INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cortical synapses are under the continuous influence of converging chemical modulators, arising from extracortical afferents as well as from cells within the cerebral cortex. A number of recent observations suggest that the endogenous cannabinoids may represent a novel class of intrinsic modulators in this brain region. First, the G_i/o protein-linked type 1 cannabinoid receptor (CB1R) is abundantly expressed throughout the cortical mantle with high levels in superficial layers 2 and 3 (Egertova and Elphick 2000; Egertova et al. 1998; Marsicano and Lutz 1999). Second, cortical neurons are capable of synthesizing the endogenous CB1R ligands anandamide and 2-arachidonylglycerol (Di Marzo et al. 1994; Stella et al. 1997) and display carrier-mediated uptake of these lipids (Beltramo and Piomelli 2000; Beltramo et al. 1997). Third, endogenous cannabinoids are rapidly inactivated by hydrolysis via a membrane-bound fatty acid amide hydrolase, which is expressed in cortical neurons (Egertova et al. 1998; Thomas et al. 1997). Last, the behavioral and cognitive effects produced by exogenous cannabinoids substantiate a role for this system in cortical processing (Feldman et al. 1997).

Acute application of natural and synthetic cannabinoids leads to a presynaptic suppression of neurotransmitter release in a number of brain regions (Gerdeman and Lovinger 2001; Hoffman and Lupica 2000, 2001; Huang et al. 2001; Morisset and Urban 2001; Schlicker and Kathmann 2001; Takahashi and Linden 2000; Vaughan et al. 1999; Wilson and Nicoll 2001). CB1R is coupled to several intracellular signaling pathways; activation of CB1R leads to a modulation of adenylyl cyclase activity and a number of voltage-dependent Ca²⁺ and K⁺ conductances (reviewed by Pertwee 1997), consistent with the effect of cannabinoids on transmitter release. In the rodent neocortex, CB1R mRNA and protein expression is dense in supragranular layers (i.e., layer 2/3) and CB1R-expressing cells appear to be mostly GABAergic interneurons (Egertova and Elphick 2000; Marsicano and Lutz 1999). Furthermore, extracellular GABA levels in the frontal cortex have been shown to be reduced following in vivo cannabinoid administration (Ferraro et al. 2001), suggesting that cannabinoids may also inhibit GABA release in the cortex. However, the direct examination of the effects of CB1R activation on inhibitory inputs received by cortical neurons has not been explored. We have found that the synthetic cannabinoid WIN55,212-2 modulates GABA release from the presynaptic terminals of local circuit interneurons that synapse onto layer 2/3 pyramidal cells in the auditory cortex. These data directly demonstrate cannabinoid suppression of inhibition in the neocortex. Portions of this work have appeared in abstract form (Trettel and Levine 2001).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Swiss-Webster mice [postnatal day 12 to 20 (P12-20); Charles River] were rapidly decapitated following CO₂ asphyxiation according to procedures approved by University of Connecticut Health Center Animal Care Committee. Brains were rapidly dissected into ice-cold saline containing (in mM) 125.0 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25.0 NaHCO₃, 2.0 CaCl₂, 2.0 MgCl₂, and 20.0 glucose, and gassed with 95% O₂-5% CO₂ (pH 7.3, 317 ± 3 mmol · kg¹, mean ± SE) and sectioned (300 µm) in the anatomically transverse plane. Cortical slices containing auditory fields (Frisina and Walton 2001; Paxinos and Franklin 2001) were incubated for 30-45 min in 32°C saline before being transferred to a recording chamber perfused with oxygenated saline (22-23°C). Ionotropic glutamate receptors were blocked with 6,7-dinitroquinoxaline-2,3-dione (DNQX, 10 µM, Tocris, Bristol, UK) and 3-[(R)-2-carboxypiperazin-4-yl]-propyl-1-phosphonic acid (CPP, 2 µM, Tocris). Layer 2/3 pyramidal neurons were visualized at ×400 (Olympus LUMPlanFI, 0.80NA) with infrared-DIC optics. These neurons responded to depolarizing current injection with regular, frequency-adapting spikes (Trettel and Levine 2001), characteristic of cortical pyramidal cells (Connors and Gutnick 1990; McCormick et al. 1985). All recordings used in these analyses were made in the whole cell voltage-clamp configuration with borosilicate glass micropipettes (R_p = 3-5 M) filled with (in mM) 120.0 CsCl, 10.0 HEPES, 1.0 EGTA, 0.1 CaCl₂, 1.5 MgCl₂, 4.0 Na₂-ATP, 0.3 Na-GTP, and 5.0 QX-314 (pH 7.3, 290 ± 4 mmol · kg¹). Signals were filtered at 2.9 kHz and digitized at 6 kHz using a HEKA EPC9 amplifier and a PCI-16 interface board (Heka Elektronic, Darmstadt, Germany). On breaking into whole cell configuration, a brief series of voltage ramps (50 ms, 2 mV/ms) were applied to promote the activity-dependent block of Na⁺ conductances by QX-314 (Sigma, St. Louis, MO). Series resistance (R_s) was then compensated to 60% or greater at 10-100 µs lag (8.7 ± 0.46 M uncompensated R_s, n = 44). During the course of the experiments, leak currents were subtracted on-line (P/4), and the input resistance (R_i) was monitored continuously with 5-mV hyperpolarizing voltage steps (50 ms). Neurons were rejected from analyses 1) if R_s was >23 M at the time of break-in or >10.5 M after compensation, 2) if R_i changed by >15% during the course of an experiment, or 3) if R_i fell below 100 M. All drugs were delivered through the bath perfusion system at 2-3 ml/min. WIN55,212-2 (Sigma), AM251 (Gift from Dr. A. Makriyannis, University of Connecticut), and DNQX were stored in 10-mM aliquots in DMSO at 20°C. WIN55,212-2 and AM251 were delivered in saline containing 0.01% BSA; final DMSO concentration did not exceed 0.03%.

Evoked inhibitory postsynaptic currents (eIPSCs) were elicited by applying 50-µs current pulses at 0.1-0.2 Hz through a saline-filled glass micropipette or a bipolar tungston electrode (World Precision Instruments, Sarasota, FL) positioned 150-200 µm lateral to the recording pipette, within layer 2. The intensity of the stimulation was adjusted so that the average eIPSC amplitude was ~70% of the maximal amplitude for each recording and ranged from 50 to 300 µA. Figure 1A illustrates the current-voltage relationship of the pharmacologically isolated GABA_A-mediated Cl conductance (n = 5). Some outward rectification of the eIPSCs could be seen before the dialysis of QX-314 was complete. Postsynaptic GABA_B responses were blocked by intracellular Cs⁺ and QX-314. Action potential-independent IPSCs (mIPSCs) were recorded in the presence of 1 µM TTX and 2.0 mM [Ca²⁺]_o. For nominally Ca²⁺-free experiments, Ca²⁺ ions were replaced with Mg²⁺and 1 mM EGTA. All IPSCs were abolished by the GABA_A antagonist bicuculline methiodide (BMI, 30 µM, Sigma; see Fig. 1B).

View larger version (15K): [in this window] [in a new window]   Fig. 1. Type 1 cannabinoid receptor activation modulates evoked inhibitory postsynaptic currents (eIPSCs) in layer 2/3 pyramidal neurons. A: pharmacologically isolated eIPSCs. A1: current-voltage plot of eIPSCs (n = 5) illustrating the observed reversal potential of the GABAA current. A2: individual sweeps (Vcom = 60  +60 mV). Scale bars = 150 pA, 25 ms. B: time course of a representative experiment in which 3 µM WIN55,212-2 was followed by co-application with AM251 (AM, 5 µM). Representative individual traces are shown above; scale bars = 175 pA, 30 ms. C: group data for eIPSCs (Vehicle, n = 3; WIN, n = 15; AM, n = 5; WIN + AM, n = 6). Stimulation artifacts have been blanked from all traces for clarity. BMI, bicuculline methiodide. * P < 0.0001 (Student's t-test).

Off-line analysis was carried out using PulseFit (Heka Elektronic) and MiniAnalysis (Synaptosoft, Decatur, GA) software. The effects of the test solutions on eIPSCs were determined by comparing the currents evoked during a 5-min baseline period (BL, 30 sweeps) to a those from a 5-min window centered around the termination of the 10-min drug exposure (30 sweeps). The mean amplitudes and the rise and decay times for the eIPSCs were compared using the Student's t-test. Miniature IPSCs were differentiated from noise by detecting inward peaks in continuous recordings that exceeded an area threshold and had exponential rise and decay time constants. The binwidth used for analyzing mIPSC frequency and kinetics before and during drug exposure was 120 s. Nonparametric Kolmogorov-Smirnov (K-S) statistics were used to compare mIPSC amplitude distributions, and the Student's t-test and one-way ANOVAs were used for determining significant changes in mIPSC frequency and rise and decay times. The paired-pulse ratio of eIPSCs (PPR = IPSC₂/IPSC₁; see Fig. 3A) was determined at 75 ms ISI with analysis bins that were identical to those used for eIPSC analysis (i.e., 30 sweeps) and significance was established using Student's t-test. Paired-pulse data are reported as the mean PPRs before and after WIN; calculating the mean PPR by dividing the P2 mean by the P1 mean yielded similar results. All data are presented as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Whole cell recordings were made from 44 layer 2/3 pyramidal neurons in primary and secondary auditory cortices. At a holding potential of 70 mV and with E_Cl = 2.4 mV, stimulation within layer 2/3 resulted in a multicomponent, inward postsynaptic current (not shown). Adding 10 µM DNQX and 2 µM CPP to the bath left a BMI-sensitive, GABA_A-mediated eIPSC (mean amplitude = 326.8 ± 54.1 pA; n = 26) that reversed polarity near E_Cl (Fig. 1A, n = 5). Application of the CB1R agonist WIN55,212-2 (3 µM) reduced the amplitude of eIPSCs in 15/15 cells tested (Fig. 1, B and C; 63.7 ± 4.8% of BL; P < 0.0001), and the magnitude of this effect was not correlated with the age of the animal used (P > 0.90). The rise and decay time constants of the eIPSCs were not altered by WIN55,212-2 exposure (P > 0.30, n = 15). The vehicle solution had no effect on eIPSC amplitude (Fig. 1C; n = 3; 99.2 ± 2.6% of BL). The effect of WIN55,212-2 on eIPSC amplitude was blocked by pretreatment with the competitive CB1R antagonist AM251 (5 µM; n = 3; BL = 447.6 ± 60 pA, WIN + AM251 = 466 ± 118 pA; P > 0.70), which alone had no effect on eIPSC amplitude (Fig. 1C; n = 5; P > 0.40). In addition, application of AM251 during WIN55,212-2 exposure in a separate group of cells reversed the depression of eIPSC amplitude to near baseline values (n = 3; see example in Fig. 1B). Because the type-2 cannabinoid receptor is not expressed in the CNS (Munro et al. 1993) and the suppression of eIPSC amplitude by WIN55,212-2 was blocked and reversed by the highly selective CB1R antagonist AM251 (i.e., K_i = 7.5 nM) (Lan et al. 1999), we conclude that the effect of WIN55,212-2 in this preparation is mediated by CB1R.

The reduction in eIPSC amplitude caused by WIN55,212-2 could involve presynaptic and/or postsynaptic mechanisms. To address this issue we examined the effects of WIN55,212-2 on mIPSCs in saline containing 1 µM TTX (Fig. 2A). In the absence of WIN55,212-2, baseline mIPSC frequency was 2.43 ± 0.25 Hz (n = 10), which was not different from the frequency of mIPSCs recorded in the presence of the vehicle control solution (Fig. 2B; 100.3 ± 3.2% of BL, n = 3, P > 0.70). Adding 3 µM WIN55,212-2 to the bath perfusate reduced the frequency of mIPSCs to 1.57 ± 0.2 Hz (Fig. 2B; n = 7, P < 0.05). WIN55,212-2 had no effect on mIPSC peak amplitude in five of seven cells (see example in Fig. 2C; P > 0.5, K-S) or on rise and decay kinetics (Fig. 2D; BL rise 10-90% = 2.48 ± 0.11 ms, _decay = 9.2 ± 1.3 ms). The reduction in mIPSC frequency with no change in peak amplitude or kinetics suggests that the CB1R-mediated reduction in eIPSCs (Fig. 1, B and C) results from a suppression of presynaptic GABA release from interneuron terminals. Because transmitter release is dependent on the voltage-gated influx of Ca²⁺ and CB1R activation has been shown to reduce Ca²⁺ conductance through N and P/Q-type Ca²⁺ channels (Caulfield and Brown 1992; Twitchell et al. 1997), we tested the hypothesis that WIN55,212-2 reduced mIPSC frequency by retarding presynaptic Ca²⁺ influx. When 100 µM Cd²⁺ was added to the extracellular solution to block voltage-gated Ca²⁺ influx, mIPSC frequency was reduced to 68.1 ± 6.1% of BL (Fig. 2E; n = 4, P < 0.03). The addition of WIN55,212-2 to the Cd²⁺-containing bath, however, did not cause a further reduction in the frequency of the remaining fraction of mIPSCs (Fig. 2E; 107 ± 6.7% of Cd²⁺ BL, n = 4). Similarly in four experiments, removing Ca²⁺ from the medium (see METHODS) reduced mIPSC frequency (Fig. 2E; 60 ± 6.6% of BL; P < 0.05) and occluded the effect of WIN55,212-2 (89.8 ± 8.2% of Ca²⁺-free BL). WIN55,212-2 had no effect on mIPSC amplitude (P > 0.5, K-S) or kinetics (P > 0.1, Student's t-test) in either Cd²⁺ or Ca²⁺-free conditions (data not shown).

View larger version (34K): [in this window] [in a new window]   Fig. 2. WIN55,212-2 reduces the frequency of Ca2+-dependent miniature IPSCs (mIPSCs). A: representative mIPSC traces for vehicle and WIN55,212-2 conditions in 2.0 mM [Ca2+]o. Scale bars = 12 pA, 200 ms. B: group time course data plotting the effect of WIN55,212-2 (n = 7, ) and vehicle solution (n = 3, ) on mIPSC frequency. Duration of drug exposure was 10 min for every cell. * P < 0.05 (ANOVA). C: cumulative histogram from a single experiment showing the effect of WIN55,212-2 () on mIPSC amplitude compared with baseline (BL; ). D: group data for the effect of WIN55,212-2 (filled bars) and vehicle (open bars) on mIPSC 10-90% rise time and decay time (67%). All values were normalized to BL for each condition. E: the effects of either Ca2+ removal (n = 4) or 100 µM Cd2+ (n = 4) and subsequent WIN55,212-2 exposure on mIPSC frequency normalized to 2.0 mM Ca2+ BL. * P < 0.05 (Student's t-test).

To further test the presynaptic locus of the CB1R-mediated suppression of GABAergic transmission, we repeated eIPSC experiments by pairing two stimuli at a 75-ms interstimulus interval and determined the PPR as IPSC₂/IPSC₁ (see Fig. 3A). As shown in the example in Fig. 3B, exposure to WIN55,212-2 increased the PPR, switching the response from depression to facilitation. By comparison, in the same cell decreasing [Ca²⁺]_o from 2.0 to 0.5 mM mimicked the effect of WIN55,212-2 on the PPR, albeit with different temporal characteristics. In five cells, the mean PPR was increased from 0.65 ± 0.05 to 1.11 ± 0.15 following WIN55,212-2 treatment (Fig. 3C; P < 0.05), indicating a reduction in the probability of transmitter release.

View larger version (10K): [in this window] [in a new window]   Fig. 3. WIN55,212-2 enhances paired-pulse facilitation at interneuronpyramidal neuron synapses. A: example traces taken during BL (dark line) and WIN55,212-2 (WIN, light line). Baseline for each pulse was the mean for a 1- to 3-ms bin immediately preceding stimulation. The 5-mV hyperpolarizing pulse used for calculating Ri is seen on the rightmost portion of each trace. Scale bars = 40 pA, 50 ms. B: individual experiment illustrating the time course of the increase in paired-pulse facilitation caused by either WIN55,212-2 (3 µM; ) or low external calcium (). C: group data showing a significant increase in PPR following WIN55,212-2 exposure (n = 5). Each line represents an individual cell before and after WIN55,212-2. Means are shown as solid dots. * P < 0.05 (Student's t-test).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The primary finding of this report is that application of the synthetic cannabinoid WIN55,212-2 depresses inhibitory synaptic transmission at GABAergic synapses received by layer 2/3 pyramidal cells in the mouse neocortex. The suppression, measured as a decrease in the amplitude of evoked IPSCs, was reliable (i.e., 15/15 cells) and was mediated by CB1R. This observation is in accord with the recent in vivo observations that WIN55,212-2 leads to decreased levels of extracellular GABA in the frontal cortex of the awake rat (Ferraro et al. 2001) and increased firing rates of prefrontal pyramidal neurons in anesthetized rats (Pistis et al. 2001). These latter results may also be partly attributable to a decrease in the spontaneous and stimulus-evoked firing rate of the interneurons. In the frontal cortex cannabinoids have previously been shown to suppress glutamate release at layer 5 synapses received by pyramidal cells (Auclair et al. 2000), which could reduce the excitatory drive on interneurons. Nonetheless, we have demonstrated that WIN55,212-2 increases the excitability of layer 2/3 pyramidal neurons in response to extracellular, intralaminar field stimulation without altering pyramidal cell membrane potential (Trettel and Levine 2001), further supporting the idea that cannabinoids depress cortical inhibition. Taken together, it appears that activation of the CB1R in the neocortex results in a suppression of GABAergic inhibition from local circuit interneurons onto pyramidal neurons.

The suppression of cortical inhibition involves a presynaptic mechanism. Two primary observations reported here, 1) a reduction in the frequency of spontaneous, action potential-independent neurotransmitter release events from GABAergic terminals and 2) an increase in paired-pulse facilitation of evoked GABA_A currents, suggest that cannabinoids depress GABA release from presynaptic terminals. Evidence against a postsynaptic mechanism of action stems from our observation that WIN55,212-2 does not alter the kinetic properties of evoked or mIPSCs. The localization of CB1R also supports a presynaptic locus. Within layer 2/3 of the neocortex, CB1R mRNA is mostly restricted to a subset of GABAergic interneurons (Marsicano and Lutz 1999). Furthermore, CB1R-immunoreactive fibers have been identified surrounding the soma of layer 2/3 pyramidal cells, which themselves do not express CB1R (Egertova and Elphick 2000). Cannabinoids have also been shown to inhibit serotonin (5-HT) (Nakazi et al. 2000) and acetylcholine release (Kathmann et al. 2001), raising the possibility that some CB1R-immunoreactive fibers may not originate from GABAergic interneurons. In the hippocampus of mice lacking CB1R, cannabinoids fail to suppress GABAergic transmission (Hajos et al. 2001; Wilson et al. 2001), further suggesting that the effect of cannabinoids on GABA release would occur via presynaptic CB1R receptors.

The mechanism(s) involved in the CB1R-mediated suppression of GABA release have not been resolved but may include modulation of presynaptic voltage-gated Ca²⁺ and K⁺ channels, leading to changes in Ca²⁺ influx, as well as direct effects on vesicle release processes downstream of Ca²⁺ entry. In the hippocampus, cannabinoids suppress GABA release through a direct G protein interaction with N-type Ca²⁺ channels (Wilson et al. 2001), resulting in an inhibition of presynaptic Ca²⁺ influx (Hoffman and Lupica 2000; Wilson and Nicoll 2001). Moreover, activation of CB1R has been shown to inhibit whole cell N- and P/Q-type Ca²⁺ currents in cultured neurons (Caulfield and Brown 1992; Twitchell et al. 1997). CB1R activation can also modulate voltage-gated K⁺ channels (Deadwyler et al. 1995; Mu et al. 2000), thereby indirectly altering Ca²⁺-dependent transmitter release. In the substantia nigra pars reticulata, Cd²⁺ has been shown to block the actions of WIN55,212-2 on GABA release (Chan and Yung 1998), further supporting the idea that the inhibition of release is ultimately mediated at the level of Ca²⁺ entry. There is also evidence for direct modulation of vesicle release, independent of Ca²⁺ influx. For example, in the cerebellum (Takahashi and Linden 2000) and periaqueductal gray (Vaughan et al. 2000) cannabinoids reduced the frequency of Ca²⁺-independent mIPSCs, suggesting that release processes downstream of Ca²⁺ entry can be regulated by CB1R signaling. In the neocortex, we observed that a fraction of mIPSCs depend on Ca²⁺ influx through voltage-gated Ca²⁺ channels and that the frequency of these events is strongly depressed by WIN55,212-2. The remaining pool of Ca²⁺-independent mIPSCs did not demonstrate WIN55,212-2 sensitivity, consistent with the idea that in the neocortex cannabinoids inhibit transmitter release at the point of Ca²⁺ entry, similar to the hippocampus. It is unclear from the present studies, however, whether CB1R activation modulates Ca²⁺ channels directly or if changes in Ca²⁺ influx are secondary to modulation of presynaptic K⁺ channels (e.g., Daniel and Crepel 2001). It is also possible that the population of terminals expressing CB1R generate only Ca²⁺-dependent mIPSCs, in which case the reduction in mIPSC frequency that we observed in response to WIN55,212-2 may still reflect inhibition of vesicle release downstream of Ca²⁺ influx. At the present time, this interpretation is difficult to exclude.

The endogenous cannabinoids anandamide and 2-arachidonylglycerol are synthesized and released from cortical neurons in an activity-dependent manner (Di Marzo et al. 1994; Stella et al. 1997). The recent demonstration that endocannabinoids act retrogradely to inhibit transmitter release in the cerebellum (Kreitzer and Regehr 2001a,b) and hippocampus (Wilson et al. 2001; Wilson and Nicoll 2001) raises the possibility that these compounds have a similar function in the cortex. A reduction in inhibition caused by CB1R activation in layer 2/3 of the neocortex could provide a mechanism whereby pyramidal cells transiently increase their responsiveness to associative inputs and switch from tonic firing to bursting. It is clear that the generation of bursts in regular spiking layer 5 pyramidal neurons is highly sensitive to apical (i.e., layer 2/3) inhibition and occurs when excitatory inputs from basal and apical dendrites are temporally correlated (Larkum et al. 1999, 2001). The release of endogenous cannabinoids from the apical dendrites of pyramidal neurons may suppress inhibition to a degree that would promote burst firing. Furthermore, the finding that endocannabinoid release from cortical neurons is enhanced by acetylcholine (Stella and Piomelli 2001) suggests that ascending inputs may gate the action of these intrinsic neuromodulators.