|Home | About | Journals | Submit | Contact Us | Français|
Hippocampal interneurons release γ-aminobutyric acid (GABA) and produce fast GABAA- and slow GABAB- inhibitory postsynaptic potentials (IPSPs). The regulation of GABAB eIPSPs or the interneurons that produce them are not well understood. In addition, while both µ-opioid receptors (µORs) and cannabinoid CB1R receptors (CB1Rs) are present on hippocampal interneurons, it is not clear how these two systems interact.
This study tests the hypotheses that: (1) all interneurons can initiate both GABAA and GABAB inhibitory postsynaptic potentials; (2) GABAB responses are insensitive to mGluR-triggered, endocannabinoid (eCB)-mediated inhibitory long-term depression (iLTD); (3) GABAB responses are produced by interneurons that express µOR; and (4) CB1R-dependent and µOR-dependent response interact.
Pharmacological and electrophysiological approaches were used in acute rat hippocampal slices. High resistance microelectrode recordings were made from pyramidal cells, while interneurons were stimulated extracellularly.
GABAB responses were found to be produced by interneurons that release GABA via either presynaptic N-type or P/Q-type calcium channels but that they are insensitive to suppression by eCBs or eCB-mediated iLTD. GABAB IPSPs were sensitive to suppression by a µOR agonist, suggesting a major source of GABAB responses is the µOR-expressing interneuron population. A small eCB-iLTD (10% eIPSP reduction) persisted in conotoxin. eCB-iLTD was blocked by a µOR agonist in 6/13 slices.
GABAB responses cannot be produced by all interneurons. CB1R or µOR agonists will differentially alter the balance of activity in hippocampal circuits. CB1R- and µOR-mediated responses can interact.
The hippocampus hosts a wide variety of GABAergic inhibitory interneurons, which are classified according to their functional, anatomical, or molecular properties (for reviews, see Freund and Buzsaki 1996; Freund 2003; Maccaferri and Lacaille 2003). GABA activates two classes of postsynaptic receptor in the central nervous system. The ionotropic GABAA receptor-channel conducts Cl− and bicarbonate ions (Bormann 1988). The metabotropic GABAB receptor principally activates a G-protein coupled, inwardly rectifying potassium (GIRK) channel (Andrade et al. 1986; Bowery 1989). Presynaptically, the GABAB receptor modulates transmitter release by activating GIRKs and inhibiting voltage-gated calcium channels (VGCCs; Andrade et al. 1986; Nicoll 2004 for review). A persistent question has been whether or not both GABAA and GABAB responses are typically induced by a given GABAergic interneuron under normal conditions, or only when there is a high probability of release or a reduction of GABA uptake (e.g., Solis and Nicoll 1992). Whether or not all interneurons can produce GABAB responses is unknown (see Nurse and Lacaille 1997) and modulation of GABAB responses is not well understood.
A basic distinction among interneurons is that they release GABA via different classes of VGCCs. Some hippocampal interneurons release GABA only by activation of presynaptic ω-conotoxin GVIA (‘conotoxin’)-sensitive, N-type VGCCS, and others only by ω-agatoxin IVA (‘agatoxin’)-sensitive, P/Q-type VGCCs (Lambert and Wilson 1996; Poncer et al. 1997). Lenz and Alger (1999) found that DSI, the retrograde signal process now known to be mediated by eCBs (Wilson and Nicoll 2001; Ohno-Shosaku et al. 2001; Alger 2002, for review), was not affected by agatoxin but was abolished by conotoxin. Wilson et al. (2001) recorded from synaptically coupled, interneuron-pyramidal cell pairs, and showed directly that endocannabinoids (eCBs) inhibit release only from interneurons that release via N-type channels.
Functional distinctions between interneuronal subtypes are less clear than morphological or neurochemical distinctions. It is not fully understood how different interneurons are regulated, or how they interact with each other, although various hypotheses are emerging (e.g., Buzsaki 2002; Gillies et al. 2002; Hefft and Jonas 2005; Glickfeld and Scanziani 2006; Karson et al. 2008). Hoffman and Lupica (2000) showed that the GABAA (but not the GABAB) component of an evoked inhibitory postsynaptic current was sensitive to an exogenous cannabinoid agonist, while both components were sensitive to the selective µOR agonist, [D-Ala2, NMe-Phe4, Gly-ol5]-enkephalin (DAMGO). Conceivably, CB1-expressing interneurons activate only GABAA receptors, while CB1R-lacking interneurons activate both receptors. In the hippocampus, opioids and cannabinoids seem to be independent of each other (Neu et al. 2007), although activation of µOR causes widespread disinhibition in CA1 (Nicoll et al. 1980; McQuiston and Saggau 2003).
Functional interactions between conotoxin- and agatoxin-sensitive interneurons have recently been described (Karson et al. 2008). If functional classes of interneurons correspond to CB1R-expressing, or µOR-expressing interneurons, this would be important, as there are many different reports that cannabinoids and opioids show a bidirectional interaction in some behavioral experiments (see Maldonado and Rodriguez 2002 for review). Little is known about the cellular substrates of opioid–cannabinoid interaction, although work on the nucleus accumbens (Schoffelmeer et al. 2006) and cultured cells suggests that µOR-CB1R heterodimers could be responsible (Rios et al. 2007). Interactions among groups of interneurons that are selectively sensitive to inhibition by either cannabinoids or opioids should also be considered.
We have taken a pharmacological and electrophysiological approach to investigating possible functional distinctions among hippocampal interneurons, in particular to the regulation of their GABAergic responses. Specifically, we tested the hypotheses that: (1) all interneurons are capable of initiating both GABAA and GABAB IPSPs, (2) GABAB responses are insensitive to mGluR-triggered, eCB-mediated iLTD, (3) GABAB responses are produced by interneurons that express µOR, and (4) CB1R-dependent and µOR-dependent responses interact. We report that GABAB responses are produced by interneurons that release GABA via presynaptic N-type VGCCs but that they are insensitive to suppression by eCB-mediated iLTD. They are sensitive to suppression by a µOR agonist, suggesting that µOR-expressing interneurons are a source of GABAB responses. Hence, GABAB IPSPs are not produced by all types of interneurons. We also describe for the first time an interaction between µOR and CB1R in the regulation of iLTD that affects only a functionally distinct set of interneurons.
All animal handling protocols were approved by the Institutional Animal Care and Use Committee of the University of Maryland School of Medicine. After the 5- to 7-week-old male Sprague–Dawley (Charles River) rats were deeply sedated with isoflurane and decapitated, we removed the hippocampi and cut them into 400-µm-thick slices in ice-cold bath solution using a Vibratome model 1000 (Technical Products International, St. Louis, MO, USA). The slices were kept at room temperature for >1 h in an interface holding chamber filled with humid 95% O2–5% CO2 before transfer to the submersion-type experimental chamber (Nicoll and Alger 1981) at 30°C.
We recorded from CA1 cells with sharp electrodes (80–150 M Ω) pulled from 1.2 mm o.d. fiber-filled pipettes containing 3 M KCH3SO4 inserted into a holder containing 3 M KCl (pHi=7.2). Responses were recorded with an Axoclamp 2B (Axon Instruments) and data were filtered at 1 kHz, digitized at 4 kHz with a Digidata 1322A (Axon Instruments), and analyzed with pClamp 9.0 routines (Clampex and Clampfit). The bath perfusion included, in mM: 120 NaCl, 3 KCl, 25 NaHCO3, 1 NaH2PO4, 2.5 CaCl2, 2 MgSO4, and 20 glucose (pHo=7.4). 2,3-Dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX, 10uM) and d-(−)-2-Amino-5-phosphonopentanoic acid (D-AP5, 20 µM) were also added to block ionotropic glutamate receptors. Evoked inhibitory postsynaptic potentials (eIPSPs; Fig. 1a) were elicited by 100-µs extracellular stimuli delivered with concentric bipolar stimulating electrodes in stratum (s.) radiatum at 0.05 Hz. Inhibitory long-term depression (iLTD) was induced chemically (10-min bath application of the mGluR agonist, (S)—3,5 dihydroxphenylglycine (DHPG: see Chevaleyre and Castillo 2003) or evoked by synaptic stimulation (two 1-s/100 Hz stimulus trains given 20 s apart, or four 1-s/100-Hz stimulus trains each given 1 s after a 1-s depolarizing dc injection, the trains given 20 s apart).
In some experiments, the GABAB eIPSP was isolated by adding 20 µM bicuculline to the perfusion solution. To compare the differential effect of various treatments on GABAA and GABAB eIPSPs when this was not done, we used two approaches:
The validity of these methods was confirmed by comparing the kinetics of the derived GABAB eIPSP from the pure-evoked GABAB response obtained in the presence of bicuculline. The responses were indistinguishable (cf. Figs. 1b and and2b2b).
NBQX disodium salt, APV, (S)-3,5-DHPG, (S)-(+)-a-Amino-4-carboxy-2-methylbenzeneacetic acid (LY 367385), 2-methyl-6-(phenylethynyl)pyridine hydrochloride (MPEP), N-(Piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (AM 251), and ω-conotoxin GVIA were obtained from Tocris (Ellsville, MO); ω-agatoxin TK was obtained from Peptides International. N-(Piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (SR141716A) was a generous gift of the National Institute on Drug Abuse and was provided by K. Gormley. All other drugs were obtained from Sigma-Aldrich (St. Louis, MO, USA). Drugs were prepared as stock solutions in water and used at 1/1,000 dilution, except for AM251 and SR141716, which were dissolved in dimethyl sulfoxide (DMSO; final concentration of DMSO in the bath was 0.01%). All drugs were applied by bath-application; the flow rate in the chamber was 0.3–0.5 ml/min. Pretreatment of slices (>1 h) with ω-agatoxin GVIA, AM251, or SR141716 was carried out in the incubation chamber described above. SR141716 or AM251 were used interchangeably to block the cannabinoid receptor. ω-agatoxin GVIA and ω-agatoxin TK were used interchangeably to block P/Q type Ca2+ channels. The group I mGluR agonist, DHPG, was applied at 50 µM, which is a functionally maximal concentration in our system.
All group data are represented as mean±SEM. Statistical tests [Student’s t tests and one-way analysis of variance (ANOVA)] were done in Sigma Plot 8 (Systat Software). The presence of iLTD was determined by a significant (p<0.05) suppression of eIPSP 20 min after the washout of DHPG or 20 min after synaptic stimulation. Exponential decay curves were fit to the data in Clampfit 9 (Axon Instruments).
All experiments were carried out in the s. radiatum region of CA1. We found that conotoxin perfusion, 250 nM, reduced the peak GABAA response to 38.8±2.29% of its baseline value (Fig. 1c, n=7), as was reported (Lambert and Wilson 1996; Poncer et al. 2000). The GABAB response was reduced to 19.2±1.27 of its baseline (Fig. 1a and b, see “Materials and methods” for methods of GABAA and GABAB component separation). The difference in maximal suppression of GABAA and GABAB responses is highly significant (p<0.001). The GABAB component also declined somewhat more rapidly than the GABAA response, the half-times for depression being 12.9±0.10 and 13.4±0.79 min, respectively (p<0.05). Agatoxin reportedly reduces evoked GABAA IPSP/Cs by 50–70% (e.g., Lenz and Alger 1999; Wilson et al. 2001), and in three experiments in which 200 nM agatoxin was washed into the bath (Fig. 2a), we confirmed and extended these results. The GABAA eIPSP was reduced to 53.2±1.72% of baseline, and the GABAB component was reduced to 37.7±2.96% of baseline by agatoxin. The GABAB response isolated with bicuculline, 20 µM, was also reduced to 34.2±1.31% of baseline (n=5; Fig. 2b), thus supporting the validity of the methods used to determine the GABAB component. As agatoxin is very expensive and its effects are essentially irreversible over the time course of hours (Wheeler et al. 1994), we pretreated slices with 1 µM agatoxin for >1 h but did not perfuse it in later experiments.
DAMGO, the µOR agonist, reduced both GABAA (to 58.9±1.72% of baseline, n=17; cells treated with either 100 or 500 nM DAMGO, p<0.001, Fig. 3a) and GABAB components of the eIPSPs (DAMGO 500 nM, 44.9±2.06% of baseline, n=5; Fig. 3b). The difference in suppression of GABAA and GABAB responses is significant (p<0.001). Interestingly, naloxone restored the GABAA eIPSP only to 90.8±1.13% of baseline, and the GABAB eIPSP only to 87.7±1.13% of baseline. In both cases, these values are significantly different from their respective controls (p<0.05, ANOVA) suggesting the existence of a µOR-mediated iLTD, although pursuing this possibility was beyond the scope of the present work. We conclude that µORs are located on cells that produce GABAA or GABAB eIPSPs, but more effectively reduce GABAB than GABAA responses.
Long-term, eCB-mediated modulation of synaptic plasticity affects some GABAA responses (Chevaleyre and Castillo 2003); however, there is little evidence about whether or not GABAB responses are similarly regulated. Activation of group I mGluRs suppresses eIPSPs by mobilizing eCBs (Maejima et al. 2001; Varma et al. 2001; Chevaleyre and Castillo 2003). We used the selective group I mGluR agonist, DHPG at 50 µM, to activate the eCB system in place of the lipophilic CB1R agonists, which are more difficult to apply and remove rapidly in acute brain slices. We first replicated the findings of Chevaleyre and Castillo (2003) and observed that a persistent iLTD was induced either with a 10-min application of DHPG (78.7±1.32% of baseline, n=27, p<0.001, Fig. 4b) or synaptic stimulation protocols (e.g., Fig. 4a): two times 100 Hz synaptic stimulation (73.5±2.42 of baseline, n=5, p<0.001, Fig. 4b) or four times 100 Hz synaptic stimulation plus a depolarizing voltage step (e.g., Fig. 4a; to 83.9±1.99% of baseline, n=5, p<0.001, data not shown). The GABAA eIPSP was subject to iLTD, and this iLTD was mediated by eCBs, as CB1R antagonists, AM 251, 3 µM, or SR141716A; 3 µM (Fig. 4c) prevented its establishment (94.5±1.84% of baseline, n=4). iLTD was also blocked by the group I mGluR antagonists LY 367385, 100 µM, plus MPEP, 5 µM, (Fig. 3c; 95.7±1.68% of baseline, n=4), confirming that eCB mobilization is mediated by mGluR I receptors. Unlike Chevaleyre and Castillo (2003), we found that synaptic stimulation given 25 min after DHPG-iLTD induction elicited a further iLTD (75.5±1.94% of baseline, n=5, p<0.001, Fig. 4d) that was comparable to DHPG-iLTD or synaptically induced iLTD alone.
Hoffman and Lupica (2000) reported that the GABAB response was not affected by the synthetic CB1R agonist, WIN55212-2. However, the overlap between WIN55212-2 sensitivity and eCB responses is imperfect, and eCB-iLTD depends on factors besides simple CB1R activation (Ronesi et al. 2004; Edwards et al. 2006). To determine directly if GABAB was affected by eCBs and iLTD, we isolated the GABAB response with 20 µM bicuculline (e.g., Fig. 5a). We then applied DHPG and observed that there was no iLTD of the GABAB response (103.8±6.00% of baseline, n=7, n.s., Fig. 5b; data from control and agatoxin-treated cells pooled, see below) arguing that eCBs do not modulate the GABAB response.
There were two possible deficiencies with this experiment on isolated GABAB eIPSPs: (1) There was no positive control showing that the protocols were sufficient to elicit iLTD; and (2) the mechanism of iLTD is not known, and unknown circuit effects involving a GABAA response could play a role. Hypothetically, blocking GABAA receptors would prevent GABAB- iLTD induction.
To test the susceptibility of GABAB responses to iLTD while avoiding these problems, we did additional experiments without isolating the GABAB responses with bicuculline. After DHPG-iLTD was induced and DHPG washed out, we used the procedures outlined in “Materials and methods,” to examine the GABAB eIPSP before and after iLTD induction, and thus compared GABAA and GABAB eIPSPs in the same experiment. The GABAA response was reduced to 67.7±3.71% of baseline (p<0.001), and the GABAB response was reduced only 97.3±5.46% of baseline by the iLTD induction protocol (n.s., n=11, Fig. 5c; for display purposes the data are shown in Fig. 5c as percent eIPSP reduction, rather than percent baseline eIPSP).
To determine if CB1Rs and µORs are expressed on the same functionally defined interneurons, we examined the effects of DAMGO in the presence of either conotoxin or agatoxin. As shown in Fig. 6, 500 nM DAMGO reduced the agatoxin-insensitive GABAA response (to 64.6±1.95% of baseline, n=5, p<0.001) and the conotoxin-insensitive GABAA response (to 36.2±2.50% of baseline, n=4, p<0.001). There was a significant difference between the DAMGO effects on agatoxin- and conotoxin-insensitive eIPSPs not only during application (p<0.001, t test) but also after applying naloxone 5 µM (102.4±2.85 % of baseline in agatoxin against 88.4±3.07% baseline in conotoxin, p<0.05, t test).
After agatoxin pretreatment, a 10-min application of DHPG induced iLTD, indicating that P/Q channels are not required for the GABAA responses that are susceptible to iLTD, or for the iLTD induction process (Fig. 7a agatoxin, 73.1±4.56% of baseline, n=8, p<0.001, t test).
We then asked if P/Q-type VGCCs were important for modulation of GABAB receptors by DHPG. There was no difference between the DHPG effects on the pharmacologically isolated GABAB eIPSP recorded in agatoxin and on the GABAB eIPSPs in non-agatoxin-treated cells (to 106.1±8.88% of baseline, n=4, n.s., agatoxin-treated cells; compared to 94.4±2.67% of baseline in control cells, n=3; control and agatoxin-treated cells pooled in Fig. 5b).
Despite the large suppression of the GABAA response by conotoxin (Fig. 1), there was a small but significant iLTD (Fig. 7b) of the remaining GABAA eIPSP after a 10-min exposure to DHPG in conotoxin in five of seven cells tested (to 74.2±3.05% of the eIPSP remaining in conotoxin, cf. Fig. 1, n=5, p<0.001). This was unexpected in view of previous evidence that conotoxin occluded the effect of CB1R agonists and abolished DSI (Lenz et al., 1998; Wilson and Nicoll 2001). Two obvious possible explanations for the residual iLTD were that: (1) it is not mediated by CB1R, or (2) that it is mediated by CB1R receptors on terminals that do not release GABA by N-type VGCCs. To determine which, we repeated the experiment in the presence of the CB1R antagonist AM251, 3 µM (plus conotoxin). The residual iLTD was blocked by AM251 (94.9±4.46% of baseline, n=3, n.s., Fig. 7c).
Whether µOR and CB1 systems interact at the level of long-term inhibitory synaptic plasticity is not known, although prior results point to an independence of the interneurons that express the two kinds of receptors (Neu et al. 2007). Nevertheless, most experiments have not directly tested for interactions. We applied DHPG for 10 min in the presence of DAMGO to determine if eCB actions are altered when µORs are also activated.
We observed two different effects: In 46% (6/13) of the cells, the presence of DAMGO increased the DHPG-induced eIPSP suppression to 69.5±3.36% of baseline (p<0.01) compared with a reduction to 79.7±1.63% in control solution, and completely blocked the iLTD normally caused by DHPG (to 104.57±2.48% of baseline, n.s., n=6). In other cells, DHPG increased eIPSP suppression (to 60.4±1.98% of baseline) and yet normal iLTD was observed (to 77.5±1.59% of baseline, n=7, p<0.001). The differences between the groups were highly significant (Fig. 8a and b). Such divergent results could mean that different groups of interneurons were primarily activated in the two cases. In fact, we observed that the paired pulse ratios (PPRs) of the remaining GABAA eIPSPs were differentially affected by DAMGO depending on whether iLTD was observed or not. DAMGO shifted the PPR of non-iLTD-responsive eIPSPs (Fig. 8a) significantly from 0.79±0.04 to 1.00±0.04 (p<0.001), i.e., there was no longer any paired-pulse depression in DAMGO. In contrast, DAMGO did not alter the PPR of iLTD-responsive eIPSPs (0.85±0.04 to 0.89±0.01, Fig. 8b). DHPG had no lasting effect on PPR in either group, and naloxone reversed the change in PPR for the non-iLTD group (Fig. 8c).
The vast amount of information about hippocampal interneurons has led to a number of generalizations about functional properties of different interneurons and their regulation. We have tested certain hypotheses regarding interneuronal responses and classes of interneurons and provide new information that not only fills in some gaps in the literature but also leads to new questions. The major findings of our study address two separate but related issues: regulation of GABAB eIPSPs and regulation of iLTD.
The source of GABAB eIPSPs has been unclear, and one possibility was that the GABAB system could be activated by all interneurons under certain conditions. A prediction of this hypothesis is that the GABAB eIPSPs should be equally susceptible to blockade by VGCCs that inhibit GABAA eIPSPs. Conotoxin reduced the GABAA eIPSP by 61% and agatoxin reduced it by 47%. In contrast, conotoxin reduced the GABAB response by 81%, and agatoxin reduced it by 62%, suggesting that, unlike the GABAA responses, the GABAB eIPSPs originate to a significant extent from interneurons using both N- and P/Q-type channels. Hoffman and Lupica (2000) reported that GABAB eIPSPs were not affected by application of the synthetic CB1R agonist, WIN55212-2, suggesting that they are not produced by the CB1R-expressing interneurons, 97% of which also express the neuropeptide, cholecystokinin (CCK; Katona et al. 1999). We now report that the GABAB eIPSPs are not altered by eCBs and are therefore not generated by CCK-interneurons. Neurogliaform interneurons are the source of GABAB eIPSPs in neocortex, where they principally target dendrites (Tamas et al. 2003). In CA1, neurogliaform neurons communicate with one another via GABAB synapses (Price et al. 2005), although this network is located in s. lacunosum-moleculare, and it is not known if these cells also communicate with the pyramidal cells. Neurogliaform cells lack CB1R receptors and represent likely candidates for the GABAB eIPSP. It was interesting to note that, although both GABAA and GABAB eIPSPs were greatly reduced by conotoxin, there was nevertheless a subtle but significant functional distinction between them: Over the time course of drug exposure, the GABAB eIPSP was reduced more rapidly than was the GABAA eIPSP. This heightened sensitivity of GABAB suggests potential functional differences between the two GABA responses that could be investigated in future work.
µORs are detected immunohistochemically on parvalbumin expressing (PV)- but not CCK-expressing interneurons, and most µOR-expressing cells were PV-containing (Drake and Milner 2002). PV cells release GABA exclusively via activation of P/Q-type VGCCs (Freund 2003; Hefft and Jonas 2005). These observations predict that large perisomatic eIPSPs that are sensitive to inhibition by DAMGO will be greatly reduced in agatoxin-treated slices. In accordance with the prediction, we find that the effects of DAMGO were much smaller in agatoxin. We cannot determine the identity of the N-type releasing interneurons that are inhibited by DAMGO. The SOM-expressing interneurons, a significant fraction of which also express µORs (Drake and Milner 2002), might be good candidates.
We show directly for the first time that GABAB eIPSPs are not susceptible to mGluR-initiated, CB1R-dependent iLTD. This is consistent with the evidence that the GABAB eIPSPs are not directly inhibited by CB1R activation (Hoffman and Lupica 2000), but differences between iLTD and processes mediated directly by CB1R activation (e.g., Chevaleyre and Castillo 2003; Edwards et al. 2006) meant that insensitivity of the GABAB response to iLTD could not be assumed. Our data further support the conclusion that the GABAB eIPSPs are not produced by CB1R-expressing interneurons and, more importantly, imply that iLTD will alter the normal balance between GABAA and GABAB eIPSPs and thereby affect neuronal circuit interactions.
Because hippocampal CB1R receptors are expressed by interneurons that release GABA via activation of presynaptic N-type VGCCS (Wilson et al. 2001), we tested the prediction that iLTD would be absent in conotoxin. We observed that a residual eCB-iLTD (10% depression compared to the control eIPSP baseline response) persisted in conotoxin. This is in agreement with a previous report from our laboratory (Varma et al. 2002) that a small proportion of conotoxin-insensitive mIPSCs were blocked by a CB1R agonist. It is likely that the small percentage of CB1Rs on non-CCK interneurons (Katona et al. 1999) accounted for the residual eCB-iLTD. Conceivably, an alternative explanation for the present result would be that CCK-interneurons can release GABA partially via a non-N-type release mechanism, although the published data (e.g., Wilson et al. 2001; Hefft and Jonas 2005) do not support this interpretation.
If indeed, iLTD is expressed mainly at terminals releasing via N-type VGCCs; blocking P/Q channels with agatoxin should have little effect on iLTD expression, and indeed, we observed that agatoxin did not affect the magnitude of iLTD, although variability in the responses could have masked a small change. The initial percentage reduction of GABAA eIPSPs caused by DHPG was increased in agatoxin because the extracellular stimulation activates both CB1R-expressing and non-CB1R expressing (agatoxin-sensitive) interneurons. A given absolute depression of release from the CB1R-expressing interneurons will represent a greater percentage reduction of the eIPSPs once the non-CB1R population is removed. Similarly, agatoxin enhances the relative magnitude, although not the absolute magnitude, of DSI by blocking interneurons that are not susceptible to DSI (Lenz et al. 1998). Alternatively, the greater suppression of GABAA eIPSPs in agatoxin could reflect the removal of a presynaptic interaction between different kinds of s. radiatum interneurons, as recently suggested (Karson et al. 2008).
An important behavioral and neurophysiological issue involves the potential interactions between cannabinoid and opioid systems. For example, CB1R-lacking mice self-administer heroin at levels markedly less than do wild-type mice (Ledent et al. 1999). If the receptors are present on separate interneurons, then one possibility is that the two groups of interneurons may interact only indirectly via their synapses on the pyramidal cells. If the interneurons bearing CB1R and µOR are independent, then activating one receptor should have no affect on eIPSP suppression caused by the other. We found that in half of the experiments, pretreatment with DAMGO did not decrease the effects of DHPG (the percentage reduction of eIPSPs by DHPG was actually slightly increased, as expected if CB1R- and µORs are present on different groups of interneurons). In these instances, the µOR and CB1R systems appear to act independently and in parallel.
However, in the remaining cases, DAMGO significantly reduced the suppressive effects of DHPG and completely blocked iLTD. Either there was some interaction between µOR- and CB1R-expressing interneurons, or DAMGO uncovered a set of interneurons on which CB1R activation has only short-term but no long-term effects. Intriguingly, when DHPG induced iLTD, the PPR of the eIPSPs was less than one, i.e., paired-pulse eIPSP depression was observed. In contrast, when DAMGO prevented iLTD, the eIPSPs were unaffected by paired-pulse stimulation: the PPR was ~1, i.e., there was neither depression nor facilitation. We propose that among the class of µOR-insensitive, DHPG-sensitive interneurons, there are two groups that can be distinguished by their paired-pulse responsiveness and the ability to undergo eCB-dependent iLTD. To our knowledge, this is the first report of such a distinction among hippocampal interneurons. A focus of future work should be on the modification of CB1R-mediated responses by µORs.
We find that GABAB responses are produced by interneurons that release GABA via N- or P/Q type calcium channels. They are insensitive to eCB regulation but suppressed by µOR agonists. GABAA responses are suppressed by both opioids and cannabinoids, and there is evidence that the receptors are both present on a subset of interneurons. In some cases, the influences of opioids and cannabinoids interact.
This work was supported by NIH R01 MH077277 and RO1 DA 014625 to B.E.A. We thank Dr. M. A. Karson for her comments on a draft of this manuscript and for help in preparing the reference list. We also thank K. Gormley at the NIDA for providing the SR141716A.
Carlos A. Lafourcade, Departments of Physiology and Psychiatry, University of Maryland School of Medicine, 655 West Baltimore Street, BRB 5-025, Baltimore, MD 21201, USA.
Bradley E. Alger, Departments of Physiology and Psychiatry, University of Maryland School of Medicine, 655 West Baltimore Street, BRB 5-025, Baltimore, MD 21201, USA. Program in Neuroscience, University of Maryland School of Medicine, 655 West Baltimore Street, BRB 5-025, Baltimore, MD 21201, USA, Email: ude.dnalyramu@reglab.