Differences in chloride gradients allow for three distinct types of synaptic modulation by endocannabinoids

Endocannabinoids are important signaling molecules that depress both excitatory synapses leading to decreases in neural circuit output and inhibitory synapses leading to increases in circuit output due to disinhibition. In this study, difference in Cl⁻ gradients were found to strongly influence whether endocannabinoids ultimately decrease or increase circuit output. These findings may explain the pro- and antinociceptive effects of endocannabinoids. The role of Cl⁻ gradient may also be relevant to other modulatory transmitters as well.

Abstract

Endocannabinoids can elicit persistent depression of excitatory and inhibitory synapses, reducing or enhancing (disinhibiting) neural circuit output, respectively. In this study, we examined whether differences in Cl⁻ gradients can regulate which synapses undergo endocannabinoid-mediated synaptic depression vs. disinhibition using the well-characterized central nervous system (CNS) of the medicinal leech, Hirudo verbana. Exogenous application of endocannabinoids or capsaicin elicits potentiation of pressure (P) cell synapses and depression of both polymodal (N_poly) and mechanical (N_mech) nociceptive synapses. In P synapses, blocking Cl⁻ export prevented endocannabinoid-mediated potentiation, consistent with a disinhibition process that has been indicated by previous experiments. In N_mech neurons, which are depolarized by GABA due to an elevated Cl⁻ equilibrium potentials (E_Cl), endocannabinoid-mediated depression was prevented by blocking Cl⁻ import, indicating that this decrease in synaptic signaling was due to depression of excitatory GABAergic input (disexcitation). N_poly neurons are also depolarized by GABA, but endocannabinoids elicit depression in these synapses directly and were only weakly affected by disruption of Cl⁻ import. Consequently, the primary role of elevated E_Cl may be to protect N_poly synapses from disinhibition. All forms of endocannabinoid-mediated plasticity required activation of transient potential receptor vanilloid (TRPV) channels. Endocannabinoid/TRPV-dependent synaptic plasticity could also be elicited by distinct patterns of afferent stimulation with low-frequency stimulation (LFS) eliciting endocannabinoid-mediated depression of N_poly synapses and high-frequency stimulus (HFS) eliciting endocannabinoid-mediated potentiation of P synapses and depression of N_mech synapses. These findings demonstrate a critical role of differences in Cl⁻ gradients between neurons in determining the sign, potentiation vs. depression, of synaptic modulation under normal physiological conditions.

NEW & NOTEWORTHY

Endocannabinoids are important signaling molecules that depress both excitatory synapses leading to decreases in neural circuit output and inhibitory synapses leading to increases in circuit output due to disinhibition. In this study, difference in Cl⁻ gradients were found to strongly influence whether endocannabinoids ultimately decrease or increase circuit output. These findings may explain the pro- and antinociceptive effects of endocannabinoids. The role of Cl⁻ gradient may also be relevant to other modulatory transmitters as well.

many modulatory neurotransmitters, such as endocannabinoids, opioids, and adenosine, depress both excitatory (glutamatergic) and inhibitory (GABAergic/glycinergic) central synapses (Freund et al. 2003; Morairty et al. 2004; Lau and Vaughan 2014). Consequently, these transmitters are capable of either decreasing neural circuit output by depressing excitatory synapses or increasing circuit output by depressing inhibitory synapses (i.e., disinhibition). This raises a question; how do neural circuits regulate which synaptic pathways undergo depression vs. disinhibition? This has been a major concern in understanding how endocannabinoids modulate pain circuits given that these transmitters have been observed to produce both depression (Morisset and Urban 2001; Liang et al. 2004) and disinhibition (Pernia-Andrade et al. 2009) of nociceptive synaptic input.

It has been hypothesized that differences in Cl⁻ gradients may act to regulate whether or not synapses are sensitive to disinhibition, potentially explaining how endocannabinoids can have both pro- and antinociceptive effects (Christie and Mallet 2009). Most neurons have low levels of intracellular Cl⁻ that are maintained by the K⁺-Cl⁻ cotransporter 2 (KCC2) that exports Cl⁻ so that activation of ionotropic GABA receptors leads to Cl⁻ influx and hyperpolarization. However, neurons can also have a high intracellular Cl⁻ due to the influence of the Na⁺-K⁺-Cl⁻ cotransporter 1 (NKCC1) that imports Cl⁻ so that GABA receptor activation elicits Cl⁻ efflux and depolarization (Kaila et al. 2014). While KCC2-expressing neurons would be sensitive to neuromodulator-induced disinhibition, NKCC1-expressing neurons would be presumably “protected” from such disinhibition.

Testing the role of differences in Cl⁻ gradients is impossible without prior knowledge of which neurons have normal vs. elevated intracellular Cl⁻ and whether their synapses undergo endocannabinoid-mediated modulation. Knowing such details, a priori, is possible in the well-characterized central nervous system (CNS) of the medicinal leech, Hirudo verbana (Fig. 1). In Hirudo, nonnociceptive pressure-sensitive afferents (P cell) are inhibited by GABA while the polymodal nociceptive (N_poly) and mechano-only nociceptive (N_mech) afferents are excited by GABA (Sargent et al. 1977; Wang et al. 2015). These opposing effects of GABA are due to differences in the Cl⁻ equilibrium potentials (E_Cl), which are regulated by KCC2- and NKCC1-like transporters in the P and N cells, respectively (Wang et al. 2015). Pharmacological inhibition of KCC2 in Hirudo produced a depolarizing shift in the E_Cl in the P cells, as well as reduced GABA-induced hyperpolarization and increased/disinhibited synaptic transmission. Inhibition of NKCC1 in the N cells, however, produced a hyperpolarizing shift in the E_Cl that was accompanied by a reduction in GABA-induced depolarization and a decrease in synaptic transmission.

Fig. 1.

A: diagram of Hirudo sensorimotor synapses including the putative GABAergic input they receive. The pressure (P), polymodal (N_poly), and mechanical (N_mech) sensory neurons all have excitatory, glutamatergic input onto shared postsynaptic targets (filled triangles) (Wessel et al. 1999; Yuan and Burrell, 2010; Higgins et al. 2013). In this study, the synapses used were the P- or N_poly-to-longitudinal motor neuron (L) and the N_mech-to-anterior pagoda neuron (AP). Tonic, GABAergic inputs (open circles) have an inhibitory effect on P cell synapses but have an excitatory effect on both N_poly- and N_mech-cell synapses (Wang et al. 2015). B: locations of identified neurons in the Hirudo central nervous system (CNS) used in these experiments (modified from Muller et al. 1981). Both the ventral (left) and dorsal (right) aspects of a single mid-body ganglion are shown. Neurons used in this study have been outlined; N1 = N_poly and N2 = N_mech.

Both the P and N_poly cells make monosynaptic, glutamatergic input onto a shared postsynaptic target, the longitudinal (L) motor neuron (Fig. 1A) (Baylor and Nicholls 1969; Yuan and Burrell 2010). Furthermore, both synapses contribute to the neural circuit that mediates whole body shortening, a defensive withdrawal reflex (Shaw and Kristan 1995). Finally, both synapses are modulated by endocannabinoids, depressing N-cell synapses and potentiating P-cell synapses for at least 1 h (Yuan and Burrell 2010, 2012; Higgins et al. 2013). Since it is known which synapses are modulated by endocannabinoids and which of these synapses are hyperpolarized vs. depolarized by GABA, it is possible to examine the potential role of Cl⁻ homeostasis on endocannabinoid-mediated plasticity in Hirudo.

The Hirudo CNS possesses the two main endocannabinoids found in the mammalian nervous system, 2-arachydonoyl glycerol (2-AG) and anandamide (AEA) (Matias et al. 2001). Like other protostomal invertebrates, Hirudo lacks orthologues of the cannabinoid receptors CB1 and CB2 (Elphick 2012). Instead, modulation by endocannabinoids in Hirudo is mediated by a transient potential receptor vanilloid (TRPV)-like receptor (Summers et al. 2014) that mediates both endocannabinoid-dependent long-term depression (eCB-LTD) in N_poly synapses (Yuan and Burrell 2010, 2013b) and endocannabinoid-induced disinhibition of P synapses (Higgins et al. 2013). N_poly cells possess TRPV channels that are believed to be activated by 2-AG, resulting in eCB-LTD (Yuan and Burrell 2010). P cells lack TRPV channels and it is thought that 2-AG acts on unknown GABAergic neurons that are TRPV positive and undergo eCB-LTD resulting in disinhibition of the P-cell synapses (Higgins et al. 2013; Summers et al. 2014). It is well-established that endocannabinoids can activate mammalian TRPV1 channels (De Petrocellis et al. 2001; Qin et al. 2008; Zygmunt et al. 2013) and that endocannabinoid activation of TRPV1 can elicit LTD of excitatory and inhibitory synapses, the latter resulting in disinhibition of neural circuits (Gibson et al. 2008; Maione et al. 2009; Chavez et al. 2010; Puente et al. 2011; Kim et al. 2012; Brown et al. 2013). The cellular features of eCB-LTD in Hirudo exhibit considerable conservation with at least the presynaptic forms of TRPV-mediated LTD in mammalian synapses (Gibson et al. 2008; Jensen and Edwards 2012) so the effects of Cl⁻ homeostasis on endocannabinoid-mediated plasticity in Hirudo are relevant to function in the mammalian CNS. In this study, differences in Cl⁻ gradients were found to play a critical role in determining how endocannabinoids modulate synapses from three distinct cell types of neurons (P, N_poly, and N_mech).

MATERIALS AND METHODS

Animal preparation.

Hirudo verbana (3 g) were obtained from commercial suppliers (Leeches USA, Westbury, NY and Niagara Leeches, Cheyenne, WY) and maintained in artificial pond water (0.52 g/l H₂O Instant Ocean, Aquarium Systems) on a 12-h light-dark cycle at 15°C. Individual ganglia were dissected and pinned in a recording chamber with constant perfusion of normal Hirudo saline (110 mM NaCl, 5 mM NaOH, 4 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, pH 7.4) at ∼1.5 ml/min. Individual neurons were identified based on their position within the ganglion, size, and electrophysiological properties. Each ganglion contains two bilateral pairs of nociceptive (N) and pressure-sensitive cells (P) and three pairs of light touch-sensitive neurons (T) (Fig. 1B). The N cells are further divided into a lateral N pair that are polymodal nociceptors (N_poly) and a medial pair that are mechanical nociceptors (N_mech) (Nicholls and Baylor 1968; Blackshaw et al. 1982; Pastor et al. 1996). L-motor neuron identification was confirmed by recording from the electrically coupled contralateral L-motor neuron and observing synchronous activity (Stuart 1970). The L cells are located laterally on the dorsal side of the ganglion and the afferent cells are located on the ventral surface (Fig. 1B). However, it is possible to record from both the L and the N_poly or P cells from the dorsal side of the ganglion given the lateral positions of the cell bodies from both these primary afferents (Fig. 1B). It is not possible to record from the N_mech cell when the ganglion is dorsal side up because this neuron is located medially on the ventral side of the ganglion. Therefore, synaptic recordings were made from the N_mech and AP neurons, which are both located on the ventral surface of the ganglion (Fig. 1B). Where they have been compared, the properties of N- or P-to-AP synapses appear to be identical to other synaptic connections made by these sensory cells (Gaudry and Kristan 2009; Higgins et al. 2013).

Drug application.

Drugs used for each experiment were kept as frozen aliquot solutions and then diluted to their final concentration in normal saline just before each respective experiment. 2-Arachidonoylglycerol (2-AG), anandamide (AEA), capsaicin (CAP), SB 366791 (SB), tetrahydrolipstatin (THL, also known as Orlistat), bumetanide (BUM), and VU 0240551 (VU) stocks were made in dimethyl sulfoxide (DMSO). 2-AG, AEA, and SB were obtained from Tocris (Ellisville, MO), while THL, BUM, VU, CAP, and DMSO were obtained from Sigma-Aldrich (St. Louis, MO).

Electrophysiology.

Techniques used in this study have been described in detail previously (Yuan and Burrell 2013b). Briefly, current clamp (bridge balanced) intracellular recordings were carried out using sharp glass microelectrodes (tip resistance: 20–35 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 3 M potassium acetate (KAc), which produced no obvious effects on synaptic inhibition in Hirudo neurons (Burgin and Szczupak 2003; Ivanov and Calabrese 2006). Manual micropositioners (Model 1480; Siskiyou, Grants Pass, OR) were used to impale individual neurons during experiments. Current pulses were delivered to electrodes using a multichannel 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).

Pretest recordings of the excitatory postsynaptic potentials (EPSPs) were conducted and then the intracellular electrodes were withdrawn. The ganglion was then treated with 2-AG (100 μM), AEA (100 μM), or capsaicin (1 μM) for 15 min, followed by a 60-min consolidation/washout period. The pre- and postsynaptic cells were then repenetrated and the posttest recordings of EPSPs were taken. Prolonged recordings with these sharp microelectrodes were not carried out because this results in rundown of EPSP amplitudes, likely due to damage to the recorded cell. For experiments using SB 366791, the pretest recording were conducted after bath application of SB 366791 for 15 min and then coapplication of SB 366791 and 2-AG, AEA, or capsaicin, and after a 60-min consolidation, posttest recordings of EPSPs were taken. For experiments using bumetanide (25 μM) or VU 0240551 (100 μM), the pretest recordings were conducted after bath application of either bumetanide or VU 0240551 for 15 min and then coapplication of bumetanide or VU 0240551 and 2-AG, AEA, or capsaicin, and after a 60-min consolidation, posttest recordings of EPSPs were taken (bumetanide and VU 0240551 were applied throughout the whole experiments). For experiments using THL (10 μM), pretest recordings of EPSPs were performed, then bath application of THL 15 min and then coapplication of THL and 2-AG, AEA, or capsaicin, and after a 60-min consolidation, posttest recordings of EPSPs were taken. All treatment groups were tested in parallel with a vehicle control group (0.01% DMSO).

For experiments testing activity-elicited synaptic modulation, low-frequency stimulation (LFS) of T cell was conducted by a well-established protocol in which the presynaptic cell was stimulated 900 times at 1 Hz. High-frequency stimulation (HFS) was delivered to the N_poly neuron that consisted of 20 trains delivered at a rate of 0.1 Hz with each train having 10 suprathreshold stimulus pulses at 25 Hz. For both LFS and HFS experiments, EPSPs and recordings were made before (pretest) and 60 min following either LFS or HFS.

To ensure that changes in synaptic transmission were not the result of changes in postsynaptic input resistance (IR) due to either experimental treatment or repenetration of the sharp electrode, IR was monitored during the pre- and posttest recordings by measuring the membrane potential change during a 500-ms, 1-nA negative current pulse. Only stable recordings (<10% change in IR) were included in the final data analysis. During the pre- and posttests, EPSPs were elicited at 0.1 Hz and the peak amplitude was calculated from the average of 5–10 EPSPs recordings.

Statistics.

EPSP amplitude measurements of the pre- and posttest recordings were normalized and presented as means ± SE. Statistical analyses using a one-way ANOVA were performed to determine main effects with Student-Newman-Keuls post hoc tests to confirm the ANOVA results. The analyses were carried out with Sigmaplot. All significance was determined at α level of at least P ≤ 0.05.

RESULTS

Effects of endocannabinoids on three different types of glutamatergic synapses.

First, the effects of the endocannabinoids 2-AG and anandamide (AEA) and the TRPV1 agonist capsaicin (CAP) were tested on the P, N_poly, and N_mech synapses. Recent studies have shown that 2-AG activates TRPV1 channels (Qin et al. 2008; Zygmunt et al. 2013). Consistent with previous findings (Yuan and Burrell 2010; Higgins et al. 2013), bath application of 2-AG (100 μM for 15 min) elicited persistent changes in both synapses when recorded 1 h after treatment; potentiation of the P-to-L synapses and depression of the N_poly-to-L synapses [P cell synapses, Fig. 2, A and B; one-way ANOVA, F_(7,39) = 11.47; P < 0.001; Student-Newman-Keuls post hoc test of 2-AG vs. DMSO, P < 0.01; N_poly synapses, Fig. 2, C and D, one-way ANOVA, F_(7,39) = 11.13; P < 0.001; post hoc 2-AG vs. DMSO, P < 0.01]. In ganglia that were cotreated with 2-AG plus the TRPV1 channel antagonist SB 366791 (SB; 10 μM), no potentiation was observed in the P-cell synapses and no depression was observed in the N_poly synapses, again consistent with previous studies (Higgins et al. 2013; Yuan and Burrell 2013b) (P synapses, Fig. 2B, 2-AG vs. SB + 2-AG, P < 0.05; N_poly synapses, Fig. 2D, 2-AG vs. SB + 2-AG, P < 0.01).

Fig. 2.

Endocannabinoid-mediated synaptic plasticity in P, N_poly, and N_mech synapses. A, C, and E, top: example of changes in excitatory postsynaptic potential (EPSP) amplitude elicited by either the P (A), N_poly, (C) or N_mech neuron (E; scale bar = 20.0 mV/50 ms), following or 2-arachydonoyl glycerol (2-AG) treatment (middle; scales bars = 2.0 mV/50 ms) or 2-AG + SB 366791 (SB) (bottom; scale bars = 2.0 mV/50 ms for P and N_poly and 1.5 mV/50 ms for N_mech). Black and gray traces are pre- and posttest EPSPs (averaged from 5–10 recordings), respectively. B, D, and F: mean ± SE EPSP (shown as %pretest EPSP amplitude) following vehicle, endocannabinoid, or capsaicin (CAP) treatment with or without SB. 2-AG, anandamide (AEA), and CAP elicited potentiation in P synapses (B) but depressed N_poly (D) and N_mech (F) synapses. For all 3 synapse types, SB blocked synaptic changes elicited by 2-AG, AEA, and CAP and no change in EPSP amplitude was observed in the vehicle control (0.01% DMSO) or SB only groups. *Statistically significant differences based on one-way ANOVA with Student-Newman-Keuls post hoc test (see results); n = 5 (different animals) for all treatment and control groups.

The effects of AEA, which is a well-known endocannabinoid that activates TRPV channels (De Petrocellis et al. 2001), have not been tested in Hirudo. AEA (100 μM) produced the same effects as 2-AG; potentiating P synapses and depressing N_poly synapses (P synapses, Fig. 2B; AEA vs. DMSO, P < 0.001; N_poly synapses, Fig. 2D, AEA vs. DMSO, P < 0.005). The TRPV1 antagonist SB 366791 (10 μM) blocked AEA-mediated potentiation of P synapses and depression of N_poly synapses (P synapses, Fig. 2B, SB only vs. SB + AEA, P > 0.05; N_poly synapses, Fig. 2D, SB only vs. SB + AEA, P > 0.05).

As an additional test of TRPV involvement, the effects of the TRPV1 activator CAP (1 μM) were tested. The effects of capsaicin were identical to those of 2-AG and AEA, potentiation of P-cell synapses and depression of N_poly-cell synapses (P synapses, Fig. 2B, DMSO vs. CAP, P < 0.05; N_poly synapses, Fig. 2D, DMSO vs. CAP, P < 0.001). SB 366791 (10 μM) blocked capsaicin-induced modulation of P and N_poly synapses (P synapses, Fig. 2B, SB only vs. SB + CAP, P > 0.05; N_poly synapses, Fig. 2D, SB only vs. SB + CAP, P > 0.05). Together these findings support the idea that endocannabinoid/TRPV-mediated signaling elicits opposing forms of plasticity in these two synapse types in Hirudo.

The effects of endocannabinoids were also examined on synapses made by the previously untested mechanical N cell (N_mech), which, like the P cell, lacks TRPV channels (Summers et al. 2014). Surprisingly, 2-AG, AEA, and capsaicin all elicited persistent depression of the N_mech synapses when measured 1 h after treatment [Fig. 2, E and F; one-way ANOVA, F_(7,39) = 19.76; P < 0.001; post hoc 2-AG vs. DMSO control, P < 0.001; AEA vs. DMSO control, P < 0.001; CAP vs. DMSO control, P < 0.001]. 2-AG-, AEA-, and capsaicin-mediated LTD was blocked by SB366791 [Fig. 2F, SB only vs. 2-AG + SB, P > 0.5; SB only vs. AEA + SB, P > 0.05; SB only vs. CAP + SB, P > 0.05], suggesting involvement of the Hirudo TRPV-like channel. How TRPV-activating ligands could affect the N_mech synapse is addressed in subsequent experiments. For all of these synaptic studies, no changes in the postsynaptic input resistance (IR) were observed during the P-to-L, N_poly-to-L and N_mech-to-AP synapses.

Cl⁻ gradients and endocannabinoid modulation.

The role of differences in Cl⁻ gradients in regulating the “sign” of endocannabinoid-mediated plasticity was examined. 2-AG-induced potentiation of P synapses requires GABAergic signaling indicating that endocannabinoids depress GABAergic (likely tonic) input and thereby disinhibiting the synapse (Higgins et al. 2013; Wang et al. 2015). Therefore, the effects of the KCC2 inhibitor on endocannabinoid-mediated potentiation/disinhibition were examined. VU 0240551 at 100 μM produces approximately a −20-mV positive shift of E_Cl in P cells, decreasing the hyperpolarizing effects of GABA and acutely increasing synaptic transmission by this afferent (Wang et al. 2015). Pretreatment with VU 0240551 followed by application of endocannabinoids or capsaicin (15 min) blocked 2-AG, AEA, and capsaicin-induced potentiation while pretreatment with VU 0240551 by itself had no effect [Fig. 3, A and B; one-way ANOVA, F_(7,39) = 9.02; P < 0.001; post hoc VU vs. VU + 2-AG, P > 0.05; VU vs. VU + AEA, P > 0.05; VU vs. VU + CAP, P > 0.05]. These results indicate that disrupting Cl⁻ export interfered with endocannabinoid-mediated disinhibition of the P-cell synapses.

Fig. 3.

Manipulating Cl⁻ homeostasis alters endocannabinoid-mediated synaptic plasticity. A, C, and E, top: example of changes in EPSP amplitude elicited by either the P (A), N_mech (C), or N_poly neuron (E; scale bar = 20.0 mV/50 ms), following AEA (middle; scale bar = 2 mV/50 ms for P and 1 mV/50 ms for N_poly and N_mech), AEA + VU 0240551 (VU) (A), or AEA + bumetanide (BUM) (C and E) (bottom; scale bar = 2.0 mV/50 ms for P and N_poly and 1.0 mV/50 ms for N_mech). Black and gray traces are pre- and posttest EPSPs, respectively. B, D, and F: mean ± SE EPSP (%pretest amplitude) following vehicle, endocannabinoid, or CAP treatment with or without VU or BUM. B: in P synapses, 2-AG-, AEA-, and CAP-elicited potentiation was blocked by VU. D: BUM did not affect endocannabinoid- or CAP-mediated depression in N_poly synapses. F: in N_mech synapses, endocannabinoids- and CAP-mediated depression was blocked by BUM. For all three synapse types, no change in EPSP amplitude was observed in the vehicle control (0.01% DMSO) or in the VU- or BUM-only groups. *Statistically significant differences based on one-way ANOVA with Student-Newman-Keuls post hoc test (see results); n = 5 (different animals) for all treatment and control groups.

Next, the effects of disrupting Cl⁻ import on eCB-LTD in the N_poly synapses were examined. As already stated, GABA has an excitatory effect on the N_poly cell and its synapse (Wang et al. 2015). If endocannabinoid is depressing GABAergic input sufficiently to disinhibit P-cell synapses, then it is possible that depression of this excitatory GABAergic input will “disexcite” N_poly synapses. Pretreatment with the NKCC1 inhibitor bumetanide has been shown to decrease the E_Cl in the N_poly cell and this was accompanied by a decrease in N_poly synaptic transmission, presumably by decreasing tonic depolarization by GABA (Wang et al. 2015). In the present experiments, 25 μM bumetanide were used, which can decrease GABA's depolarizing effect on N_poly. Pretreatment with bumetanide did not prevent 2-AG-, AEA-, or capsaicin-mediated LTD of N_poly synapses [Fig. 3, C and D; one-way ANOVA F_7,39 = 4.89; P < 0.001; post hoc BUM vs. BUM + 2-AG, P < 0.05; BUM vs. BUM + AEA, P < 0.05; BUM vs. BUM + CAP, P < 0.05]. These results indicate that even though bumetanide decreases depolarizing GABAergic input, this does not contribute to eCB-LTD of N_poly synapses and that the observed depression is due entirely to the direct activation of the TRPV-like channel in the presynaptic N_poly neuron (Yuan and Burrell 2010, 2012, 2013a,b). However, the depolarized E_Cl may “protect” these synapses from the disinhibition that would normally be observed in neurons that receive GABAergic input and had low intracellular Cl⁻ levels.

Finally, the effects of bumetanide were tested on endocannabinoid-mediated depression of the N_mech synapses. Recall that N_mech cells also have an elevated/depolarized E_Cl, like N_poly (Wang et al. 2015) and N_mech synapses undergo depression when treated with 2-AG, AEA, or capsaicin (Fig. 2, E and F) even though they lack TRPV channels (Summers et al. 2014). Bumetanide treatment (25 μM) completely blocked 2-AG-, AEA-, and capsaicin-mediated LTD of N_mech synapses [Fig. 3, E and F; one-way ANOVA, F_7,39 = 11.19; P < 0.001; post hoc BUM vs. BUM + 2-AG, P > 0.05; BUM vs. BUM + AEA, P > 0.05; BUM vs. BUM + CAP, P > 0.05]. These results suggest that endocannabinoid-mediated modulation of the N_mech synapses utilizes the same mechanisms observed in P synapses but with the opposite effect. That is, endocannabinoids depress tonic GABAergic input to the N_mech, but since E_Cl is depolarized relative to resting potential in these afferents, this results in a decrease in excitatory modulation of the N_mech synapses or “disexcitation.” This represents a new and entirely novel mechanism of endocannabinoid modulation of excitatory synapses. Again, no statistically significant changes in postsynaptic input resistance (IR) were observed during the P-to-L, N_poly-to-L, and N_mech-to-AP synaptic recordings.

Patterns of activity that elicit endocannabinoid/TRPV-mediated synaptic modulation.

Given that exogenous activation of endocannabinoids elicits three distinct mechanisms of synaptic plasticity, two of which were directly influenced by differences in Cl⁻ gradients, what patterns of circuit activity could elicit these forms of synaptic change? Previously, low-frequency stimulation (LFS; 1 Hz for 900 s) of the touch-sensitive afferents (T cells) was found to elicit endocannabinoid/TRPV-dependent depression of N_poly synapses (Yuan and Burrell 2010, 2013b) and this LTD was blocked by SB 366791 or by an inhibitor of the DAG lipase responsible for 2-AG synthesis, tetrahydrolipstatin (THL) (Lee et al. 1995). These experiments were repeated and, consistent with previous results, T-cell LFS heterosynaptically depressed N_poly synapses. However, T-cell LFS did not affect P or N_mech synapses (Fig. 4, t-test, N_poly synapses, P < 0.01; P and N_mech synapses, P > 0.05). LFS-induced depression of N_poly synapses was still observed following inhibition of Cl⁻ import by bumetanide (25 μM) [Fig. 5, A and B; one-way ANOVA, F_(3,19) = 27.74; P < 0.001; post hoc LFS vs. no stim control, P < 0.001; BUM only vs. BUM + LFS, P < 0.001]. However, the level of LFS-induced depression in bumetanide-treated synapses was significantly less than that observed in the LFS-only group (Fig. 4, A and B: LFS vs. BUM + LFS, P < 0.05). This result suggests that in addition to the direct activation of TRPV channels on N_poly neurons, T cell LFS elicits a decrease in an excitatory input that requires a Cl⁻ conductance (presumably a GABAergic input) that contributes to eCB-LTD of the N_poly synapse, albeit to a small degree.

Fig. 4.

Effects of T cell with low-frequency stimulation (LFS) on synaptic transmission by the 3 synapses in this study. While LFS did depressed the N_poly synapses as previously observed (Yuan and Burrell 2010), no change was observed in either the P or N_mech synapses. No effect was observed in saline control (no stim) among 3 synapses. Data are presented as the mean ± SE percentage of the pretest EPSP (*P < 0.01, t-test); n = 5 (different animals) for all treatment and control groups.

Fig. 5.

Endocannabinoid-mediated synaptic modulation elicited by either LFS or high-frequency stimulation (HFS) and the role of Cl⁻ homeostasis. A, C, and E, top: example of changes in EPSP amplitude elicited by either the N_poly (A), P (C). or N_mech neuron (E; scale bars = 20.0 mV/50 ms), following LFS (A) or HFS (C and E; scale bars = 2.0 mV/50 ms for N_poly and P and 1.0 mV/50 ms for N_mech), LFS + BUM (A) or HFS + tetrahydrolipstatin (THL) (C, E;) (bottom; scale bars = 2.0 mV/50 ms). Black and gray traces are pre- and posttest EPSPs, respectively. B, D, and F: mean ± SE EPSP (%pretest amplitude) following LFS or HFS with or without SB, THL, VU, or BUM. B: in N_poly synapses, LFS elicited synaptic depression. This depression was not blocked by BUM. D: in P synapses, HFS elicited synaptic potentiation and this potentiation was blocked by SB, THL, and VU. (F) In N_mech synapses, HFS elicited synaptic depression and this depression was blocked by SB, THL, and BUM. For all 3 synapse types no change in EPSP amplitude was observed in the saline (no stim) or in the SB-, THL-, VU-, or BUM-only groups. *Statistically significant differences based on one-way ANOVA with Student-Newman-Keuls post hoc test (see results); n = 5 (different animals) for all treatment and control groups.

If T-cell LFS does not affect the P or N_mech synapses, then what patterns of activity are effective in inducing endocannabinoid-mediated modulation of these pathways? Given that noxious stimuli produce sensitization of defensive withdrawal reflexes elicited by P-cell activity (Lockery and Kristan 1991; Ehrlich et al. 1992; Modney et al. 1997; Burrell et al. 2001; Burrell and Sahley 2005), the possibility that high-frequency stimulation (HFS) of the N_poly cell could elicit potentiation of P synapses was examined. HFS (20 trains at 0.1 Hz with 10 pulses at 25 Hz per train) of N_poly significantly potentiated P synapses when retested 1 h following later [Fig. 5, C and D; one-way ANOVA, F_(7,39) = 12.87; P < 0.001; post hoc HFS vs. no stim control, P < 0.001]. Furthermore, this potentiation was blocked by either SB 366791 (10 μM; Fig. 5D, SB only vs. SB + HFS, P > 0.05) or THL (10 μM; Fig. 5D, THL only vs. THL + HFS, P > 0.05), indicating that increased signaling was 2-AG and TRPV dependent. When the Cl⁻ export inhibitor VU 0240551 (100 μM) was applied during HFS stimulation, HFS-induced potentiation was blocked (Fig. 5D, VU only vs. VU + HFS, P > 0.05), indicating that normal Cl⁻ export was required for this activity-dependent form of endocannabinoid-mediated plasticity, identical to what was observed during exogenous application of endocannabinoids.

HFS of the N_poly neuron depressed the N_mech synapses [Fig. 5, E and F; one-way ANOVA, F_(7,39) = 5.29; post hoc HFS vs. no stim control, P < 0.001]. This LTD was also blocked by SB 366791 (Fig. 5F, SB only vs. SB + HFS, P > 0.5) or THL (THL only vs. THL + HFS, P > 0.05), indicating a requirement for 2-AG synthesis and TRPV activation. HFS-elicited depression of N_mech synapses was blocked by bumetanide treatment, implying that normal Cl⁻ import was also required for this activity-dependent form of endocannabinoid-mediated plasticity (Fig. 5F, BUM only vs. BUM + HFS, P > 0.1). Again, identical to the results observed during exogenous application of endocannabinoids. No statistically significant changes in postsynaptic IR were observed during the P-to-L, N_poly-to-L, or N_mech-AP synapse recordings.

DISCUSSION

This study provides the first experimental confirmation of the hypothesis that differences in Cl⁻ gradients can control whether or not synaptic circuits are sensitive to endocannabinoid-mediated disinhibition (Christie and Mallet 2009). In addition, the role of Cl⁻ gradient control as a contributor to synaptic plasticity has been extended by the discovery of a completely novel form of neuromodulation observed in the N_mech synapses in which endocannabinoid-mediated depression is blocked by inhibition of Cl⁻ import. These findings may explain how endocannabinoids in the spinal cord can have both pro- and antinociceptive effects (Morisset and Urban 2001; Liang et al. 2004; Hohmann et al. 2005; Christie and Mallet 2009; Pernia-Andrade et al. 2009; Kato et al. 2012).

The pattern of observed endocannabinoid-mediated synaptic modulation was not simply a product of exogenous drug application but could also be produced following distinct patterns of afferent activity. In P synapses, HFS of N_poly elicited potentiation that was blocked by inhibition of 2-AG synthesis or TRPV activation. These effects were observed even though P cells lack TRPV channels (Pastor et al. 1996; Summers et al. 2014). It is proposed that endocannabinoids depress tonic GABAergic input to the P cell presynaptic terminals, disinhibiting those synapses so that spike-evoked neurotransmitter release is now enhanced (Fig. 6A). P synapses are thought to receive tonic inhibitory GABAergic input (Wang et al. 2015) and endocannabinoid-mediated potentiation of P synapses requires functioning GABA_A receptors (Higgins et al. 2013). In this study, inhibition of Cl⁻ export by the KCC2 inhibitor VU 0240551 also blocked endocannabinoid-mediated potentiation by disrupting the Cl⁻ gradient, confirming that the role of GABA is based on its activation of ligand-gated Cl⁻ channels and not metabotropic GABA receptors.

Fig. 6.

Diagram illustrating the three mechanisms of which endocannabinoids modulated synaptic transmission. All sensory neurons receive tonic GABAergic input from an unknown interneuron that has an inhibitory effect on P synapses and an excitatory on N_poly and N_mech synapses based on (Wang et al. 2015). The N_poly cell is of transient potential receptor vanilloid (TRPV) positive, while the P and N_mech neurons are TRPV negative (Pastor et al. 1996; Summers et al. 2014). A: endocannabinoids depress inhibitory GABAergic input to the P cell, increasing P glutamatergic transmission via a disinhibition mechanism that is dependent on activity of a KCC2-like Cl⁻ exporter. B: endocannabinoids depress exhibitory GABAergic input to the N_mech cell, decreasing N_mech glutamatergic transmission via a disexcitation mechanism that is dependent on activity of a NKCC1-like Cl⁻ importer. C: endocannabinoids depress the N_poly synapses directly via activation of their TRPV channels (Yuan and Burrell, 2010, 2012, 2013a,b). Although endocannabinoids presumably depress excitatory GABAergic input to the N_poly neurons, this did not appear to affect N_poly synapses.

Endocannabinoid-mediated modulation of N_mech synapses utilizes a mechanism that is identical to the P synapses except that the sign of the synaptic change is reversed due to N_mech having elevated levels of intracellular Cl⁻. In N_mech neurons, which also lack TRPV receptors, LTD can be induced by exogenous application of endocannabinoids (or capsaicin) and by HFS of the N_poly neuron that is 2-AG and TRPV dependent. Inhibition of Cl⁻ import by the NKCC1 blocker bumetanide prevented endocannabinoid/TRPV-mediated depression elicited by exogenous drug application or HFS. From these results it is inferred that endocannabinoids depress a tonic excitatory/depolarizing input that requires Cl⁻ efflux in the N_mech cell (Wang et al. 2015) and that a decrease in this excitatory input leads to depression of the N_mech synapse (Fig. 6B). It is thought that this excitatory, Cl⁻-based current represents a GABAergic input, but this has not been directly tested yet. Although the identities of at least two GABAergic neurons are known (Cline 1983), it is not known whether these cells are the source of the tonic GABAergic input. Future experiments will probe the GABA signaling component of this modulatory process.

2-AG- and capsaicin-elicited LTD of N_poly cell synapses has already been shown (Yuan and Burrell 2010, 2012, 2013a), but the capacity of AEA to also produce depression is a new discovery and is relevant given the presence of both 2-AG and AEA in the Hirudo CNS (Matias et al. 2001). That T-cell LFS can elicit LTD in the N_poly synapse that is 2-AG and TRPV dependent has also been previously reported (Yuan and Burrell 2010, 2013b). N_poly cells possess a TRPV-like receptor (Summers et al. 2014) so the effects of endocannabinoids modulation are due to the direct activation of this TRPV channel (Fig. 6C). As in N_mech neurons, N_poly cells are excited by GABA due to the high intracellular Cl⁻ and bumetanide treatment alone can reduce this response to GABA and acutely disexcite N_poly synapses (Wang et al. 2015). Although manipulation of the Cl⁻ gradient by bumetanide did not alter depression following exogenous application of 2-AG, AEA, or capsaicin, bumetanide treatment did cause a small but significant decrease in eCB-LTD elicited by T-cell LFS. Therefore, eCB-LTD in N_poly synapses is primarily due to direct activation of TRPV channels on the presynaptic, N_poly cell, with the indirect, disexcitation mechanism making a relatively small contribution. However, the depolarized E_Cl may have an additional role in that it may “protect” the N_poly and from endocannabinoid-mediated disinhibition that would occur in neurons that receive GABAergic input and possess the low intracellular Cl⁻ levels that are more commonly observed in adult cells.

It is interesting that different forms of endocannabinoid-mediated modulation were elicited by distinct patterns of afferent activity. T-cell LFS elicited endocannabinoid-mediated plasticity only in N_poly synapses, while N_poly HFS produced endocannabinoid-mediated changes in the P and N_mech synapses. HFS does elicit homosynaptic LTP in N_poly, but this is mediated by NMDA receptors and not endocannabinoid/TRPV signaling (unpublished data). This suggests that the CNS has distinct endocannabinoid synthesizing sites that are selectively activated by distinct patterns of afferent stimulation. Previous studies in which the DAG lipase inhibitor THL was injected into either the presynaptic N_poly or postsynaptic L motor neuron have shown that the latter is the site of 2-AG synthesis activated by LFS (Yuan and Burrell 2013a,b), although this has not been confirmed with nonpharmaceutical approaches. The site of endocannabinoid synthesis activated by HFS is not known. These findings are similar to endocannabinoid modulation of striatal synapses that also exhibit long-lasting depression or disinhibition depending on the pattern of afferent activity (Adermark and Lovinger 2009).

These results suggest the presence of distinct endocannabinoid “modules”; combinations of excitatory and inhibitory circuits along with endocannabinoid-releasing neurons, which are accessed by different patterns of activity and have distinct functional roles. For example, eCB-LTD in N_poly synapses is elicited by T-cell LFS and this translates into a depression of defensive withdrawal behaviors elicited by N_poly (Yuan and Burrell 2013b). This strongly resembles gate control of pain in which stimulation of nonnociceptive (Aβ) afferents can diminish nociceptive signaling (Melzack and Wall 1965). Repetitive Aβ stimulation in mammals is known to induce LTD in nociceptive synapses (Wallin et al. 2003; Sdrulla et al. 2015) that is similar to the depression of the Hirudo N_poly synapses. The endocannabinoid-mediated depression observed here may represent a previously unrecognized form of neuroplasticity contributing to gate control.

HFS of N_poly, on the other hand, elicited endocannabinoid-mediated potentiation/disinhibition of P synapses and is remarkably similar to TRPV-mediated disinhibition that contributes to sensitization of nonnociceptive signaling (allodynia) in mammals (Kim et al. 2012). This endocannabinoid-mediated potentiation may represent a heretofore unrecognized form of heterosynaptic facilitation that contributes to central sensitization of primary somatosensory synapses and provides a potential activity-dependent mechanism by which disinhibition allows nonnociceptive afferents to become recruited into the nociceptive circuitry (Torsney and MacDermott 2006; Lu et al. 2013).

HFS also resulted in endocannabinoid-mediated depression of N_mech synapses, which would seem to oppose central sensitization. What this would imply is that a noxious stimulus that produces high-frequency activity in a nociceptive afferent (such as N_poly) can elicit depression in other nociceptive pathways (N_mech in this case). This has been observed before and referred to as diffuse nociceptive inhibitory control (DNIC) or conditioned inhibition of pain (Ernst et al. 1986; Yarnitsky 2010). The physiological mechanisms that mediate DNIC are poorly understood although descending modulatory input has been implicated. Based on the findings in the present study, it is proposed that endocannabinoid-mediated disexcitation of nociceptive synapses may represent an additional form of neuromodulation contributing to DNIC. Obviously, further studies are required to support or refute this hypothesis.

Although studied in the context of somatosensory afferents, the cellular features of endocannabinoid-mediated synaptic modulation observed here, the role of differences in Cl⁻ gradients and the patterns of afferent activity that initiate specific endocannabinoid modules have broad applicability to circuits throughout the CNS. Endocannabinoid-mediated modulation via TRPV channels is observed in a wide range of neural circuits including those contributing to cognition, stress, addiction, fight-or-flight behaviors, and, of course, nociception (see review by Edwards 2014), although these findings would also be relevant for neuromodulation mediated by CB1 or CB2 receptors. There are also an increasing number of examples in which variability of Cl⁻ gradients between adult neurons is observed in both vertebrates and invertebrates (Duchen 1986; Chavas and Marty 2003; Ruiz et al. 2003; Norekian and Malyshev 2005; Cheung et al. 2006; Szabadics et al. 2006; Gilbert et al. 2007; Pfeiffer et al. 2009; Pugh and Jahr 2011; Haam et al. 2012). These include examples in which neurons with different Cl⁻ gradients are found within the same or overlapping neural circuits (Gilbert et al. 2007; Haam et al. 2012) and neurons in which different cellular compartments in neurons exhibit different Cl⁻ gradients (Khirug et al. 2008).

The findings in this study represent a new and important mechanism for both the regulation of neuromodulatory processes and the organization of these processes at the microcircuit level. The role of differences in Cl⁻ gradient observed here is unlikely to be limited to only endocannabinoid-based modulation but also impacts other modulatory transmitters known to depress both glutamatergic and GABAergic/glycinergic synapses, e.g., opioids (Lau and Vaughan 2014), glutamate (Zak et al. 2015), or adenosine (Morairty et al. 2004). Consequently, the integration of Cl⁻ gradients into adult neuromodulatory processes represents a new avenue to study neural and behavioral plasticity throughout the CNS.

GRANTS

This work was funded by National Science Foundation IOS-1051734 (to B. D. Burrell) and the University of South Dakota Division of Basic Biomedical Sciences.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Y.W. performed experiments; Y.W. analyzed data; Y.W. and B.D.B. interpreted results of experiments; Y.W. prepared figures; Y.W. drafted manuscript; B.D.B. conception and design of research; B.D.B. edited and revised manuscript; B.D.B. approved final version of manuscript.