INTRODUCTION

Recent studies have demonstrated that the excitability of hippocampal interneurons can be modulated via a number of receptor proteins, including opioid receptors (Svoboda et al. 1999), muscarinic receptors (Hájos et al. 1998;McQuiston and Madison 1999a), serotonin receptors (McMahon and Kauer 1997), metabotropic glutamate receptors (McBain et al. 1994), and nicotinic acetylcholine receptors (nAChRs) (Albuquerque et al. 1997; Alkondon et al. 1997, 1999; Frazier et al. 1998b; Hefft et al. 1999; Jones and Yakel 1997; McQuiston and Madison 1999b). Although the exact subunit composition of the native nAChR is still unknown, the pharmacological analysis of the agonist-evoked nicotinic currents together with the in situ hybridization studies on cultured hippocampal neurons enabled us to define a set of criteria to identify α7-containing receptors (referred to in this study as α7 nAChRs) and α4β2-containing receptors (referred to in this study as α4β2 nAChRs) (see Albuquerque et al. 1997). Thus nAChRs that are fully activated by agonists such as choline, nicotine, and ACh and inhibited reversibly by methyllycaconitine (MLA) and irreversibly by α-bungarotoxin (α-BGT) are classified as α7 nAChRs. On the other hand, nAChRs that are activated by ACh and nicotine, but not by choline, inhibited by dihydro-β-erythroidine (DHβE), but not by α-BGT, are classified as α4β2 nAChRs. The activation of α7 and α4β2 nAChRs in GABAergic interneurons in the CA1 field of the hippocampus results, respectively, in fast and slowly decaying nicotinic currents (Alkondon et al. 1997,1999; Frazier et al. 1998b; Jones and Yakel 1997; McQuiston and Madison 1999b;Sudweeks and Yakel 2000). If one considers that stimulation of GABAergic synapses can cause either inhibition or disinhibition depending on the neuronal connectivity (Alkondon et al. 1999, 2000b; Ji and Dani 2000), it is essential to identify primary neuron targets for nAChR-dependent GABA release, the strength of which will dictate the significance of nAChR activation to overall hippocampal output and hence to mnemonic processing.

Patterns of activity of GABAergic interneurons have been shown to provide spatial and temporal cues for modifying synaptic weight and thereby to promote encoding and retrieval of memory in the hippocampus (Paulsen and Moser 1998; Wallenstein and Hasselmo 1997). The interneurons are diverse in type and control the input and output activity by spanning their dendrites and axons, respectively, in a lamina-specific manner in the hippocampus (Freund and Busaki 1996). To name a few examples, the basket and axo-axonic interneurons, which contain parvalbumin and are located in the stratum pyramidal (SP) region, suppress by a perisomatic innervation the generation of Na⁺ spikes and action potentials and hence the output of pyramidal neurons (Miles et al. 1996). Calbindin D_28k-containing interneurons, particularly those referred to as type I cells (Gulyás et al. 1999) and located predominantly in s. radiatum (SR), innervate pyramidal cell dendrites at the Schaffer collateral termination zone (Gulyás and Freund 1996) and are implicated in the control of dendritic Ca²⁺ spikes (Miles et al. 1996). Somatostatin-containing interneurons with horizontal dendrites, found in s. oriens (SO), project their axon terminals to s. lacunosum moleculare (SLM) where they modify the excitatory drive arriving at distal apical dendrites of pyramidal neurons. In layer SLM of the CA1, near the SR border, three main interneuron types, i.e., a basket cell type with axon projecting to pyramidal layer, a perforant path-associated cell with axon ramifying in the SLM, and a Schaffer-associated type with axon terminating in SR and SO, have been described (Vida et al. 1998). These SLM interneurons, when activated, result in inhibitory postsynaptic currents (IPSCs) in pyramidal cells (Lacaille and Schwartzkroin 1988;Vida et al. 1998), and such activation suppresses selectively spikes evoked by Schaffer collateral stimulation (Dvorak-Carbone and Schuman 1999).

Taking into account the strong evidence that interneurons have a central role in controlling the activity of neuronal circuitry in the brain, particularly in the hippocampal CA1 region (Dvorak-Carbone and Schuman 1999; Tsuhokawa and Ross 1996), an understanding of the various factors that govern interneuron activity is essential for predicting the output of the hippocampus.

In the present study, we evaluated the role of activation of α7 and α4β2 nAChR on GABAergic transmission to various neuronal populations present in the CA1 region. Using whole cell patch-clamp recording to measure GABAergic PSCs and post hoc tracing of biocytin-filled structures to identify the neuron types in the CA1 area of rat hippocampal slices, we found that nAChR activation enhances GABAergic transmission and that the magnitude of this effect can be predicted based on the target neuron type and the subtype of nAChR involved.

METHODS

Hippocampal slices

Slices (250-μm thickness) of hippocampus along with adjacent cortex were obtained from the brain of 15- to 24-day-old male Sprague-Dawley rats according to a procedure similar to that described earlier (Alkondon et al. 1997). Animal care and handling were done strictly in accordance with the guidelines set forth by the Institutional Animal Care and Use Committee of University of Maryland at Baltimore. Slices were stored at room temperature in artificial cerebrospinal fluid (ACSF), which was bubbled with 95% O₂-5% CO₂ and had the following composition (in mM): 125 NaCl, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, 2 CaCl₂, 1 MgCl₂, and 25 glucose. Neurons in the CA1 field of the slices were visualized by means of infrared-assisted video microscopy for whole cell patch-clamp recordings (Alkondon et al. 1997). The identity of the neurons was later confirmed by postrecording reconstruction of biocytin-filled neurons in fixed slices according to the procedure reported earlier (Alkondon et al. 2000b; Svoboda et al. 1999). The images of neurons were drawn using the Neurolucida program (MicroBrightField, Colchester, VT).

Electrophysiological recordings

The PSCs were recorded from the somata of CA1 neurons according to the whole cell patch-clamp technique (Hamill et al. 1981), using an LM-EPC7 patch-clamp system (List Electronic, Darmstadt, FRG). The ability of an agonist to generate action potentials was assessed by recording action currents in cell-attached configuration at a pipette potential of −60 mV (Alkondon et al. 2000a). This technique had the added advantage of detecting in the same neuron the firing frequency followed later of measuring in the whole cell mode the nicotinic currents or agonist-elicited GABAergic PSCs. All signals were initially filtered at 3 kHz and either recorded onto a video tape for later analysis or directly sampled by a microcomputer using the pCLAMP 6 program (Axon Instruments, Foster City, CA). The slices were superfused with ACSF at 2 ml/min. Atropine (1 μM) was added to the ACSF to block muscarinic receptors. Patch pipettes were pulled from borosilicate glass capillary (1.2 mm OD) and when filled with the internal solution had resistances between 3 and 6 MΩ. The internal solution consisted of (in mM) 10 EGTA, 10 HEPES, 130 Cs-methane sulfonate, 10 CsCl, 2 MgCl₂, 5 QX-314, and 0.5% biocytin (pH adjusted to 7.3 with CsOH; 340 mOsm). Membrane potentials were corrected for liquid junction potentials. All recordings were performed at room temperature (20–22°C).

Agonist and antagonist application

Antagonists were applied via bath superfusion. A U-tube, made in the laboratory from 100-μm-ID glass capillary and having a pore of 250-μm diam, was placed ∼100 μm above the slice and used to deliver the agonists to the somata and dendrites of the neuron under study (Fig. 1A for schematic diagram). In 70% of the experiments, both choline (10 mM) and ACh (1 mM) were tested on the same neuron. Typically, choline was applied first, followed 5 min later by a pulse of ACh. This interval was found to be sufficient to obtain a full recovery from desensitization of nAChRs that occurs after the first pulse of the agonist. Using this protocol, it is possible to calculate the ratio of the responses induced by the two agonists in each neuron studied and to estimate the contribution of each nAChR subtype being activated by these agonists. Neurons located no more than 40–80 μm below the slice surface were used so that the applied agonist would have easy access to the neuronal structures. This choice, however, resulted in the studied neurons having fewer axonal arborizations, a feature that otherwise could have been used as one of the identifying criteria for the neuron types (seeresults). The agonist pulse delivered from the U-tube covered the structures around the cell somata within a radius of ∼125 μm, as assessed by the flow pattern of phenol red-containing ACSF (see Fig. 1A). This arrangement allowed coverage of ∼40–70% of the dendritic branches in each neuron and permitted us to evaluate the impact of activation of nAChRs present in large number of synapses made onto the neuron under study. Also we could correlate neuron type with the magnitude of GABAergic inhibition. Agonist pulses shown in the figures indicate the amount of time the valve that controls the outflow of agonist solution through the U-tube pore was closed. Thus the start of the agonist pulses shown represents the activation of the valve rather than the actual beginning of agonist outflow. The inflow and outflow in the U-tube was adjusted to prevent agonist leakage and/or drawing up of the slice into the U-tube. This flow setting created a delay of ≥500 ms between the time of activation of the valve and the time when the neuron was actually exposed to the agonist. The presence of adjacent cortex in the slices provided additional stability during recordings from neurons in the SO and pyramidal layers where drawing up of the slice into the U-tube would have occurred frequently.

Focal application of ACh was done by a pressure-ejection system. In this, a patch pipette with a tip diameter of <0.5 μm was filled with ACh (1 mM) and positioned ∼10 μm from the somata of the interneuron being studied. A picospritzer (PLI-100, Medical Systems, Greenvale, NY) was used to apply a pressure of 10 psi for 25 or 50 ms to eject the agonist close to the cell somata.

Data analysis

The net charge flowing through GABA_Areceptors during a 12-s agonist application was calculated by integrating the area under the PSCs using the pCLAMP 6 program. The data were sampled at 500 Hz (2,000-μs sampling interval between the points) so that sufficient baseline data could be gathered before the effect of the agonist began. It was confirmed from the analysis of a short pulse (2 s) of agonist-induced PSCs that a sampling rate of 500 Hz was adequate to resolve rapid current fluctuations and thereby to assess the net charge carried by individual GABAergic PSC transients. Sampling the same data segment at 500 Hz and again at 5 kHz resulted in <1% change in the net charge. The frequency, 10–90% rise time, and amplitude of GABAergic PSCs were measured using the continuous data recording (CDR) program (Dempster 1989). All results are presented as means ± SE. One-way ANOVA with Dunnett's post test (comparison of data among various groups of neurons), and two-tailed unpaired t-test with Welch correction (comparison between choline and ACh in various types of neurons) were performed using GraphPad InStat 3.05 for Windows 95 (GraphPad Software, San Diego, CA). The distributions of the amplitudes and intervals between the events were compared using Kolmogorov-Smirnov test, and the difference between the groups was considered significant if the P value was <0.01.

Drugs and toxins used

ACh chloride, (−)bicuculline methiodide, choline chloride, lidocaine N-ethyl bromide (QX-314), biocytin, and atropine sulfate were obtained from Sigma Chemical (St. Louis, MO). Methyllycaconitine citrate (MLA) was a gift from Professor M. H. Benn (Dept. Chemistry, Univ. Calgary, Alberta, Canada). DHβE was a gift from Merck, Sharp, and Dohme (Rahway, NJ). Stock solutions of all drugs were made in distilled water.

RESULTS

In this study, we evaluated in the CA1 region of rat hippocampal slices the net effects of activation of multiple GABAergic synapses via two pharmacologically identified nAChR subtypes that are presumed to contain α7 and α4β2 nAChRs, respectively (Albuquerque et al. 1997; Alkondon et al. 1999; Mike et al. 2000). To assess in each neuron the maximal contribution of the two receptor subtypes in modulating GABAergic transmission, we tested a saturating concentration of choline (10 mM) to activate selectively α7 nAChRs and ACh (1 mM) to activate α7 and α4β2 nAChRs. Previous studies have shown that 10 mM choline is equieffective with 1 mM ACh in activating α7 nAChRs, choline does not activate α4β2 nAChRs, and ACh activates both receptor subtypes (Albuquerque et al. 1997; Alkondon et al. 1999). It has been assumed, therefore, in this study that the choline-evoked responses represent the sole activation of α7 nAChR and that the ACh-induced responses represent the combined activation of α7 and α4β2 nAChRs. Thus any difference in the response magnitude between the two agonists (i.e., ACh − choline), if found, was considered to represent the activation of α4β2 nAChRs. To obtain a rough estimate of the relative contribution of the two nAChR subtypes in modulating GABAergic transmission, the ratio of the net charge of PSCs induced by the two agonists (i.e., ACh/choline) was calculated considering that a ratio of one represents a contribution from only α7 nAChR, a ratio of two suggests equal contribution from both receptor subtypes, and a ratio higher than two indicates a larger contribution from α4β2 nAChR than α7 nAChR. Outward GABAergic PSCs were recorded at 0 mV to avoid any cationic currents arising from nAChR activation in the recorded neurons. To evaluate the overall magnitude of nAChR activation on GABA transmission, the net charge flowing through GABA_A receptor channels was measured and quantified. All recordings were performed on visually identified neurons in the slices and were confirmed later from the images of biocytin-filled neurons (Fig. 1, B–F). In 95 neurons studied, 59 neurons, included in the subsequent detailed analysis, had biocytin labeling that was of sufficient quality to allow positive identification of neuron types based on cell location, cell soma, and dendritic branching. Axonal arborization, however, was incomplete in many neurons and therefore was used only as a secondary criterion for identification of neuron types. In this study, interneurons were grouped primarily on the basis of the laminar location of their cell somata. SP interneurons referred to in this study, however, were those that had cell somata located in either the SP or SO, multiple vertical dendrites going toward SLM, and axons projecting mainly to the SP region.

Nicotinic modulation of GABAergic input to pyramidal neurons

Postrecording reconstruction of presumed pyramidal neurons (see Figs. 1B and 2A) indicated that they had several basal dendrites, a single axon branching in the SO, and a single apical dendrite emerging as a continuation of cell somata passing through the SR and subsequently branching extensively in the SLM. Several secondary branches also emerged from the apical dendrite in the SR. These characteristics were used to confirm the identity of pyramidal neurons (Fig. 2A). Functionally, application of choline or ACh to pyramidal neurons induced PSCs that occurred in quick succession and summated to produce a large tonic outward current (Fig. 2, B, C, andE). Both agonists elicited PSCs in all 12 pyramidal neurons analyzed. Exposure of the slices to bicuculline (10 μM) markedly reduced the frequency of spontaneous and agonist-induced PSCs (Fig.2B), thus confirming the origin of PSCs as arising from activation of GABA_A receptor channels. Exposure of the slices to the α7-selective competitive antagonist MLA (10 nM;n = 3) completely abolished the effect of choline (Fig.2C), confirming the sole involvement of α7 nAChRs in choline-induced PSCs. On the other hand, exposure of the slices to 10 μM DHβE (Fig. 2C), a specific antagonist of α4β2 nAChR, decreased the net charge of GABAergic PSCs induced by ACh by 83% (mean ± SE: 83 ± 4.6%; n = 3), indicating the primary involvement of α4β2 nAChRs in ACh-induced PSCs. Choline-induced PSCs persisted in the presence of 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX, 10 μM) and 2-amino-5-phosphonovaleric acid (APV, 20 μM; n = 6; Fig. 2D), indicating that the glutamate pathway is not required for the elicitation of GABAergic PSCs. On the other hand, exposure of the slices to 0.5 μM TTX (n = 4) abolished most of the PSCs elicited by the agonist (Fig.2E), indicating the recruitment of Na⁺channels in the action of the agonist. The inhibitory effect of TTX also revealed that nAChR agonist did not modify the sensitivity of postsynaptic GABA_A receptors.

The magnitude of the effect of nAChR activation on GABAergic input was determined from the net charge passing through the GABA_A receptors during 12-s application of the agonists. As there are no cationic currents at 0 mV after exposure to bicuculline (see Fig. 2B), the measured net charge reflects solely the anionic current flow through GABA_A receptors. The net charge induced by ACh varied about fivefold (600–2,840 pC; n = 12), and that induced by choline varied about fourfold (186–740 pC;n = 12) among the pyramidal neurons studied (Fig.2F). In each pyramidal neuron examined, the ACh-induced response was larger than that induced by choline. The mean ACh/choline ratio of ∼ 4 (Table 1) is significantly (P < 0.01) larger than the hypothetical ratio of 2 that would have been expected if α7 and α4β2 nAChRs contributed equally to ACh-induced PSCs. Thus a ratio >2, as observed here, reflects a larger contribution of α4β2 nAChRs than α7 nAChRs in modulating GABAergic transmission onto pyramidal neurons.

Nicotinic modulation of GABAergic input to SO interneurons

All interneurons studied in this layer had predominantly horizontally oriented dendrites (n = 8). The axons tended to project either to the SO and SP (neuron in Fig.3A) or to the SLM and to some extent to the SO (neurons in Fig. 3, B and C). SO interneurons with axons projecting mainly to the SLM would correspond to the oriens-lacunosum moleculare (O-LM) interneurons described earlier (McBain et al. 1994; Svoboda et al. 1999) and are known to modify the excitatory drive arriving at distal apical dendrites of pyramidal neurons. It is possible that the remaining SO interneurons, which lack projections to SLM, may correspond to trilaminar and backprojection cells described by others as having horizontal dendrites and axons projecting to other layers (Cobb et al. 1997; Sik et al. 1995). Thus it is likely that more than one type of SO neuron might have been included in our study. Interneurons projecting predominantly to the SLM (n = 3) failed to exhibit choline-induced PSCs (Fig. 3,E and F), and those projecting mostly to the SO and SP (Fig. 3D) showed small responses with a mean net charge of ∼33 pC (n = 5). All SO interneurons (n = 8), however, responded to ACh with PSCs (Fig. 3,D–F). A wide variability in the magnitude of PSC responses to ACh was observed, and PSC summation did not result in robust tonic outward currents (Fig. 3, D–F) as seen in pyramidal neurons. The net charge carried via GABA_Areceptors in response to 1 mM ACh varied 17-fold in the range of 70 and 1,230 pC (Fig. 3G). The mean ACh/choline ratio of ∼4 (Table 1) is significantly larger (P < 0.05) than the hypothetical ratio of 2, suggesting the prevalence of α4β2 nAChRs over α7 nAChRs in modulating GABAergic input to SO interneurons.

Nicotinic modulation of GABAergic input to SP interneurons

SP interneurons found in either the SP or SO regions (n = 7) had two to six vertically oriented dendrites originating from the cell somata, running parallel to CA1 pyramidal neuron apical dendrites, traversing through SR, and reaching SLM. As shown in Fig. 4, A andB, the SP interneurons displayed this unique dendritic architecture and, in addition, had their axon terminating in the SP. These features are characteristics of basket and axo-axonic cells in the hippocampus (Buhl et al. 1996). Choline induced PSCs (Fig. 4, C and D) in six of seven SP interneurons tested, but the net charge of GABAergic PSCs varied about sevenfold between 22 and 161 pC (Fig. 4E). The mean net charge of ACh-induced GABAergic PSCs (Fig. 4, C and D) was ∼136 pC; the individual values ranged from 19 to 492 pC (Fig.4E). Thus notably in SP interneurons, the mean ACh/choline ratio was close to unity (Table 1), suggesting that compared with α7 nAChRs, α4β2 nAChRs play a minor role in modulating GABAergic input to most SP interneurons (SPIs).

Nicotinic modulation of GABAergic input to SR interneurons

SR interneurons (n = 12) invariably had vertically oriented dendrites traversing one or more layers, but some main or secondary branches were also directed obliquely or horizontally (Fig.5, A and B). Although a major part of the SR interneuron dendrites remained in the SR region, 50% of the neurons extended their dendrites into the SLM and 25% of the neurons had their dendrites additionally in the SO. Rarely did we encounter SR neurons with dendrites extending into three layers. In contrast, of the 12 SR interneurons in which we could visualize some axonal branches in biocytin images, all had axon projections into the SR, 2 neurons had their axons extended into the SLM. SR interneurons studied here corresponded to those reported to contain calbindinD_28k that are known to inhibit the Schaffer collateral termination zone (Gulyás and Freund 1996). Both choline and ACh induced PSCs (Fig. 5,C and D) in all 12 neurons studied. The net charge of choline-induced PSCs varied by 19-fold in the range from 41 to 780 pC (Fig. 5E). When ACh was used as the agonist, in addition to a tonic PSC response, discrete individual PSC events could also be discerned in the responses (Fig. 5, C andD). The net charge induced by ACh varied by 10-fold, in the range from 140 to 1,400 pC (Fig. 5E). In SR neurons, the mean ACh/choline ratio was nearly 4 (Table 1), which is found to be significantly different (P < 0.01) from the hypothetical ratio of 2 for equal contribution, thus indicating a predominance of α4β2 nAChR over α7 nAChR in modulating GABAergic input to SR interneurons. We did not find any consistent correlation in SR interneurons between the magnitude of choline- or ACh-induced PSCs and their direction of dendritic or axonal projections.

Nicotinic modulation of GABAergic input to SLM interneurons

In 8 of 10 SLM interneurons verified (see Fig.6A), all dendrites and axonal branches remained solely in the SLM, and in the remaining 2 neurons (see Fig. 6B), dendrites remained in the SLM and SR, whereas axons terminated in the SLM, SR, and SO. SLM interneurons studied here appear to belong to perforant path-associated and Schaffer-associated interneuron types (Vida et al. 1998). Both choline- and ACh-induced PSCs that summated to yield large tonic outward currents (Fig. 6, C and D). In general, there was less variability in the magnitude of the responses (∼3-fold for either agonist) among the neurons studied. For example, the net charge of choline-induced PSCs ranged from 430 to 1,300 pC and that of ACh-induced PSCs ranged from 940 to 2,600 pC (Fig. 6E). Although the mean ACh/choline ratio in SLM neurons was not significantly different (P > 0.05) from the hypothetical ratio of 2.0 (Table 1), the value of 1.83 suggests that α7 nAChR contributed more than α4β2 nAChR in ACh-induced PSCs onto SLM interneurons.

Comparison of the nicotinic modulation of GABAergic input in different types of neurons in the CA1 region

Initially we compared the actual mean net charge of GABAergic PSCs induced by the two agonists in each group of neurons in the CA1 region (Fig. 7A). The magnitude of choline-induced PSCs, an indicator of the strength of α7 nAChR-mediated inhibition, was highest in SLM interneurons and lowest in SO interneurons, and it decreased in the order SLM>pyramidal neuron>SR>SPI>SO. Also, there were significant differences between various groups in the magnitude of choline-induced PSCs. For example, the mean net charge of choline-induced PSCs in the SLM was significantly higher than that in the SO and SPI (P < 0.001), or the SR (P < 0.01). Further, the mean net charge of choline-induced PSCs in pyramidal neurons was significantly higher than that in the SO (P < 0.01) or the SPI (P < 0.05).

ACh-induced PSCs, which represent the strength of inhibition exerted by α7 and α4β2 nAChRs combined, followed pattern similar to that of choline-induced PSCs except that the response was larger in SO interneurons than in SP interneurons. The net charge of ACh-induced PSCs in either SO or SPI neurons was significantly lower than that of the stratum pyramidal pyramidal neuron (SPP) or SLM (Fig. 7A). To get a better idea of the contribution of α4β2 nAChR alone, we subtracted the net charge of choline-induced PSCs from that of ACh-induced PSCs and plotted the results separately for each group of neurons (Fig. 7B). This analysis was undertaken assuming that at the test concentrations used in this study, choline and ACh are equally effective in activating α7 nAChRs and that choline does not activate α4β2 nAChRs (Alkondon et al. 1999; Mike et al. 2000). The subtraction analysis indicated that the contribution of α4β2 nAChR-induced PSCs is highest in pyramidal neurons with the rank order being pyramidal neuron>SLM>SR>SO>SPI (Fig. 7B). Most notably, the mean value obtained in SPI neurons was found to be significantly lower than that in the SPP (P < 0.001), SR or SLM (P < 0.05).

Local modulation by ACh of GABAergic transmission in SR interneurons

The initiation of GABAergic PSCs by U-tube application of the nicotinic agonists, considering the area covered by the agonist, could result from a direct activation of nAChRs at presynaptic/preterminal regions of GABAergic terminals, but it could also result from an action potential-dependent GABA release stemming from somatodendritic nAChRs on the interneurons within the area exposed to the agonist flow. Whether or not a polysynaptic component contributes to the GABAergic PSCs measured in the above-mentioned experiments was assessed in two additional sets of experiments performed on SR interneurons. In the first set of experiments, ACh (1 mM) was applied focally by pressure ejection for brief periods (25–50 ms) close to the somata of the recorded interneuron. These brief pulses of ACh triggered a burst of PSCs that had higher amplitude than the ones that occurred spontaneously (Fig. 8); similar effects were observed in all four interneurons tested. In addition, most of the larger-amplitude events had shorter rise times when compared with the majority of events in the recording (Fig. 8E), suggesting that the nAChR-activated GABAergic synapses must be located close to the cell somata of the recorded interneuron. Addition of TTX (0.5 μM) to the recording medium abolished the effect of focally applied ACh (Fig. 8D), indicating that nAChR-mediated enhancement of GABAergic transmission is due to a local depolarization near the axon terminals, may involve preterminal rather than presynaptic nAChRs in their action, and further rules out any involvement of a postsynaptic action of ACh at the GABA_A receptors.

In the second set of experiments, we tested the effect of 10 μM ACh to find out whether this low concentration of the agonist is able to affect GABAergic transmission and if so, by an action-potential-dependent or independent mechanism. Figure9A illustrates an control experiment from an SR interneuron in which U-tube application of choline induced action currents in a cell-attached recording mode. The nicotinic current induced by another pulse of choline in the same interneuron after breaking the patch membrane to achieve the whole cell configuration (Fig. 9A, bottom) indicates that the action currents recorded in the cell-attached mode were due to the activation of somatodendritic nAChRs present in that interneuron. However, when 10 μM ACh was applied via a U-tube to the interneurons (n = 6), no action currents were induced (see trace in Fig. 9B, top), suggesting that very little somatodendritic nAChRs were activated by this low concentration of the agonist. As illustrated in Fig. 9B (bottom), in the same interneuron in which ACh (10 μM) failed to induce action currents in cell-attached mode, the agonist triggered a burst of GABAergic PSCs in the whole cell configuration. This result indicates that the agonist must be acting at the nAChRs located near the axon terminals and may not involve an action potential-dependent polysynaptic pathway. To assess quantitatively the effect of a low concentration of ACh (10 μM) on the frequency and amplitude of GABAergic PSCs, the effect of a 5-min bath application of ACh (10 μM) was studied in six other interneurons. As illustrated in Fig.9C, the bath application of ACh (10 μM) enhanced both the peak amplitude and the frequency of GABAergic PSCs to a significant extent (by Kolmogorov-Smirnov test).

DISCUSSION

The present study demonstrates for the first time that α7 and α4β2 nAChRs regulate GABAergic transmission in a neuron-type-specific way in the CA1 region of the hippocampus. The strength of nicotinic facilitation of GABAergic input to different types of neurons varied substantially (see Fig.10), indicating that the cholinergic nicotinic system may indeed control the hippocampal synaptic function in a lamina-specific fashion.

Correlation between neuron type and the strength of nAChR-modulated GABAergic inhibition

The major finding of this study is that CA1 pyramidal neurons receive GABAergic input that is facilitated by activation of α7 and α4β2 nAChRs. Previous studies demonstrated the presence of α7 nAChRs in somatodendritic membranes of interneurons located in SR, SO, and SLM, as evidenced by the ability of choline and ACh to evoke fast decaying nicotinic currents (referred to as type IA currents) and their sensitivity to blockade by MLA or α-bungarotoxin (α-BGT) (Alkondon et al. 1997, 1999; Frazier et al. 1998b; Jones and Yakel 1997; McQuiston and Madison 1999b). Subsequent studies described the existence of fast synaptic transmission mediated via α7 nAChRs in CA1 interneurons (Alkondon et al. 1998; Frazier et al. 1998a; Hefft et al. 1999). Two groups of laboratories also examined the distribution of somatodendritic α4β2 nAChRs in various CA1 interneurons and found that they are present in SO, SLM, and to a lesser extent in SR interneurons (Alkondon et al. 1999, 2000a; McQuiston and Madison 1999b). The presence of α4β2 nAChRs was confirmed by the ability of ACh to evoke slowly decaying nicotinic currents (referred to as type II currents) that were insensitive to blockade by α-BGT or MLA but were blocked by DHβE. The presence of several nAChR subunit mRNAs was shown in the SO and SR interneurons, raising the possibility that either α7 or α4β2 nAChRs could also be associated with other subunits or that new subtypes containing α2 subunits may exist (Sudweeks and Yakel 2000). Because it was found that not all the neurons (e.g., pyramidal neurons) expressing mRNA exhibited functional nicotinic currents (Sudweeks and Yakel 2000), it is premature at this stage to predict the number of possible combinations and the number of subtypes of functional nAChRs in the CA1 neurons. Therefore we have restricted our discussion to include only the pharmacologically defined α7 and α4β2 AChR subtypes with respect to their ability to modulate GABAergic transmission. Because the interneurons located in the SO (McBain et al. 1994), SR (Gulyás and Freund 1996), and SLM (Lacaille and Schwartzkroin 1988) each project to pyramidal neurons, one would predict that activation of either α7 or α4β2 nAChR would increase the frequency of GABAergic PSCs in CA1 pyramidal neurons. Although earlier studies indicated the role of α7 and α4β2 nAChRs in enhancing GABAergic transmission to pyramidal neurons (Alkondon et al. 1997; Ji and Dani 2000), none of the previous studies addressed the relative contribution of α7 and α4β2 nAChRs in mediating inhibition to pyramidal neurons. Such information would help to elucidate the participation of brain cholinergic nicotinic system in a number of neurological disorders. In this regard, the present results are the first to reveal the relative contribution of the two nAChR subtypes to brain function, i.e., α4β2 nAChR-mediated inhibition of CA1 pyramidal neurons is at least three times stronger than the inhibition derived from activation of α7 nAChRs.

Interneurons are heterogeneous in type, location, dendritic placement, and axonal termination zone, and this anatomical diversity provides a lamina-specific inhibitory control of the activity of pyramidal neurons in the CA1 region (see review Freund and Busaki 1996). Because interneurons are intrinsically active and are constantly being stimulated by feed-forward and feed-back glutamate pathways (see Fig.10), it is conceivable that inhibition of interneurons would have an impact similar to that achieved by activation. Very few studies have addressed this issue, although the significance of this mechanism is apparent from the anatomical evidence that interneurons synapse onto other interneurons (Gulyás et al. 1996, 1999) and the physiological evidence that spontaneous and evoked IPSCs can be recorded from the interneurons (Cossart et al. 2001;Hájos and Mody 1997; Morin et al. 1996). The importance of interneurons innervating other interneurons is not completely understood. It has been proposed, however, that such interplay can result in disinhibition of pyramidal neurons (Alkondon et al. 1999; Ji and Dani 2000; Tóth et al. 1997) and sustain gamma oscillations (Wang and Buzsáki 1996). The present results indicating that nAChRs play a dynamic role in modulating GABAergic input to different populations of interneurons in a receptor-subtype-dependent fashion add new dimensions to our understanding of hippocampal inhibitory mechanisms and further underscore the importance of interneuron-to-interneuron synapses to brain function.

Target interneurons appear to play an important role in the magnitude of GABAergic inhibition exerted by nAChR activation in the CA1 region (Fig. 10). We found that SO and SP interneurons receive the lowest magnitude of nAChR-dependent inhibition (see Figs. 3, 4, and 7). This can be attributed to the lower percentage (6.4%) of GABAergic inputs onto the somata and dendrites of these neurons, particularly SP interneurons (Gulyás et al. 1999). Conversely, SR interneurons receive one of the highest degrees of nAChR-dependent inhibition, particularly that mediated by α4β2 nAChRs (see Figs. 5and 7). This can be attributed partly or wholly to the fact that SR interneurons receive the largest percentage (29%) of GABAergic inputs onto their somata and dendrites (Gulyás et al. 1999). SLM interneurons receive the highest magnitude of α7 nAChR-dependent inhibition in addition to receiving a strong α4β2 nAChR-mediated inhibition (see Figs. 6 and 7). The GABAergic PSCs induced by choline and ACh in SLM interneurons can arise from the activation of nAChRs near the axon terminals of both SO interneurons as well as other SLM neurons because both neuron types send axonal projection to the SLM region (see Fig. 10). The demonstration of the presence of MLA-sensitive (α7) and DHβE-sensitive (α4β2) nAChRs on the somatodendritic sites of SO interneurons (McQuiston and Madison 1999b; Sudweeks and Yakel 2000) and SLM interneurons (Alkondon et al. 2000a) supports the possibility that the nAChRs are present near the axon terminals of both interneuron types.

Correlation between cholinergic innervation and the strength of nAChR-modulated GABAergic inhibition

Comparison of the present results with other published work suggests that there is no obvious correlation between the density of cholinergic afferents and the magnitude of nAChR-modulated GABAergic transmission in the CA1 region. For instance, a recent study (Schäfer et al. 1998) reported that the density of cholinergic terminals, as assessed from the light microscopic analysis of vesicular ACh transporter antibody-stained fibers, is highest in the SP region, followed by SO, SLM, and SR. This is in quite a contrast to the layer dependence of the magnitude of ACh-activated GABAergic PSCs that followed the order: pyramidal neuron > SLM > SR > SO > SPI (see Fig. 7B). This discrepancy can be accounted for by considering the fact that muscarinic ACh receptors are additional targets besides nAChRs for the actions of the endogenous cholinergic transmitter. Nevertheless, the presence of cholinergic fibers in different fields of the CA1 (Schäfer et al. 1998) assures that endogenous ACh is available for the activation of nAChRs. However, it is not certain whether ACh would interact with the nAChRs in a synaptic or nonsynaptic manner. The prevalence of nonsynaptic specializations apposing cholinergic axon varicosities (Mrzljak et al. 1995; Umbriaco et al. 1995) suggests that, with the exception of selected cases where a cholinergic impulse would gate through synaptic α7 nAChRs located in somatodendritic regions (Alkondon et al. 1998; Frazier et al. 1998a), the majority of nicotinic cholinergic activity would be mediated by presynaptic/preterminal nAChRs located on GABAergic neurons (Alkondon et al. 1997, 1999; Léna et al. 1993; McMahon et al. 1994). The efficacy of both low and high ACh concentration in modulating GABAergic transmission (see Figs. 8 and 9) indicates that this mechanism is operational under a wide range of physiological conditions.

Implications of nAChR-modulated GABAergic input to function of CA1 pyramidal neurons

nAChRs can alter the function of CA1 pyramidal neurons, the main output neurons of the hippocampus, in at least three distinct ways. First, nAChR-mediated facilitation of GABAergic transmission to pyramidal neurons exerts an inhibitory effect during cholinergic neuron firing. This would be accomplished via activation in the interneurons of either somatodendritic nAChRs or presynaptic/preterminal nAChRs. What is the significance of nAChR-triggered GABAergic PSCs recorded in the pyramidal neurons? The most plausible explanation is that cholinergic nicotinic inhibition would suppress weak excitatory signals arriving at the pyramidal neuron dendrites such that only strong signals would be propagated. This could be a mechanism to filter extraneous signals (Paulsen and Moser 1998) and increase the attentional function. This type of mechanism is presumed to be involved in the actions of nicotine in cigarette smokers (Stolerman et al. 2000). Our study suggests that α4β2 nAChRs would have a greater role than α7 nAChRs in this process.

Secondly, nAChR-mediated GABA release disinhibits CA1 pyramidal neurons via inhibiting the interneurons. When α7 nAChRs are activated, SLM interneurons are inhibited more than other interneurons, resulting in a selective disinhibition of the dendritic segments of pyramidal neurons innervated by SLM axon terminals. When α4β2 nAChRs are activated, both SR and SLM interneurons are inhibited, resulting in disinhibition of dendritic areas innervated by both neuron types. Disinhibition would be less prominent in dendritic compartments innervated by SO and SP interneurons because these interneurons receive the least nAChR-mediated inhibition. Thus nAChRs appear to disinhibit feed-forward inhibitory zones (i.e., SR and SLM interneuron target zones) more than feed-back inhibitory zones (i.e., SO and SP interneuron target zones) at the pyramidal neuron dendrites (see Fig.10).

Third, nAChR-mediated GABA release can cause neuronal hyperpolarization, which in turn affects neuronal function via several mechanisms. For example, hyperpolarization removes inactivation of inward currents (Cobb et al. 1995), and this action resets the membrane potential of the neurons such that a subsequent excitatory input will be more effective in eliciting action potentials. This mechanism can increase the efficiency of the neuron network as a whole. The observation that all CA1 neurons receive some degree of nAChR-dependent inhibition supports this notion. The nAChR-mediated GABAergic inhibition of SLM interneurons deserves special mention because α7 nAChR activation can result in a neuron hyperpolarization, a signal that is able to trigger rebound burst firing in SLM interneurons even in the absence of excitation (Lacaille and Schwartzkroin 1988). Burst firing in SLM interneurons suppresses spikes in pyramidal neurons evoked by stimulation of Schaffer collaterals (Dvorak-Carbone and Schuman 1999), and this allows selective activation of the pyramidal cells via the perforant pathway. Such selective regulation of intrinsic (e.g., Schaffer collateral) and extrinsic (e.g., perforant path) afferent inputs is considered important in switching between encoding and retrieval modes of associative memory systems (Hasselmo and Schnell 1994;Paulsen and Moser 1998; Wallenstein and Hasselmo 1997).

In summary, we propose that α7 and α4β2 nAChR-activated GABAergic inputs to pyramidal neurons and interneurons in the CA1 field of the hippocampus mediate several functions including inhibition, disinhibition, and removal of inactivation of intrinsic membrane conductances. Such mechanisms are critical for encoding information as well as retrieval of memory in the CNS.

FOOTNOTES

• Address for reprint requests: E. X. Albuquerque, Dept. of Pharmacology and Experimental Therapeutics, University of Maryland School of Medicine, 655 W. Baltimore St., Baltimore, MD 21201 (E-mail:[email protected]edu).