ARTICLE

TOP ABSTRACT ARTICLE REFERENCES

The activity of neocortical pyramidal neurons (PNs) is regulated by diverse classes of GABAergic interneurons. These inhibitory neurons innervate functionally segregated domains of PNs to control action potential timing, the efficacy of excitatory inputs, and the synchronous activity of PNs (Larkum et al. 1999; Somogyi et al. 1998; Szabadics et al. 2001). Many interneurons form axosomatic contacts with PNs (Somogyi et al. 1998), and interneuron discharge rates in vivo can be very high (Mountcastle et al. 1969), suggesting that physiological regulation of this inhibition is likely essential for neocortical function. Several lines of evidence indicate that the endocannabinoid system may modulate the strength of inhibition in the neocortex. First, the type 1 cannabinoid receptor (CB1R) is abundantly expressed throughout the cortical mantle (Mailleux and Vanderhaeghen 1992) where it is mostly limited to interneurons of layers 2/3 and 6 (Egertova and Elphick 2000; Marsicano and Lutz 1999). Second, activation of CB1R reduces neocortical GABA levels (Ferraro et al. 2001) by inhibiting Ca²⁺-dependent GABA release (Trettel and Levine 2002), consistent with the effects of CB1R agonists in other brain regions (reviewed by Schlicker and Kathmann 2001). Third, neocortical neurons synthesize the endocannabinoid arachidonylethanolamide (anandamide) (Devane et al. 1992) and display carrier-mediate uptake (Beltramo et al. 1997) and enzymatic degradation of this lipid-derived compound (Beltramo and Piomelli 2000). Last, exogenous cannabinoids have pronounced effects on neocortical function (Feldman et al. 1997). Despite this evidence, the role of endocannabinoid signaling in the neocortex is largely unexplored.

In the hippocampus and cerebellum, endocannabinoids function as retrograde messengers that mediate depolarization-induced suppression of inhibition (DSI) (reviewed by Kreitzer and Regehr 2002; Wilson and Nicoll 2002). DSI is a transient reduction in presynaptic GABA release following postsynaptic depolarization (Llano et al. 1991; Pitler and Alger 1992) that enhances neuronal excitability (Wagner and Alger 1996) by allowing target neurons to regulate the strength of afferent inhibition. Stimuli that result in voltage-dependent Ca²⁺ entry (Alger et al. 1996; Llano et al. 1991) or activation of metabotropic glutamate (mGluR) or muscarinic acetylcholine receptors (Kim et al. 2002; Maejima et al. 2001; Varma et al. 2001) are sufficient to initiate the release of endocannabinoids from hippocampal and cerebellar neurons. These molecules then diffuse retrogradely to presynaptic terminals to activate CB1R, resulting in the depression of GABA release. Such signaling mechanisms may also be essential for the physiological regulation of cortical inhibition. Therefore in the present study, we sought to determine whether depolarization of neocortical PNs triggers endocannabinoid-mediated retrograde signaling at inhibitory synapses.

To determine whether endocannabinoids act as retrograde messengers in the neocortex, we examined synaptic inhibition of layer 2/3 PNs in 250-µm-thick slices of mouse auditory and visual cortex (P12-20; Swiss-Webster, Charles River). Briefly, whole cell voltage-clamp recordings were made from PNs at 32°C, and GABA_A-mediated inhibitory postsynaptic currents (IPSCs) were evoked using bipolar stimulation (25-100 µs, 100-900 µA). To assay the effects of PN depolarization, evoked IPSCs (eIPSCs) were paired with repeated, brief epochs of membrane depolarization (pairing protocol; see Fig. 1A). Brain slices were perfused at 1.5-2 ml/min in a submersion-type chamber with normal artificial cerebrospinal fluid (ACSF) containing (in mM) 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 2 CaCl₂, 2 MgCl₂, 17.5 glucose, 0.01 6,7-dinitroquinoxaline-2,3-dione (DNQX), and 0.002 3((+/)-2-carboxypiperazin-4-yl)propyl-1-phosphonate (CPP), equilibrated with 95% O₂-5% CO₂ (pH 7.3, 305 mmol/kg). Borosilicate glass recording pipettes had resistances of 3-5 M when filled with internal solution. For BAPTA experiments, the internal solution contained (in mM) 105 CsCl, 10 HEPES, 15 Cs₄-BAPTA, 1.5 CaCl₂, 1.5 MgCl₂, 3.5 Na₂-ATP, and 0.3 Na-GTP (pH 7.3, 285 mmol/kg). For all other experiments, the internal solution was composed of (in mM) 120 CsCl, 10 HEPES, 2 EGTA, 0.2 CaCl₂, 1.5 MgCl₂, 3.5 Na₂-ATP, and 0.3 Na-GTP (pH 7.3, 282 mmol/kg). Lidocaine N-ethyl bromide (QX-314, 5 mM) was included in the pipette solutions to block g_Na. All recordings were made at -70 mV, and synaptic currents were filtered at 2.9 kHz and digitized at 6 kHz using a Heka EPC9/2. AM404 and AM251 were generously provided by Dr. Alex Makriyannis (University of Connecticut, Storrs, CT). Neurons were rejected from analyses if the series resistance was >30 M at the time of break-in or >12 M after compensation at 100 µs lag, if input resistance (R_i) changed by 15% during the course of an experiment, or if R_i fell <150 M. Off-line analysis was carried out using PulseFit (Heka Elektronic), MiniAnalysis (Synaptosoft), and Origin (OriginLab, Northampton MA) software and significance was tested using the Student's t-test and one-way ANOVAs. All data are presented as means ± SE.

View larger version (12K): [in this window] [in a new window]   Fig. 1. Pyramid neuron (PN) depolarization results in suppression of GABAergic afferents. A: voltage-clamp pairing protocol. Fifty inhibitory postsynaptic currents (IPSCs) were evoked at 0.3 Hz and paired with 150-ms voltage steps to 0 mV during trials 20-30. Individual sweeps correspond to areas 1 (prepairing), 2 (pairing), and 3 (postpairing) in B for control conditions. Scale bar: 350 pA, 300 ms. B: group time course of evoked IPSC (eIPSC) amplitude normalized to the 5 trials preceding pairing (baseline, BL). Data for control cells (n = 10; ) and for 15 mM BAPTA (n = 5; ) are shown. Dashed lines mark the 10 pairing trials in this and subsequent time courses.

Somatic whole cell recordings were obtained from 51 layer 2/3 PNs. At V_hold = 70 mV, stimulation near the soma produced large eIPSCs (1.73 ± 0.21 nA) that reversed polarity near E_Cl and were abolished by bicuculline methiodide (30 µM; not shown). To test the effects of somatic depolarization on GABAergic inhibition, we paired extracellular stimulation with brief epochs of membrane depolarization (Fig. 1A). As shown in Fig. 1B, two pairing trials were sufficient to depress eIPSC amplitude (Fig. 1B; 90.8 ± 1.7% of baseline; P < 0.05; n = 10), and at the end of 10 pairings, the suppression reached 63.2 ± 4.8% of baseline (Figs. 1, A and B, and 4; P < 0.001; n = 10). There was no relationship between the magnitude of IPSC suppression and the initial size of the eIPSCs (P > 0.8). The recovery from suppression was rapid: at 60 s following pairing, eIPSC amplitude was 90.4 ± 5.7% of baseline (P > 0.25 compared with baseline).

To determine the synaptic locus of the suppression, we first measured the paired-pulse ratio (PPR = eIPSC₂/eIPSC₁). PN depolarization suppressed eIPSC₁, and this was paralleled by a sharp increase in the PPR from 1.11 ± 0.1 to 1.81 ± 0.1 (Fig. 2, A and B; P < 0.01; n = 6). These data suggest that the suppression of eIPSC amplitude was associated with a reduction in the probability of presynaptic transmitter release (e.g., Zucker 1989). We also measured whole cell currents evoked by pressure ejection of GABA onto PN soma to determine if PN depolarization resulted in changes in postsynaptic GABA_A receptors. Immediately following the pairing voltage protocol, the amplitude of the GABA current was 103.5 ± 5.3% of baseline (Fig. 2C; P > 0.1; n = 7). Therefore the most likely explanation for the observed suppression of eIPSC amplitude was that PN depolarization affected presynaptic function via retrograde signaling. To investigate the role of postsynaptic Ca²⁺ in triggering the release of a retrograde messenger, we loaded PNs with 15 mM BAPTA. Under these conditions the suppression of eIPSC amplitude was significantly attenuated (Figs. 1B and 4; 92.3 ± 6.7% of baseline; P < 0.001 compared with control; n = 5). These data suggest that the synthesis and/or release of the retrograde signal required a rise in postsynaptic intracellular Ca²⁺, similar to the induction of retrograde signaling in other brain systems (Llano et al. 1991; Pitler and Alger 1992).

View larger version (23K): [in this window] [in a new window]   Fig. 2. Depolarization-induced IPSC suppression is associated with changes in presynaptic but not postsynaptic function. A: time course illustrating the effect of pairing on the paired-pulse ratio (PPR) for 6 cells. The single stimulus was replaced by 2 stimuli (ISI = 75 ms) and PPR was calculated as eIPSC2/eIPSC1. The mean amplitude of IPSC1 is plotted on the left y axis () and the mean PPR is plotted on the right y axis (). Error bars were omitted from PPR values for clarity. B: changes in PPR for individual cells. Values represent means of the 5 trials before, during, and after pairing. *, P < 0.05. C: pressure application of 50 or 100 µM GABA (2-4 psi; 10 ms) to PN soma delivered through a glass micropipette positioned 20-40 µm from the cell. Currents were evoked every 30 s and the 10 PN depolarizations are indicated by the square wave symbol. Data represent the mean values of the peak GABA current amplitude (IGABA) for 7 cells. Representative sweeps from a single experiment ([GABA] = 50 µM) are shown. Scale bar: 500 pA, 50 ms. GABA currents were completely blocked by 30 µM bicuculline (data not shown).

We next asked whether the retrograde signal was an endocannabinoid. Pretreatment of slices with the selective CB1R antagonist AM251 (2 µM) completely abolished the pairing-induced suppression of eIPSCs (Figs. 3A and 4; 97.4 ± 1.6% of baseline; n = 8), while having no effect on baseline eIPSC amplitude (data not shown). In addition, CB1R activation by the synthetic cannabinoid WIN55,212-2 (5 µM) reduced eIPSC amplitude to 54.8 ± 5.0% of baseline (Fig. 3B; P < 0.01; n = 6) and occluded the effects of PN depolarization (Figs. 3B and 4; 95.5 ± 2.8% of baseline; P > 0.8). The strong correlation between the magnitudes of WIN-mediated inhibition and depolarization-induced suppression of eIPSC amplitude (Fig. 3C; r² = 0.92; n = 6) suggests that most if not all of the WIN-sensitive inhibitory afferents to PNs are suppressed by depolarization. The rapid recovery from eIPSC suppression (i.e., Fig. 1B) prompted us to explore the role of endocannabinoid uptake using a selective inhibitor of the anandamide transporter, AM404 (Beltramo et al. 1997). As shown in Fig. 3D, AM404 (25 µM) reduced the latency to peak suppression, increased the magnitude of the suppression (Fig. 4; 36.9 ± 12.0% of baseline compared with 62.7 ± 5.9% of baseline, P < 0.05; n = 5), and significantly retarded eIPSC recovery (P < 0.05; n = 5). Taken together, these data suggest that CB1R activation by endocannabinoids mediates the transient suppression of eIPSCs following repeated epochs of PN depolarization.

View larger version (30K): [in this window] [in a new window]   Fig. 3. Retrograde synaptic suppression is mediated by endocannabinoids. A: group time course illustrating the effect of preincubating slices with AM251 (2 µM; n = 8) on pairing-induced eIPSC suppression. B: application of 5 µM WIN55,212-2 (WIN, ) reduced eIPSC amplitude and occluded suppression. For each cell (n = 6), suppression under control conditions () was assessed prior to application of WIN55,212-2; cells were exposed to WIN55,212-2 for 10 min before testing in all experiments. The magnitude of eIPSC inhibition caused by WIN55,212-2 is plotted as a function of the amount of pairing-induced eIPSC suppression for the 6 cells in C. The data are fit by linear regression with slope = 1.07 and r2 = 0.92 (P < 0.001). D: time course showing the effects of extracellular AM404 (25 µM) on eIPSC suppression. As in B, control () and AM404 () conditions represent data taken from the same group of cells (n = 5). AM404 was applied for 10 min prior to examining depolarization-induced suppression of inhibition (DSI) and was present throughout the experiment. The 10 PN depolarizations are indicated by the square wave symbol. AM404 alone had no effect on baseline eIPSC amplitude (data not shown).

View larger version (19K): [in this window] [in a new window]   Fig. 4. Summary of pairing-induced suppression of eIPSCs in the neocortex. Magnitude of suppression for all conditions plotted as a percentage of the prepairing baseline (control, n = 10; BAPTA, n = 5; AM251, n = 8; WIN55,212-2, n = 6; and AM404, n = 5). *, P < 0.05 compared with control.

To our knowledge, this is the first report of the cellular effects of endocannabinoid signaling in the neocortex. Two other forms of retrograde signaling at neocortical layer 2/3 synapses have been reported previously. In one type, back-propagating action potentials triggered the release of GABA from the dendrites of bipolar interneurons and suppressed glutamate release from afferent PNs via GABA_B receptor activation (Zilberter et al. 1999). In the other form, glutamate released from depolarized PN dendrites acted on fast-spiking interneurons via mGluRs to depress GABA release (Zilberter 2000). The present results show that endocannabinoids are also involved in the suppression of GABA release from neocortical interneurons. It is possible that both retrograde glutamate and endocannabinoid signaling occur at an overlapping population of layer 2/3 synapses received by PNs. Alternatively, different retrograde messengers may mediate suppression of distinct classes of interneurons. Because these interneurons innervate segregated functional domains of PNs, e.g., apical dendrites, soma/basal dendrites, and the axon initial segment (Somogyi et al. 1998), selective modulation of this compartmentalized inhibition could play a critical role in shaping the responsiveness and firing patterns of PNs.

Several observations support our conclusion that endocannabinoids mediate the suppression of inhibition following the depolarization of neocortical PNs. First, the suppression of eIPSCs was completely blocked by the diarylpyrazole-type CB1R antagonist AM251. Second, the suppression was presynaptically expressed, similar to the effects of the cannabinomimetic WIN55,212-2 in the neocortex (Trettel and Levine 2002). Furthermore, WIN55,212-2 occluded the suppression, suggesting a common target for both WIN55,212-2 and the endogenous retrograde signal. Third, PNs synthesize and release endocannabinoids in a Ca²⁺-dependent manner (Di Marzo et al. 1994; Stella et al. 1997), and the suppression of inhibition in the present study was dependent on a rise in intracellular Ca²⁺. Last, inhibiting endocannabinoid uptake with AM404 had a pronounced effect on eIPSC suppression, suggesting an important role for reuptake during retrograde signaling. Together, these results indicate that in the neocortex, similar to the hippocampus and cerebellum, endocannabinoids released from PNs act retrogradely to suppress GABA release from interneurons that express CB1R.

The physiological significance of endocannabinoid signaling in the neocortex may ultimately be reflected in the activity of PNs. The ability of PNs to fire bursts of action potentials is thought to play a key role in cortical information processing (Kepecs et al. 2002; Lisman 1997). As associative neurons, PNs receive excitatory input over widespread, functionally segregated dendritic domains. The integration or coupling of temporally coincident distal (i.e., associative, nonspecific intracortical) and basal (i.e., specific thalamo-cortical) inputs leads to burst firing (Larkum et al. 2001) and is highly sensitive to GABAergic inhibition (Larkum et al. 1999). Regulation of this inhibition likely plays an essential role in shaping PN dendritic integration and discharge pattern. Thus activity-dependent release of endocannabinoids from PNs could transiently suppress afferent inhibition to a degree that promotes the coupling of excitatory inputs, thereby representing a functional role for endocannabinoid-mediated DSI in the neocortex.