Differential modulation of nociceptive versus non-nociceptive synapses by endocannabinoids

For most of the experiments in this study, excitatory post-synaptic potentials (EPSPs) were recorded from synapses made by either the lateral (polymodal) nociceptive (N-) cell or one of the pressure-sensitive (P-) cells onto the L motor neuron (Figure 1). The L motor neuron innervates the longitudinal muscles and is active during the defensive withdrawal reflex, whole-body shortening [19]. This motor neuron receives monosynaptic, glutamatergic input from the both the P- and N-cells [10, 20]. Previous studies have shown that 2AG (60 μM for 15 mins) induced LTD in nociceptive synapses made by the lateral N-cell onto the L motor neuron (Figure 2A, B; data re-presented from [10]; t=6.45, p≤0.0001, n=5). In contrast, the non-nociceptive P-to-L synapse was actually potentiated 1 hr following bath application of 60–100 μM 2AG for 15 min (N=10) relative to control synapses in which 2AG application was omitted (Figure 2A, B; t=3.37, p≤0.005, n=10). Potentiation did not appear to be due to postsynaptic changes in intrinsic excitability given that input resistance (IR) in the L motor neuron was unchanged (post-test IR was 100.6±2.9% of pre-test levels in the 2AG-treated group and 103.2±3.7% in the vehicle control group). The effects of 60 and 100 μM 2AG on the P-to-L synapse were indistinguishable with 60 μM 2AG producing a 48.2±10.4% increase (n=4) and 100 μM producing a 56.1±22.1% (n=6) increase in EPSP amplitude (independent t-test p=0.79); therefore the data from these two concentrations were combined. N-to-L synapses treated with 100 μM 2AG still underwent depression (64.6±10.3% of initial EPSP amplitude; n=3) that was no different from the effects observed at 60 μM (63±4%; independent t-test p=0.87). These results indicate that the potentiating effect of 100 μM 2AG on the non-nociceptive synapse was not a consequence of the concentration of 2AG used.

To determine if these opposing effects of 2AG were a feature of other nociceptive and non-nociceptive synapses, N- and P-cell input to the anterior pagoda (AP) cell was also examined. Although the function of the AP cell is not known, the monosynaptic, glutamatergic P-to-AP synapses have been used to study a variety of forms of synaptic plasticity in the leech [21‐25]. However, the N-to-AP synapse had not been previously characterized and this was done prior to conducting the 2AG experiments. In general the N-to-AP synapse was quite similar to the N-to-L connection. N-to-AP was substantially reduced by CNQX (20 μM) indicating glutamatergic transmission (Additional file 1: Figure S1A). In experiments with high MgCl₂ saline (15 mM), which blocked all chemical synaptic transmission [20], the majority of the N-to-AP EPSP was eliminated except for a small (<1 mV) early component that appeared to be a rectifying electrical synapse between the N-and AP-cell (Additional file 1: Figure S1B). This putative electrical EPSP was also observed in the CNQX experiments. Experiments with high divalent saline (18 mM MgCl₂/15 mM CaCl₂) eliminated later components of the N-to-AP EPSP indicating the presence of polysynaptic elements (Additional file 1: Figure S1C). While positive current associated with the N-cell action potential did propagate to the AP via electrical coupling, negative current injected into the N-cell did not spread to the AP (data not shown). Interestingly, evidence of electrical coupling was also observed in the AP-to-N direction. Injection of 500 msec current pulses showed that negative current, but not positive current, could spread from the AP to the N-cell (Additional file 1: Figure S1D).

In terms of eCB modulation, 2AG (100 μM) depressed N-to-AP synapses (n=9 and 5 respectively; t=2.91, p≤0.05), but potentiated P-to-AP synapses relative to controls (n=6 and 5 respectively; t=2.36, p≤0.05) identical to N- and P-cell inputs to the L motor neuron (Figure 2C). No obvious changes in postsynaptic IR were observed during the P-to-AP (2AG group=111.5±3.2%, control=106.8±5.6) or N-to-AP (2AG group=104.6±3.8%, control=97.7±6.2%) synapses.

To test whether 2AG-induced potentiation of the non-nociceptive P-cell synapses is mediated by a TRPV-like receptor, experiments were carried out using the selective TRPV1 antagonist SB366791 (10 μM), which blocks eCB-mediated LTD in leech nociceptive synapses [10]. When SB366791 was co-applied with 2AG, potentiation of the non-nociceptive P-to-L synapse was prevented (Figure 3A; n=6). Application of SB366791 alone had no effect on the P-to-L EPSP (N=5). One-way ANOVA of the 2AG+SB366791, SB366791, 2AG and vehicle control groups showed a significant treatment effect (F_2,18=6.01, p≤0.05). Post-hoc analysis showed that the 2AG group was statistically different from the control group (p≤0.05), but that the 2AG+SB366791 and SB366791 groups were not. Postsynaptic IR was unchanged in the 2AG+SB366791 (110.1±4.6% of pre-test levels), SB366791-only (101.7±5.6%) and vehicle control groups (101±3.9%; n=6). These findings indicate that the difference in the effects of 2AG on nociceptive vs. non-nociceptive synapses is not due to activation of different endocannabinoid receptor.

eCBs are known to depress inhibitory synaptic transmission and eCB-induced disinhibition has been shown to increase synaptic signaling within spinal nociceptive circuits, although the identity of the presynaptic input(s) being enhanced was not known [4]. Therefore, we examined the potential role of GABA-A receptors during 2AG-induced potentiation of the non-nociceptive synapse in the leech. The leech CNS contains GABAergic neurons and the GABA-A receptor antagonist bicuculline has been shown to disinhibit other circuits that utilize P-cell input [26‐28]. To test whether functional GABA-A receptors are required for 2AG-mediated potentiation, ganglia were pre-treated with bicuculline (100 μM) followed by the pre-test measurements of the P-to-L EPSP, 15 min application of 2AG, washout for 60 mins and finally post-test EPSP measurements. The 2AG treatment that normally produces potentiation was blocked in the non-nociceptive P-to-L synapses that had undergone bicuculline+2AG treatment (Figure 3B; n=5). P-to-L synapses that were pre-treated with bicuculline, but subsequent 2AG treatment omitted, were unchanged between the pre- and post-tests (Figure 3B, n=5). One-way ANOVA of the 2AG+bicuculline, bicuculline, 2AG and vehicle control group showed a significant effect of treatment (F_2,24=7.13, p≤0.005). Post-hoc analysis showed that the 2AG group was statistically different from the control group (p<0.01) and that the 2AG+bicuculline and bicuculline groups were not. No change in postsynaptic input resistance was observed in the 2AG+bicuculline (104±4.4%) or bicuculline only group (96±4.6%). These findings support the hypothesis that 2AG-induced potentiation of non-nociceptive synapses is mediated by disinhibition as a result of a decrease in GABAergic input.

If 2AG potentiates non-nociceptive synapses due to a decrease in GABAergic input, why are nociceptive synapses not similarly disinhibited? GABA has been observed to depolarize the lateral N-cells used in these experiments [29] so it is possible that GABA excites rather than inhibits the nociceptive synapse. To confirm this earlier finding and to determine what effect GABA has on the P-cell, intracellular recordings of the N- and P-cells were made in response to GABA delivered via puffer electrodes. GABA reliably elicited depolarization in the lateral N-cells, consistent with earlier findings, but produced hyperpolarization in the P-cells (Figure 4A). Both the N-cell (n=6) and P-cell (n=3) responses were significantly inhibited following 15 min bath-application of 100 μM bicuculline (Figure 4A; N-cell, paired t-test t=8.80, p≤0.001; P-cell, t=4.56, p≤0.05). No changes in N-cell (n=5) or P-cell (n=5) GABA response were observed in control experiments in which responses were recorded prior to and then 15 mins following treatment with normal saline (Figure 4A; N-cell, paired t-test t=1.65, p>0.05; P-cell, t=0.81, p>0.05).

Next, P-to-L and N-to-L EPSPs were recorded prior to and then during acute bicuculline treatment (100 μM for 10–15 mins). As expected, the P-to-L EPSP amplitude increased during bicuculline treatment (n=7) compared to control synapses (n=5); that is, disinhibition of the non-nociceptive synapse was observed (Figure 4B; t=2.71, p≤0.05). However, bicuculline treatment actually decreased the amplitude of the N-to-L EPSP (n=8) compared to control synapses (n=5) suggesting that GABA has a tonic excitatory effect on this nociceptive synapse (Figure 4B; t=2.35, p≤0.05). No change in postsynaptic input resistance was observed during either the P-to-L synapse (bicuculline=96.7±4.9, control=105.7±6.4%) or N-to-L synapse (bicuculline=107.7±2.2%, control=98.7±3%) experiments. These findings are consistent with the idea that this nociceptive synapse does not undergo 2AG-induced disinhibition, at least in part, because the presynaptic neuron is excited, not inhibited, by GABA.

Leeches (Hirudo verbana) were obtained from commercial suppliers (Leeches USA, Westbury, NY and Niagara Leeches, Cheyenne, WY) and maintained in artificial pond water (0.50g/L H₂O Hirudo salt) on a 12 hour light/dark cycle at 18°C. Ganglia were dissected and pinned in a recording chamber with constant perfusion of normal leech saline (≈1.5 ml/min). All dissections and recordings were carried out in normal leech saline (110 mM NaCl, 4 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM NaOH, and 10 mM HEPES, pH=7.4). Drugs were dissolved in leech saline from stock solutions and final concentrations were made just prior to respective experiments. The following drug was obtained from Tocris (Ellisville, MO): 2-arachidonoyl glycerol (2AG). Drugs obtained from Sigma-Aldrich (St. Louis, MO) included CNQX, dimethyl sulfoxide (DMSO), and bicuculline.

Techniques used in this study have been described in detail in [10]. Briefly, current clamp (bridge balanced) intracellular recordings were carried out using sharp glass microelectrodes (tip resistance 35–40 MΩ) made from borosilicate capillary tubing (1.0 mm OD, 0.75 mm ID; FHC, Bowdoinham, ME) using a horizontal puller (Sutter Instruments P-97; Novato, CA). Microelectrodes were filled with 3M potassium acetate. Manual micropositioners (Model 1480; Siskiyou Inc., Grants Pass, OR) were used to impale individual neurons during experiments. Current was delivered to electrodes using a multi-channel programmable stimulator (STG 1004; Multi-Channel Systems; Reutlingen, Germany) and the signal was recorded using a bridge amplifier (BA-1S; NPI, Tamm, Germany) and digitally converted for analysis (Axoscope; Molecular Devices, Sunnyvale, CA).

The presynaptic lateral nociceptive (N) and pressure (P) cells and the postsynaptic longitudinal (L) motor neuron and anterior pagoda (AP) cell were identified based on their position with the ganglion (Figure 1), size, and characteristic electrophysiological properties (size and shape of action potential). L motor neuron identification could be confirmed by recording from the electrically coupled contralateral L motor neurons and observing synchronous activity [61]. For experiments utilizing N-to-L and P-to-L synapse recordings, the ganglion was pinned dorsal side up so that the L motor neurons could be located on the dorsal side along with access to the lateral-most N- and P-cells. For N-to-AP and P-to-AP synapse recordings, the ganglion was pinned ventral side up. Following pre-test recordings of the excitatory postsynaptic potentials (EPSPs), the ganglion was superfused with 2AG for 15 minutes and then returned to normal saline. In vehicle control experiments, 2AG was replaced with saline containing 0.01% DMSO. After one hour, the EPSP was retested (post-test). Separate electrode impalements of the same presynaptic and postsynaptic neuron were made for pre- and post-test recordings. Chronic intracellular recordings of these neurons were not carried out because this results in progressive rundown of the EPSP within 10–15 mins most likely due to damage caused by movements of the tissue during the electrode impalement (there are muscle fibers and connective tissue present in the leech CNS). Input resistance was recorded at the pre- and post-test level and only consistent, stable recordings were included in the data analysis (see Results section). The peak EPSP amplitude was recorded every 10 seconds and calculated by averaging 5–10 EPSP (pre- or post-test) sweeps.

For N-to-AP characterization experiments, the ganglion was pinned ventral side up as the lateral N-cell and AP-cell were both located on the ventral side. Treatments for the set of characterization experiments (high Mg²⁺ saline, high divalent saline, and CNQX) were perfused for 10 to 15 minutes between pre-test and post-test recordings while controls were carried out in constant perfusion of normal leech saline. Coupling experiments were performed by injecting single 500 msec hyperpolarizing current pulses with increasing amounts of current into the AP- or N-cell and recording any subsequent response from the opposite cell.

The effects of GABA on the N- and P-cells were tested using a “puffer” electrode (a patch electrode with a resistance of ≈ 5 MΩ) connected to a picospritzer. The puffer electrode was positioned approximately 100 μm from the N or P soma to prevent movement artifact and a 200 msec pulse of GABA was applied at 10 psi. The membrane potential of the P-cell was maintained at −50 mV. The membrane potential of the N-cell was maintained at −60 mV in order to prevent the cell from firing when GABA was applied. The concentration of GABA within the puffer electrode was 1 M, consistent with similar studies in the leech by Sargent et al. [29]. This high concentration in the electrode was required due to the substantial amount of dilution that occurred by the time the expelled GABA reached the neuropil, which is relatively deep in the ganglion. That responses to GABA appear to arise from N- and P-cell processes in the neuropil is based on the delay between the GABA puff and the P- or N-cell response which ranged between 100–300 msec.