitors perfused in the bath or applied intracellularly in the postsynaptic neuron. The eLTP was prevented when the paired depolarizing protocol was delivered in the presence of 2 µm BIS I, a diffusible PKC inhibitor (Toullec et al. 1991; Daw et al. 2000) that can act pre- or postsynaptically (n = 4, Fig. 8A,B), and that did not modify the eCB-induced astrocyte Ca2+ elevations (n = 58 astrocytes from n = 6 slices) (Fig. 8C). In contrast, eLTP was still present when the PKC inhibitor chelerythrine (1 µm; (Ling et al. 2002; Kwon and Castillo 2008) was included in the postsynaptic neuron through the recording pipette (n = 9, Fig. 8A,B). The effectiveness of chelerythrine was assessed in parallel experiments by monitoring the slow Ca2+-activated K+ current (sIAHP), which is sensitive to PKC activation (Engisch et al. 1996; Dutar and Nicoll 1988; Seroussi et al. 2002). We performed paired recordings with one neuron filled with control solution and the other neuron filled with the intracellular solution containing 1-μm chelerythrine (Fig. 8D). While bath application of the PKC activator PMA (phorbol 12-myristate 13-acetate, 15 μm) (Engisch et al. 1996) inhibited the sIAHP in control neurons; the sIAHP was unaffected in the chelerythrine-filled neurons (n = 4; Fig. 8E,F). Taken together, these results indicate that activation of PKC and mGluRs, located presynaptically rather than postsynaptically, mediate the long-term enhancement of presynaptic transmitter release.
Present results show that eCBs released from pyramidal neurons, in addition to their well-known effects suppressing synaptic activity, can induce long-term enhancement of synaptic transmission through activation of astrocytes. eCB-astrocyte signaling potentiates the probability of transmitter release in heteroneuronal synapses when the eCB-evoked astrocyte Ca2+ signal coincides with postsynaptic activity. In addition, these data show the transmitters and molecular pathways involved in this eLTP, which results from the concerted activity of the 3 elements of the tripartite synapse, that is, the eCB-evoked astrocyte Ca2+ signal that stimulates the gliotransmitter release of glutamate, the NO production from the postsynaptic neuron, and the activation of presynaptic mGluRs and PKC signaling (Fig. 8G).
The effects of eCB signaling on synaptic transmission and plasticity has been studied in different brain areas, where it has been found to induce inhibition of neurotransmitter release through activation of presynaptic CB1Rs, which leads to different forms of short-term and long-term depression, such as DSI, DSE, or LTD (e.g., Kreitzer and Regehr 2001a; Alger 2002; Wilson and Nicoll 2002; Freund et al. 2003; Chevaleyre et al. 2006; Heifets and Castillo 2009; Navarrete and Araque 2010). Present results demonstrate that eCBs can trigger LTP of synaptic transmission in the hippocampus through stimulation of astrocytes. They reveal that eCBs, which are largely considered a retrograde signal that evokes synaptic depression, may also serve as a lateral signal through activation of astrocytes that lead to synaptic transmission enhancement in synapses relatively distant from the eCB source, that is, heteroneuronal synapses. Therefore, present findings indicate that eCBs can exert additional effects to classical synaptic depression, which is also supported by recent reports in mice and rats (Navarrete and Araque 2010; Xu et al. 2012; Tamagnini et al. 2013), goldfish Mauthner cell synapses (Cachope et al. 2007), and synapses in the spinal locomotor network of the lamprey (Song et al. 2012). The fact that eCBs may have opposite effects on neurotransmission reveals a higher complexity of eCB signaling on synaptic transmission regulation than previously thought. They may have relevant implications in network function because the opposite effects of eCBs on synaptic transmission are spatially defined and depend on the target cells. Indeed, while eCBs are known to lead to transient or LTD through activation of presynaptic CB1Rs in homoneuronal synapses close to the eCB source (<20 µm) (Wilson and Nicoll 2001; Chevaleyre and Castillo 2004; Chevaleyre et al. 2006; Navarrete and Araque 2010), they transiently (Navarrete and Araque 2010) or persistently (present results) potentiate synaptic transmission at more distant synapses through activation of astrocytes. Homoneuronal synapses may be influenced by the neuronal NO release and the eCB-induced astrocytic glutamate signaling triggered by the neuronal depolarization; however, these synapses exhibit synaptic depression induced by eCBs released from the eCB-releasing neuron (see Fig. 3A,B; cf. Kreitzer and Regehr 2001b; Ohno-Shosaku et al. 2001; Chevaleyre et al. 2006; Navarrete and Araque 2010). In contrast, heteroneuronal synapses, which are more distant from the short-range eCB signaling direct effect, display the persistent synaptic potentiation due to the coordinated activity of astrocytic mGluR and neuronal NO signaling pathways. Thus, the final outcome of this complex and spatially controlled signaling is that astrocytes selectively “feed forward” to the adjacent neurons, but do not “feedback” to the eCB-releasing neuron, because the direct activation of presynaptic CB1Rs by eCBs leads to synaptic depression, which overpowers the mGluR-mediated signaling that would lead to eLTP (cf. (Navarrete and Araque 2010). Consequently, eCBs released from active neurons depress neurotransmission in close synapses; while simultaneously lead to lateral potentiation of more distant synapses, generating a core/shell spatial arrangement of synaptic depression/potentiation.
We show that eCB-induced LTP necessarily requires the involvement of astrocytes, which play an essential role by responding to eCBs with Ca2+ elevations and stimulating glutamate release, thus expanding the original signal to more distant points and transforming it into a potentiating signal. While the role of astrocyte Ca2+ and gliotransmission has been questioned by recent reports based on negative results obtained using transgenic mice (Fiacco et al. 2007; Petravicz et al. 2008; Agulhon et al. 2010), including IP3R2−/− mice that in the present study showed an impairment of the eLTP, our data support a high physiological relevance of these phenomena on synaptic function (e.g., Haydon and Carmignoto 2006; Jourdain et al. 2007; Perea and Araque 2007; Di Castro et al. 2011; Panatier et al. 2011; Navarrete et al. 2012). Therefore, present results further support the tripartite synapse concept, showing that astrocytes are actively involved in synaptic transmission and plasticity, exerting relevant roles in synaptic function, being able to spatially extend intercellular signals and to transform the sign of the effects of intercellular messengers. Furthermore, present study provides further evidence to the accumulating data showing the contribution of astrocytes to synaptic plasticity (Martineau et al. 2006; Panatier et al. 2006; Perea and Araque 2007; Henneberger et al. 2010; Takata et al. 2011; Min and Nevian 2012; Navarrete et al. 2012), indicating that astrocytes are actively involved in storage information in the brain.
The coincidence of the astrocyte Ca2+ signal evoked by Ca2+ uncaging and postsynaptic activity has been previously shown to lead to the generation of hippocampal LTP (Perea and Araque 2007). Our present results indicate that the astrocyte-mediated LTP can be triggered by an endogenous stimulus (i.e., eCBs released form neurons) that physiologically elevates Ca2+ in astrocytes. Cholinergic-evoked Ca2+ elevations in astrocytes have been reported to mediate cholinergic-induced hippocampal LTP (Navarrete et al. 2012). Because astrocytes express a wide variety of G protein-coupled receptors that can mobilize intracellular calcium, whether different neurotransmitters may trigger similar mechanisms for LTP may have a strong impact on our current understanding of the signaling mechanisms that can lead to LTP. Additionally, present results identified the molecular mechanisms and signaling pathways involved in the astrocyte-mediated LTP. We show that astrocyte CB1Rs activated by eCBs elevate Ca2+ and stimulate glutamate release that acting on type I mGluRs persistently enhance synaptic transmitter release when it coincides with a postsynaptic signal (i.e., NO), through activation of presynaptic PKC. mGluR-dependent LTP is known to occur in different areas of the brain including the neocortex, hippocampus, striatum, and nucleus accumbens (for a review see Anwyl (2009)). Present results are consistent with the mechanisms proposed to mediate mGluR-mediated LTP, but our data further reveal that astrocytes are the source of glutamate responsible for the generation of eLTP.
The eCB-induced LTP requires activation of group I mGluRs by glutamate released from astrocytes after being stimulated by eCBs. The subcellular localization of these mGluRs is not fully elucidated. Although these mGluRs can be located at postsynaptic sites (Lujan et al. 1996), physiological data indicate that these mGluRs are located presynaptically because: 1) eLTP is characterized by the enhancement of transmitter release probability without changes in synaptic potency, which indicates a presynaptic rather than a postsynaptic underlying mechanism (Fig. 1E); and 2) eLTP is associated with changes in the PPF (Fig. 1C), which is consistent with a presynaptic mechanism. Additional electrophysiological studies also suggest the existence of presynaptic mGluRs in SC, where they can regulate synaptic transmission (e.g., Rodriguez-Moreno et al. 1998; Fiacco and McCarthy 2004; Sanchez-Prieto et al. 2004; Perea and Araque 2007; Navarrete and Araque 2010; Bonansco et al. 2011; Navarrete et al. 2012). Besides this compelling physiological evidence, we show for the first time, to our knowledge, morphological evidence at the ultrastructural level of the presence of group I mGluRs at SC presynaptic terminals. Furthermore, because mGluR1a is located exclusively at postsynaptic sites in asymmetrical synapses between CA1 pyramidal cells and SC (Lujan et al. 1996), the presynaptic localization of mGluR1 is likely due to the presence of the 1b isoform. Interestingly, because synaptic potentiation has been described to occur only when these receptors are activated by exogenous glutamate or glutamate released from astrocytes, perhaps they are exclusively activated by gliotransmitters and therefore are specific molecular targets of astrocyte–neuron signaling.
In summary, we present evidence indicating that eCBs can induce LTP through the necessary involvement of astrocyte activity and gliotransmitter release, and revealing the different signaling pathways at cellular and molecular level underlying this eLTP, such as the interaction between pyramidal cells and astrocytes, and the eCBs, mGluR, NO, and PKC signaling. While eCBs are considered retrograde signals that depress neurotransmission, we show that they lead to a more distant regulation of synaptic activity inducing a lateral potentiation through activation of astrocytes. This eCB-induced LTP requires the coordinated activity of the 3 elements of the tripartite synapse: astrocyte gliotransmission, postsynaptic activity, and presynaptic activation of mGluRs. Our results show a novel mechanism that reconciles the previous results regarding the NO and eCB signaling role in LTP (Schuman and Madison 1991; Carlson et al. 2002; Crosby et al. 2011; Hardingham et al. 2013) and astrocyte-mediated plasticity (Panatier et al. 2006; Perea and Araque 2007; Henneberger et al. 2010; Takata et al. 2011; Navarrete et al. 2012). The effects on synaptic modulation by eCBs are spatially defined, establishing a core of synaptic depression in close synapses to the eCB source surrounded by a shell of synaptic potentiation in more distant synapses. Because present results were obtained in juvenile mice, it is possible that the described mechanisms correspond to specific events occurring during development. Further studies are required to evaluate whether astrocyte-induced eLTP is a general phenomenon also present in adults, where the expression of some signaling molecules involved might be downregulated.
Hence, these results reveal novel mechanisms underlying the consequences of eCB signaling as well as astrocyte activity on hippocampal synaptic plasticity, and may affect to the physiological roles of eCB system in learning and memory processes in particular, and in the overall brain function.
This work was supported by grants from Ministerio de Economia y Competitividad, Spain (MINECO; BFU2010-15832), European Union (HEALTH-F2-2007-202167), and Cajal Blue Brain to A.A. Grants from Spain (MINECO; BFU-2009-08404 and CSD2008-00005) to R.L. Grants from Spain (MINECO; Consolider, CSD2010-00045; Ramón y Cajal Program, RYC-2012-12014; and BFU2013-47265) to G.P.
The authors declare no competing financial interests. We thank Dr Atsu Aiba (Animal Resources, Center for Disease Biology and Integrative Medicine, Faculty of Medicine, University of Tokyo) for the donation of the mGluR1b rescue mice. Conflict of Interest: None declared.
- nitric oxide
- protein kinase c
- signal transduction
- hippocampus (brain)
- long-term potentiation
- microscopy, immunoelectron
- neuronal plasticity
- precipitating factors
- receptors, metabotropic glutamate
- synaptic transmission
- brain function
- long-term depression