Neuron,

Vol. 6, 889-900, June, 1991, Copyright

@ 1991 by Cell Press

Synaptic Localization of Adrenergic Disinhibition in the Rat Hippocampus Van A. Doze, Gal A. Cohen, and Daniel V. Madison Department of Molecular and Cellular Physiology Beckman Center for Molecular and Genetic Medicine Stanford University School of Medicine Stanford, California 94305-5426

Summary Norepinephrine is an endogenous neurotransmitter that reduces synaptic inhibition onto pyramidal neurons in the hippocampus by an action at an a-adrenergic receptor. The physiological mechanism of this disinhibition was previously not known, except that it occurred at a site presynaptic to the inhibited pyramidal cell. In this paper we present evidence that adrenergic disinhibition is restricted to the early phase of the evoked inhibitory postsynaptic potential in area CA1 of the hippocampus. The locus of disinhibition does not appear to reside in the interneuronal terminal, axon, or cell body. Instead, adrenergic agonists appear to reduce evoked synaptic inhibition by depressing excitatory synapses that acti{*ate the interneuron. Introduction Norepinephrine (NE) can influence the excitability of the hippocampus in several ways. It can cause both t mall hyperpolarizations and depolarizations of the pyramidal cell membrane potential (Segal and Bloom, .‘974a; Segal, 1981; Madison and Nicoll, 1986a) and c:an change the size of population spikes recorded r:xtracellularly (Mueller et al., 1981; Madison and Ni(:oll, 1988a; Mynlieff and Dunwiddie, 1988). NE reduces a calcium-activated potassium current in these neurons (Madison and Nicoll, 1982; Haas and Konnerth, 1983), resulting in adecrease in action potential trequencyaccommodation (Madison and Nicoll, 1982, 1986a). NE also increases the amplitude of voltagedependent calcium currents in pyramidal cells (Madiq,on and Nicoll, 1986a; Gray and Johnston, 1987; I)oerner and Alger, 1988). However, the most protound hippocampal action of NE, in terms of the regulation of synaptic function, is that it markedly reduces ~;ynaptic inhibition (Mody et al., 1983; Leung and ,vliller, 1988; Madison and Nicoll, 1988a). This disinhi‘lition has been measured as an increase in the num‘)er of population spikes resulting from synaptic acti,/ation (Mody et al., 1983; Madison and Nicoll, 1988a) *md as a reduction in the amplitude of inhibitory post,;ynaptic potentials (IPSPs) recorded in CA1 pyramidal ,:ells (Leung and Miller, 1988; Madison and Nicoll, 1988a). The adrenergic disinhibitory action has been ;hown to be mediated through an a-adrenoceptor Madison and Nicoll, 1988a).

The IPSP in area CA1 of the hippocampus consists of two components, an early and a late phase. The early (fast) IPSP occurs immediately after the excitatory postsynaptic potential (EPSP) and is believed to arise from a feedback (recurrent) circuit as well as a feedforward circuit (Kandel et al., 1961; Andersen et al., 1964a, 1964b; Alger and Nicoll, 1982a). Thus, it can be elicited with either orthodromic or antidromic stimulation. Theearly IPSP results from an increase in the chloride ion conductance and is mediated by a y-aminobutyric acid type A (CABAJ receptor on the postsynaptic pyramidal cell body (Andersen et al., 1980; Dingledine and Langmoen, 1980). The late (slow) IPSP peaks approximately 150 ms after the early IPSP and arises from a feedforward circuit (Alger and Nitoll, 1982a, 1982b; Alger, 1984; Newberry and Nicoll, 1984,1985). It is only observed with orthodromic stimulation. The late IPSP is believed to be mediated by a GABAe receptor coupled to a potassium channel located postsynaptically on the dendrites of the pyramidal cell (Alger and Nicoll, 1982b; Gahwiler and Brown, 1985; Dutar and Nicoll, 1988a). The functional characteristics of the inhibitory circuitry in the CA1 region of the hippocampus, adapted from Lacaille and Schwartzkroin (1988b), are illustrated in Figure 1. Synaptic inhibition in CA1 is mediated by at least three different kinds of interneurons. Two of these interneuronal types, the stratum oriens/ alveus interneuron and the basket cell, receive strong recurrent excitatory synaptic input from CA1 pyramidal cells and feedforward excitatory input from Schaffer collateral or commissural fibers. These interneurons in turn directly inhibit pyramidal cells via GABAergic synapses onto pyramidal cell somata (Knowles and Schwartzkroin, 1981; Schwartzkroin and Kunkel, 1985; Lacaille et al., 1987). Activation of these interneurons is believed to give rise to the early IPSP. For simplicity, the stratum oriens/alveus interneurons and basket cells are represented together in Figure 1, since they are similarly connected in the local synaptic circuitry. The third type of interneuron, the stratum lacunosum-moleculare interneuron, receives only feedforward excitatory input and in turn forms GABAergic synapses onto pyramidal cell dendrites (Lacaille and Schwartzkroin, 1988a). Activation of lacunosum-moleculare interneurons is believed to give rise to the late IPSP (Lacaille and Schwartzkroin, 1988b). In Figure 1, the inset trace shows the two phases of the IPSP, with arrows indicating the schematic synapse believed to be the major contributor to each phase (Newberry and Nicoll, 1985; Lacaille and Schwartzkroin, 1988b).The small numeralson thediagram indicate the potential sites where adrenergic agonists might act to reduce synaptic inhibition within this circuitry. These possibilities are that adrenergic ago-

NWKNl 890

Figure 1. A Schematic Diagram ot the Major Functional Local Inhibitory Circuitry in the Rat Hippocampal CA1 Region This diagram does not depict a strict anatomically accurate version of the circuitry, but represents a summary of the functional feedback and feedforward natureof inhibition in this part of the hippocampus. The late IPSP (inset) arises from the feedforward circuit. It is mediated by CABA receptors localized primarily on the dendrites of the pyramidal cell. The early IPSP (inset) arises from both the feedforward and the feedback circuit. It is mediated by CABAA receptors localized primarily on the Early somata of pyramidal cells. It is not certain Late IPSP that a given pyramidal cell activates its own IPSP feedback interneuron. but that distinction is not critical for evaluating the data in this paper. The small inset numbers represent potential sites for the adrenergic disinhibitorvaction and will be referred to throuehout the text. P, CA1 pyramidal cell; OYA, interneuron; B, basket cell interneuron; L-M, lacunosom-moleculare interneuron; SCIComm, Schaffer collaterals/commisInset: An example of the IPSPs occurring after stimulation of the Schaffer collateralslcommissural pathway. Arrows indicate believed to produce each phase of the IPSP.

yF

orienslalveus sural inputs. the synapse

nists may reduce excitatory synaptic activation of pyramidal cells, subsequently reducing the recurrent (feedback) activation of interneurons (location I), reduce transmission through the feedforward part of the circuit (location 2), change the postsynaptic response to the inhibitory transmitter GABA (location 3), reduce GABA release from interneuronal terminals (location 4), reduce action potential conduction in interneurons or reduce action potential-stimulated GABA release from interneuronal terminals (location 5), decrease the excitability of the interneurons (location 6), decrease excitatory synaptic activation of interneurons (location 7). Two of these possibilities have been ruled out in a previous report (Madison and Nicoll, 1988a). First, it was shown that NE does not reduce the excitatory activation of pyramidal cells (location 1 in Figure I), because it does not reduce the excitatory synaptic field potential in area CAI. Second, it was shown that NE does not reduce evoked IPSPs by an action on the postsynaptic pyramidal cell. NE reduces evoked IPSPs without causing any change in the IPSP reversal potential (Madison and Nicoll, 1988a). This demonstrated that NE does not cause any change in the transmembrane ionic gradients that drive the early IPSP. However, the function of GABAA receptors is highly regulated and can be influenced by changes such as phosphorylation in the postsynaptic cell (Chen et al., 1990). This raises the possibility that NE might decrease inhibition by reducing GABA receptor function. However, NE causes no change in the amplitude or reversal potential of hyperpolarizations in the pyramidal cell induced by iontophoretically applied GABA (Madison and Nicoll, 1988a). This shows that NE induces no change in the GABA sensitivity of the pyra-

midal cell. These results led to the conclusion that there is no detectablechange in the postsynaptic pyramidal cell that can account for the disinhibitory effect of NE. Thus, the postsynaptic pyramidal cell (location 3 in Figure 1) was ruled out as the site of adrenergic disinhibitory action. Although the adrenergic disinhibitory effect appears to occur in the synaptic circuitry at a site presynaptic to the pyramidal cell, several potential sites of action remain. Using a combination of pharmacological probes and electrophysiological techniques, we have examined the question of the synaptic localization of adrenergic disinhibition. First, we determined whether adrenergic agonists reduce both the early and late phases of the IPSP, or whether they have a selective effect on only one component. Second, we studied, on a more microscopic level, the location within the synaptic circuitry of the a-adrenergic disinhibitory effect. The results provide evidence in favor of the hypothesis that adrenergic agonists selectively decrease the early IPSP by inhibiting excitatory synaptic activation of stratum oriens/alveus interneurons and/or basket cells. Results Disinhibitory Action of Adrenergic Agonists The disinhibitory action of an adrenergic agonist is apparent in any one of three electrophysiological measures in hippocampal slices: the extracellularly recorded population spike, the intracellularly recorded excitatory postsynaptic potential (EPSP), and the IPSP. Orthodromic stimulation normally causes a single population spike, a single action potential upon the intracellular EPSP, and a large intracellular IPSP.

A~renergic WI

Disinhibition

in Hippocampus

With strong electrical orthodromic stimulation, both phases of the IPSP are evoked, although the extent of temporal overlap between these two IPSPs often makes it difficult to differentiate between phases by visual examination. Application of adrenergic agonists such as NE, (l10 PM) or epinephrine (EPI, 0.3-10 PM) causes the appearance of multiple population spikes (Figure 2A), similar in appearance to those caused by GABA antagonists such as bicuculline (Schwartzkroin and Prince, 1980). The intracellular EPSP appears to increase in both amplitude and duration and often evokes multiple action potentials, as opposed to a single action potential in control conditions (Figure2B). In addition, the amplitude of the IPSP is diminished during application of EPI (Figure 2C) or NE. The apparent increase in the amplitude and duration of the intracellularly recorded EPSP occurs as a result of the decrease in the temporally overlapping IPSP, since extracellularly r$?corded field EPSPs are not affected by application of NE (Madison and Nicoll, 1988a) or EPI (Figure 2D). Although NEispresumablytheendogenousagonist causing disinhibition, in many of these experiments we used EPI as an agonist, because its effects are qualitatively similar. W e found EPI to be a more potent agonist than NE at producing this disinhibitory effect (Madison and Doze, 1990, Sot. Neurosci., abstract).

Control

Epinephrine

Wash

i -I

2 mV

10 m s

B

I

C A +.--J5mV 200 m s

4;ri,,, 5 ms

Early versus late IPSP In characterizing the effects of NE or EPI on the intracellularly recorded IPSP, we noted two unusual aspects of the disinhibitory effect. First, while NE and EPI both reduced the amplitude of the intracellular II’SP, this blockade was not complete even at supramaximal concentrations of agonist (NE at IO PM, 5j% + 8%, n = 7; EPI at 10 PM, 58% + 12%, n = 22). Second, close examination of the records revealed that a major effect of NE or EPI was an increase in the apparent time to peak of the IPSP rather than a simple decrease in the IPSP amplitude (see Figure 2B). This effect closely resembled that of bicuculline, which blocks only the early IPSP (Schwartzkroin and Prince, 1980). This suggested that the effect of adrenergic agonists might be to reduce the early IPSP preferentially. To test for a selective effect of adrenergic agonists on one phase of the IPSP, we isolated pharmacologically the early or late IPSP and then tested the effects of EPI on each isolated IPSP phase. To isolate the early phase of the IPSP, we took advantage of the recent development of antagonists selective for the GABAe rg?ceptor, such as 2-hydroxysaclofen (Kerr et al., 1988), which block the late IPSP. Conversely, picrotoxin was used to isolate the late IPSP by blocking the GABAJ chloride-mediated early IPSP. Because synaptic activity results in a small, NE-sensitive, calcium-activated potassium current in the pyramidal cell that overlays the IPSP (Madison and Nicoll, 1982; Newberry and Nicoll, 1984), we made the recordings in this series of experiments with 100 m M 8-bromoadenosine 3’,5’-

Figure 2. Effects of Epinephrine pocampal Slices

on Synaptic

Responses

in Hip

In all figures, EPI was applied to the slice in the ACSF perfusate. (A) Effects of EPI on population spikes recorded in stratum pyramidale of area CAI, in control, after EPI (.I P M for 15 min), and following wash in drug-free ACSF (for 60 min). Note the appearance of multiple population spikes in the presence of EPI. (B) Intracellular recordings from a CA1 pyramidal cell with a microelectrode containing 2 M KMeSO+ in control, after EPI (IO P M for 15 min), and following wash (for 294 min). Note also that the cell fires multiple action potentials corresponding to multiple population spikes in (A). (C) Intracellular recordings from the same pyramidal cells as in (B) at a slower time scale to show the IPSP. Note the reduction in the amplitude of the IPSP caused by EPI (IO P M for 15 min). In both (B) and (Cl, the stimulating electrode was placed more than 1 m m from the recording site to evoke a purely polysynaptic IPSP. (D) Field EPSPs recorded in stratum radiatum of area CA1 in control, after EPI (1 P M for 15 min), and following wash (for 60 min). Note that EPI caused no change in these potentials, indicating a lack of effect on excitatory transmission onto pyramidal cells (location 1 in Figure 1). The membrane potential of the cell in (B) and (C) was -61 mV.

cyclic monophosphate (8-Br-CAMP) in the microelectrode. This nonhydrolyzable CAMP analog blocks the calcium-activated potassium current in CA1 pyramidal cells (Madison and Nicoll, 1982,1986b), but has no effect on IPSPs (Newberry and Nicoll, 1984). When we blocked the early IPSP with picrotoxin, we found that the remaining isolated late IPSP was insensitive to EPI. In the experiment illustrated in Figure 3, EPI was applied first in control artificial cerebro-

Neuron 692

A

Control

Epinephrine

A

Wash

Control

Epinephrine

1 I

Wash 3

I

‘1

1if--Picrotoxin 4

k--Picrotoxin 7

5 -1L

+ Epinephrine

P-OH-saclofen

+Epinephrine

-IL 4/---J-A.#.@ 6

+2-OH-saclofen

4 I

5

2-OH-saclofen 6

1

4G 200

ma

Picrotoxin

B

9

6

z

‘Or

p

2.5 -

t

A---

4.---

-L----

-A/----

-]5 200

Epinephrine

2-OH-saclofen Epinephrine

mV

ms

0 15 9 g

12.5

c

Epi

Epi

2-OH-saclofen

: I=

0 0

50

100

150

200

Minutes Figure 4. Blockade of the Late IPSP Does Not Prevent the Disinhibitory Effects of Epinephrine 1

I

140

210

~ 260

_

I 350

Minutes Figure 3. Epinephrine Reduces the Early but Not the Isolated Late IPSP in CA1 Pyramidal Cells All records in this figure are from the same CA1 pyramidal cell recorded with an intracellular microelectrode filled with 2 M KMeSO., and 100 m M &Br-CAMP. (Arm1) The effects of EPI alone on the orthodromically evoked IPSP. Illustrated are the IPSP in control ACSF (A,), after 15 min in bath-applied EPI at 1 u M (A,), and after 20 min wash with drug-free solution (A,). Note that EPI reduces the IPSP by approximately 50%. (A& The effects of EPI after the early IPSP has been blocked by the bath application of picrotoxin (100 PM). Illustrated are the IPSP in picrotoxincontaining ACSF (Ad), after 15 min of EPI added to the picrotoxin ACSF (AS) and after 15 min wash with picrotoxin-containing ACSF (A,,). Note that EPI had no effect on the IPSP after the early phase had been blocked by picrotoxin. (A,..J The effects of 2-hydroxysaclofen (100 PM) on the IPSP in the presence of picrotoxin. Illustrated are the IPSP in picrotoxin alone (A,), 15 min after addition of 2-hydroxysaclofen in the presence of picrotoxin to the bath (A&, and 25 min after returning to ACSF containing picrotoxin alone (A,). The remaining IPSP is sensitive to 2-hydroxysaclofen, as expected for a late CABAB-mediated IPSP. (B) A plot of the IPSP amplitude versus experimental time. Each point represents the peak amplitude of a single, nonaveraged IPSP. The numerals inset in the plot indicate the times that the traces in (A) were taken. The drugs were bath-applied for the periods indicated by the bars. The membrane potential of this cell was -59 mV.

spinal fluid (ACSF) to show the extent of IPSP blockade. Following recovery of the IPSP upon washing EPI, picrotoxin (100 PM) was added to the ACSF to block the early IPSP. Reapplication of EPI in this condition did not reduce the remaining late IPSP. EPI (1 or IO

All records in this figure are from the same CA1 pyramidal cell recorded with an intracellular electrode filled with 2 M KMeSO+ (A,.,) The effects of EPI on the IPSP in control ACSF (A,), after 15 min in EPI at 10 p M (A,), and after 25 min wash with drug-free ACSF (A,). (A,,) The effects of EPI in the presence of 2-hydroxysaclofen (500 PM). Illustrated are sample records after 30 min in 2-hydroxysaclofen alone (Ad), 15 min after the addition of EPI at 10 u M (A,), and 35 min after returning to ACSF containing 2-hydroxysaclofen alone (A& Note that EPI had the same effect on the IPSP after the late phase had been blocked by 2-hydroxysaclofen. The remaining IPSP was blocked by picrotoxin, as expected for an early GABA*-mediated IPSP (data not shown). (B) The graph shows the time course of the experiment illustrated in (A). Each point represents the peak amplitude of a single, nonaveraged IPSP. The numerals inset in the plot indicate the times that the traces in (A) were taken. The drugs were bathapplied for the periods indicated by the bars. The membrane potential of this cell was -56 mV.

PM) had no effect on the isolated late IPSPs (-4% i 12%, n = 7, NS). Before the addition of picrotoxin, EPI reduced the IPSP by 47% 3t 6% (n = 7). The difference between the effect of EPI in control ACSF versus that in picrotoxin-containing ACSF, was significant at p < 0.01. The EPI-insensitive late IPSP that remained in the presence of picrotoxin was, however, reduced 50% k 8% (n = 5) bythe selective GABAe receptor antagonist 2-hydroxysaclofen (100 PM), as expected (see Lambert et al., 1989). In contrast, when we blocked the late IPSP with 2-hydroxysaclofen, we found that the isolated early IPSP was still fully sensitive to EPI. In the experiment illustrated in Figure 4, EPI was applied first in control ACSF to show the extent of IPSP blockade. Following recovery of the IPSP upon washing EPI, 2-hydroxysaclofen (500 PM) was added to the ACSF to block

A(,renerg~c 8Y3

Disinhibition

in Hippocampus

2

Figure5. Norepinephrine Does Not Decrease Miniature Spontaneous (TTX-lnsensitive) IPSPs

Norepinephrine

s

All records in this figure are from the same CA1 pyramidal cell recorded with an intracellular microelectrodecontaining3M KCI and 100 m M QX-314. The preparation was bathed in TTX at 1 P M throughout the experiment to prevent the occurrence of action potential-dependent spontaneous IPSPs.MiniaturespontaneouslPSPsaredepolarizing because of the cell’s reversed chloride gradient caused by the recording electrolyte. Each sweep is a 1 s long, nonavPiWObXi” eraged record. Sweeps are acquired once [__~.~~ -I every2 s, and those illustrated areconsecutive. (A,) In control (TTX-containing) ACSF, 3 miniature spontaneous IPSPs occurred in .. this cell at a rate of between 3 and 5 Hz. (A*) Application of NE at 10 p M for 10 min caused no decrease in the rate of these spontaneous events. (A3 The subsequent wash also produced no apparent changes in the frequency of these events. (&) These miniature spontaneous IPSPs were, howeyer, completely blocked by the application of picrotoxin (100 PM). (B) The graph shows the frequency of spontaneous miniature IPSPs (i;l Hz) versus experimental time. Each frequency bin is the rate (in Hz) averaged over a 5 s epoch. Picrotoxin, which decreases the amplitude of IPSPs by blocking GABAJchloride channels, caused an apparent decrease in the frequency of spontaneous miniature IPSPs as a result of this reduction in amplitude. The numerals inset in the plot indicate the times that the traces in (A) were taken. Drugs uere applied for the periods indicated by the bars. The membrane potential of this cell was -60 mV.

the

late

reduced tion

IPSP.

of EPI in this condition early IPSP. Before the applica-

Reapplication

the remaining

of 2-hydroxysaclofen

(100 or 500 PM),

EPI reduced

IPSP by 48% f 9% (n = 5). In the presence of GABAe receptor blockade, EPI reduced the IPSP by 49% + 11% (n = 5). The isolated early IPSP that remained in the presence of 2-hydroxysaclofen was abolished by picrotoxin (data not shown). Taken together, the experiments illustrated in Figures 3 and 4 demonstrate that it is the early IPSP that is sensitive to adrenergic agonists, whereas the late IPSP is insensrtive. the

Miniature Spontaneous IPSPs To begin a dissection of the synaptic locus of adrenergic disinhibition, we addressed first the possibility that adrenergic agonists reduce GABA release from interneuronal terminals. This was done by examining spontaneous IPSPs recorded in pyramidal cells (Alger and Nicoll, 1980; Haas and Rose, 1982). There are two types of spontaneous IPSPs. The largest and most numerous IPSPs result from the spontaneous action potential discharge of interneurons, since they are blocked bytetrodotoxin (TTX) (Alger and Nicoll, 1980). After blocking action potentials with TTX, a population of small amplitude, miniature spontaneous IPSPs remain which are likely to represent spontaneous release of GABA from interneuronal synaptic terminals (Ropert et al., 1990; Edwards et al., 1990; Otis et al., 1991). In the presence of extracellular TTX (1 PM), we recorded

KCI-filled

intracellularly

microelectrodes

from

pyramidal

to reverse

cells,

with

the transmem-

brane chloride gradient. These experiments were performed in the presence of the glutamate antagonists 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 10 PM) and 2-amino-5-phosphonovalerate (APV; 50 PM) to block spontaneous EPSPs. Under these conditions, miniature IPSPs can be seen as small transient depolarizations of the membrane potential. In our experiments, these events occurred at an average frequency of approximately 2 Hz and had amplitudes in the range of l-4 mV. These miniature spontaneous IPSPs were completely blocked by any one of three antagonists: the GABAJchloride channel blocker picrotoxin (100 PM), the GABAA receptor antagonist bicuculline (10 PM), or the GABAA receptor antagonist SR-95531 (MeinvilleandVicini,1987),thusconfirmingthatthese events were inhibitory (GABAergic) in nature. NE (IO pM)did not affect thefrequencyof miniature spontaneous IPSPs (Figure 5). The basal level of miniature spontaneous IPSPs in our pyramidal cell recordings was 1.9 * 1.2 Hz (n = 12), including 2 cells in which no miniature IPSPs were detected. In 7 cells from which miniature IPSPs were recorded, the rate of miniature spontaneous IPSPs was 2.7 k 0.7 Hz in control and 3.0 f 0.8 Hz in NE (IO PM) (difference not significant). Similar findings were observed with EPI (10 PM, n = 3, data not shown). These results indicate that adrenergic agonists do not depress the spontaneous release of GABA from interneuronal terminals (location 4 in Figure 1). Monosynaptic Evoked IPSPs To test the hypothesis that adrenergic agonists may decrease action potential-evoked release of GABA

Sl

Ret

+

GABA synapse Direct electrical stlmulatlon from bipolar stimulating electrode

C

Control

Epinephrine

Wash

DALA

Wash

Picrotoxin + 2-OH-Saclofen

S. Oriens li+---rr~r-

s. Radi”‘““p

J//+

k

$&...--

2/-

z 200ms

Figure 6. Epinephrine

Does

Not Reduce

Monosynaptic

Evoked

IPSPs in Area CA1

(A) Schematic diagram of the placement of electrodes in the hippocampal slice to elicit a stratum oriensialveus (Sl) and stratum radiatum (S2) monosynaptic IPSP. Stimulating electrodes were placed within 1 mm of the intracellular recording electrode. (6) A schematic diagram showing the method of generating purely monosynaptic evoked IPSPs. The IPSP is recorded with an intracellular electrode (Ret) in a pyramidal cell (P). A nearby stimulating electrode (Stim) activates excitatory fibers (E) and directly stimulates interneurons (I). Polysynaptic activation of interneurons by excitatory glutamate fibers is prevented by bathing the preparation in a mixture of CNQX (IO PM) and APV (50 PM), leaving only the monosynaptic component of the IPSP mediated by the direct interneuronal stimulation. (C) Effects of EPI and DALA, an enkephalin analog, on monosynaptic evoked IPSPs. All records in the figure were taken from the same CA1 pyramidal cell recorded with an intracellular electrode filled with 2 M KMeSO+ The preparation was bathed in 10 uM CNQX and 50 uM APV throughout the experiment to block glutamate synapses. IPSPs were evoked by electrical stimulation through a microelectrode placed close to the recording site, either in stratum radiatum or stratum oriens, to stimulate interneurons directly. Stimulus current was adjusted to give a monosynaptic IPSP of just-maximal size. Note the presence of a late IPSP when the stimulation is given in stratum radiatum, but not in stratum oriensialveus. EPI (IO PM for 15 min) had no effect on these monosynaptic IPSPs. They were reduced, however, by the application of DALA (IO uM for 10 min) and by a combination of 2-hydroxysaclofen (300 uM for 30 min) and picrotoxin (100 uM for IO min). The membrane potential of this cell was -64 mV.

from interneuronal terminals, or may inhibit another aspect of interneuronal excitability, we examined the actions of adrenergic agonists on monosynaptic evoked IPSPs. Electrical stimulation of the stratum radiatum in CA1 evokes an IPSP in the pyramidal cell that consists of a polysynaptic component, mediated through excitatory synaptic activation of interneurons, and a monosynaptic component, resulting from direct activation of the interneurons by the stimulating electrode. The excitatory synaptic activation (Sah et al., 1990) and polysynaptic component of the IPSP are eliminated by bathing the slice in the glutamate antagonists CNQX and APV (Collingridge et al., 1988; Davies and Collingridge, 1989; Davies et al., 1990). Under these conditions, purely monosynaptic evoked IPSPs are produced by electrical stimulation in stratum radiatum or stratum oriens/alveus. Thus, any effect of adrenergic agonists on monosynaptic evoked IPSPs would not involve the glutamate synapses that drive the interneurons, but, rather, would be dignos-

tic for an effect on the interneuron or pyramidal cell (see Figure 68). In the experiment shown in Figure6A, an intracellular electrode was placed into a pyramidal cell in the CA1 area of a slice. Two stimulating electrodes were also placed in the slice, one in stratum oriens and one in stratum radiatum, at a location near the recording electrode (Figure 6A). A second recording electrode, for detecting extracellular field potentials, was placed in stratum pyramidale also near the intracellular electrode. Application of a combination of CNQX (IO PM) and APV (50 PM) completely blocked the extracellular field EPSP (data not shown) and intracellular EPSP, leaving only a monosynaptic evoked IPSP. Application of EPI (IO PM) did not reduce monosynaptic IPSPs elicited by stratum radiatum or stratum oriens/alveus stimulation (Figure 6C). In contrast, two agents known to inhibit interneurons did reduce monosynaptic evoked IPSPs. The GABAe receptor agonist baclofen and the stable enkephalin analog

Ac!renerg~c Dlslnhibition

in Hippocampus

895

Table 1. The Effects of Norepinephrine WenslAlveus-Induced Monosynaptic -

and Other Agonists on the Amplitude of Stratum RadiatumIPSPs Recorded lntracellularly in CA1 Pyramidal Neurons Stratum

Compound EDinephrine horepinephrine Baclofen CALA

Radiatum

Stratum

Percent Change of IPSP (+_ SD)

Oriens/Alveus Percent Change of IPSP (+ SD) --___

Concentration

n

10 10 1 IO

15

-1

f

11

5

-1

f

3

+4

f

6

3

+6

+ 5

PM pM PM PM

n refers to the total number of ceils tested of the orthodromically evoked IPSP. a Data significant at p < 0.01. -

with

n

-___ and Stratum

a

IO

-56

f

14d

7

-45

f

13

-50

+ 13"

4

-40

& 11"

each agonist.

The percent

o-Ala*-Mets-enkephalinamide (DALA) have both been shown to hyperpolarize and/or inhibit the discharge of interneurons (Nicoll et al., 1980; Madison and Nitoll, 1988b). Baclofen (1 PM) reduced the stratum orienslalveusand straum radiatum-induced monosynaptic evoked IPSPs by 45% + 13% and 56% + 14%, respectively (see Davies et al., 1990). Similarly, DALA (‘10 vM) potently reduced both the stratum radiatumand the statum orienslalveus-induced monosynaptic evoked IPSPs. These results are summarized in Table 1. As further illustrated in Figure 6C, stratum radiatum stimulation produced both an early and a late monosynaptic evoked IPSP. This two-phase IPSP was not completely blocked by picrotoxin (100 PM), but was blocked by a mixture of picrotoxin and 2-hydroxysaclofen (300 FM). The stratum oriens-induced monosynaptic evoked IPSP, however, consisted of an early component only, which was completely abolished by picrotoxin. Monosynaptic evoked IPSPs were also eliminated by the application of TTX (1 PM, n = 5, data not shown). The failure of NE and EPI to affect monosynaptic evoked IPSPs excludes several potential sites of adrenergic action, including the interneuronal axon, the interneuronal terminal, and the pyramidal cell. It does not, however, excludethe possibilitythat theseagents are acting on the interneuronal somata or dendrites. We formed this conclusion because significant monosynaptic IPSPs can be evoked by stimulating electrodes located at some distance from the recording :,ite. In fact, stimulating electrodes must often be placed, more than a millimeter from the recording electrode to produce an IPSP that is completely blocked by CNQX and APV. This suggests that monosynaptic evoked IPSPs may result from electrical stimulation of the extensive axonal arbors of interneurons, rather than by direct stimulation of the somata of these neurons. Thus, it is difficult to use the monosynoptic evoked IPSP as a diagnostic tool for changes at the interneuronal soma. The finding that a significant monosynaptic component is present in IPSPs normally evoked by electrical stimulation in the CA1 region may explain in part why adrenergic agonists can only partially reduce IPSP ami,litude even with the late IPSP blocked (see Figure 4).

change

of IPSP was taken

13"

as the change

in the amplitude

--~

However, in normal conditions, even purely polysynaptic IPSPs are not completely blocked by adrenergic agonists because of the presence of the late IPSP.

Monosynaptic

Spontaneous IPSPs

As a further test for potential inhibitory adrenergic effects on interneurons, we examined spontaneous TTX-sensitive IPSPs. These events, as distinct from TTX-insensitive miniature IPSPs, are generated by the action potential discharge of interneurons. Thus, they serve as an index of the excitability state of the interneuron (Alger and Nicoll, 1980). Recording in the presence of CNQX (10 PM) and APV (50 wM) ensures thatthespontaneous IPSPsareonlythosearisingfrom the intrinsic activity of interneurons, rather than from tonic glutaminergic activation of interneurons. In recordings from pyramidal cells loaded with KCI we found that the addition of CNQX/APV tothe perfusate decreased the frequency of spontaneous IPSPs by an average of about 14% (from 13.1 + 4.1 to Il.4 + 5.0 Hz, n = 15; p < 0.05). As all events were blocked by picrotoxin, the portion of l-TX-sensitive spontaneous IPSPs that were sensitive to CNQX/APV was most likely the result of glutamate released onto the interneuron, exciting it to fire action potentials, which in turn generated IPSPs. As with the TTX-insensitive events, monosynaptic spontaneous IPSPs were also completely blocked by the GABA* antagonists picrotoxin (100 wM), bicuculline (10 PM), or SR-95531(1 PM), thus confirming that these synaptic events were GABAergic in nature. If adrenergic agonists inhibited the interneuron, their application should decrease the rate of monosynaptic spontaneous IPSPs. Conversely, if adrenergic agonists excited interneurons, then the frequency of monosynaptic spontaneous IPSPs recorded in pyramidal cells should increase. The frequency of monosynaptic spontaneous IPSPs, as illustrated in Figure 7, was greatly increased by the application of NE (10 vM) or EPI (IO FM). This indicates that adrenergic agonists excite, rather than inhibit, interneuronal activity. Monosynaptic spontaneous IPSP frequency was also increased by application of the selective a-adrenoceptor agonist 6-fluoroepinephrine (10 PM; Adejare et al., 1988) and to a much lesser extent by phenylephrine

NeLlrOll 896

trated in Figure 7, the monosynaptic IPSPs were completely abolished by the onist SR-95531 (I PM, n = 3), as well as (100 PM, n = 7, data not shown). The adrenergic agonists increase interneuronal gests that inhibition of the interneuronal

spontaneous GABAA antagby picrotoxin finding that firing sugsoma (loca-

tion

of adrenergic

6 in Figure

1) is not

the

mechanism

disinhibition. Discussion

Mi”“feS

Figure 7. Epinephrine Spontaneous IPSPs

Increases

the Frequency

of Monosynaptic

All records in this figure were taken from the same CA1 pyramidal cell recorded with an intracellular electrode filled with 3 M KCI and 100 m M QX-314 and were made in the presence of the glutamate blockers CNQX (10 vM) and APV (50 PM) in the bath to isolate interneurons from excitatory synaptic input and to eliminate any spontaneous EPSPs impinging onto the pyramidal cell. QX-314was used to prevent large spontaneous depolarizing IPSPs from eliciting action potentials. (A,) Monosynaptic spontaneous (lTX-sensitive) IPSPs recorded in control (CNQX/APV) ACSF. Each sweep is a 1 s long, nonaveraged record. The records in this and other panels are consecutive. Sweeps were taken at a rate of 1 every 2 s. (A,) Monosynaptic spontaneous IPSPs recorded after the addition of EPI (IO PM for 10 min) to the bath. (A,) Monosynaptic spontaneous IPSPs after washing with EPIfree ACSF (for 1 hr). (A,) Monosynaptic spontaneous IPSPs after applying the CABAA receptor antagonist SR-95531 (1 NM for 8 min). (B) Frequency versus time plot for the entire experiment. Each bin is the frequency (in Hz) averaged over a 5 s epoch. In addition to EPI (IO PM), 6-fluoroepinephrine (6F-EPI, 10 PM), phenylephrine (PE, 10 PM), isoproterenol (ISO, 10 PM), norepinephrine (NE, 10 PM) and SR-95531 (SR, 1 PM) were applied at the time periods indicated by the bars. Note that SC95531 was added before the NE wash was complete, showing that the increased events as well as the baseline events were mediated through a GABAA receptor. As with picrotoxin in Figure 5, SR-95531, a CABAn antagonist, decreases the amplitude of IPSPs, leading to an apparent decrease in frequency. The drugs were bath-applied for the periods indicated by the bars. The membrane potential of this cell was -57 mV.

(IO PM).

The

frequency

of monosynaptic

spontaneous

IPSPs was not, however, much affected by the selective fl-adrenoceptor agonist isoproterenol (IO PM). These results are summarized in Table 2. This rank order of agonist potency EPI > NE >> ISO, is consistentwith mediation byan a-adrenoceptor.Thesefindings are in accord with previously published in vivo studies which showed that NE potently increased the activityof hippocampal interneurons bytheactivation of an a-adrenoceptor (Pang and Rose, 1987). As illus-

The results presented in this paper provide evidence that the disinhibitory action of adrenergic agonists arises through a mechanism that resides at a highly restricted locus. We have ruled out several possible sites of adrenergic disinhibition: the excitatory synapses onto pyramidal cells (location 1 in Figure I), because the intracellular EPSP and field EPSPs are not reduced by adrenergic agonists (Figures 2B and 2D; Madison and Nicoll, 1988a); the feedforward inhibitory circuit through the lacunosum-moleculare interneuron (location 2 in Figure I), because the late IPSP is not affected by adrenergic agonists (Figures 3 and 4); the pyramidal cell (location 3 in Figure I), because the reversal potential of the IPSP, the size and reversal potential of iontophoreticallyapplied GABAmediated hyperpolarizations (Madison and Nicoll, 1988a), and the monosynaptic evoked IPSPs (Figure 6) are not affected byadrenergic agonists; the interneuronal terminal or axon (locations 4 and 5 in Figure I), because miniature IPSPs (Figure 5) and monosynaptic evoked IPSPs (Figure 6) are not affected by adrenergic agonists; and the interneuronal soma, because monosynaptic spontaneous (TX-sensitive) IPSPs are increased (Figure 7), not decreased, by adrenergic agonists (location 6 in Figure 1). Thus, by the process of elimination, we have formed the hypothesis that a-adrenoceptor activation reduces the evoked early IPSP by inhibiting the excitatory synaptic activation of interneurons, perhaps at a presynaptic location. This action occurs only at excitatory inputs to the stratum orienslalveus interneurons and/or basket cells (location

7 in Figure

1). Determining

whether

this

action

is

pre- or postsynaptic to the interneuron will require further experiments involving recordings from interneurons. The insensitivity of the late IPSP to adrenergic agonists supports the idea that the early and late IPSP are produced by two different populations of interneurons. This is in agreement with several anatomical and physiological studies indicating the presence of heterogeneous populations of interneurons in the hippocampus (Knowles and Schwartzkroin, 1981; Schwartzkroin and Kunkel, 1985; Lacaille et al., 1987; Lacaille and Schwartzkroin, 1988a, 198813; Kawaguchi and Hama, 1988). These morphological and electrophysiological studies suggest that the stratum oriens/alveus interneurons and basket cells mediate the early IPSP, whereas lacunosum-moleculare interneurons (which send numerous processes into stratum radiatum) may

Ad energic 89;

Disinhibition

in Hippocampus

%le 2. The Effects of Noreoinephrine and Other lntracellularly in CA1 Pyramidal Neurons -

Agonists

on the Frequency .

of Monosynaptic I.

Spontaneous

Compound

Concentration

n

Percent Change in Frequency (+ SD)

Eplnephrine Norepinephrine Isoproterenol DALA 6-f luoroepinephrine Phenylephrine

10 10 10 10 IO 10

4 9 4 7 2 2

+I50 +704 +25 -67 +200 +63

PM PM PM PM uM PM

n refers to the total number of cells tested with each agonist. The percent of monosynaptic spontaneous IPSPs and is for all cells tested, including 3 Data significant at p < 0.01.

give riseto the late IPSP. Our resultsare highlyconsistent with this idea and further suggest that only stratum oriens/alveus interneurons and basket cells are involved in adrenergic disinhibition. It has been demonstrated that activity-dependent mechanisms are responsible for some forms of disinhibition. Two such mechanisms have been identified. First, during repetitive activation of evoked IPSPs, chloride accumulation in the postsynaptic pyramidal cell changes the IPSP reversal potential, leading to a decrease in IPSP amplitude (McCarren and Alger, 1985; Thompson and Gahwiler, 1989a, 1989b). Second, during repetitive activation of IPSPs, GABA accumulation in the extracellular space has been shown to activate auto-inhibitory GABAe receptors, presumably on interneuronal terminals (Bowery et al., 1983; Dutar and Nicoll, 1988b; Deisz and Prince, 1989; Thompson and Gahwiler, 1989c; Davies et al., 1990; Harrison, 1990). Because adrenergic agonists increase the frequency of spontaneous IPSPs, it is conceivable that the adrenergic inhibition of evoked IPSPs may occur through one of these activity-dependent mechanisms. However, the data, previously published and also presented in this paper, rule out both of these activity-dependent mechanisms as the basis for adrenergic disinhibition. First, NE reduces evoked IPSPs without changing the reversal potential of the IPSP (Madison and Nicoll, 1988a); thus, chloride accumulation in the postsynaptic pyramidal cell cannot account for adrenergic disinhibition. Second, 2-hydroxysaclofen, applied at concentrations shown to block presynaptic GABAe receptors (Davies et al., 1990), does not prevent adrenergic agonists from reducing the IPSP (see Figure 4); thus, GABA accumulation in the extracellular space, acting on auto-inhibitoryCABAs receptors, does not account for adrenergic disinhibition. One significant controversy in the literature on noradrenergic effects in cortical tissue is whether NE exerts a net excitatory or inhibitory effect. Most of the cellular electrophysiological effects of NE that have been reported in the hippocampus are excitatory in nature. These excitatory effects include an increase in the population spike amplitude (Mueller et al., 1982), depolarization of pyramidal cells(Madison and Nicoll,

IPSPs Recorded --

f + f f

46” 67” 50 24a

change in frequency “no effect.”

No Effect 0 0 3 0 0 0

was taken as the change

in the number

1986a), inhibition of calcium-activated potassium currents (Madison and Nicoll, 1982; Haas and Konnerth, 1983), and disinhibition (Mody et al., 1983; Leung and Miller, 1988; Madison and Nicoll, 1988a). In contrast, most studies of the effects of NE in vivo suggest that the predominant effect of NE in the hippocampus is inhibitory, decreasing the spontaneous firing of principal neurons (Segal and Bloom, 1974a, 1974b; Mueller et al., 1982; Curet and de Montigny, 1988). Similarly, in animal models of altered noradrenergic function, decreased noradrenergic function in the cortex gives rise to greater excitatory influences, such as a decrease in seizure thresholds, whereas increased noradrenergic cortical function results in greater inhibitory influences and is potently anti-epileptogenic (see reviews by Chauvel and Trottier, 1986; Ferrendelli, 1986). In conjunction with these in vivo findings, our results raise the possibility that the most important effect of NE, in terms of the normal function of the cortical areas, may not be the excitatory effects listed above, but rather the profound increase in tonic inhibition that is evident as an increase in the rate of monosynaptic spontaneous IPSPs (Figure 7). The localization of the disinhibitory action of NE to a very confined site in the synaptic circuitry is unexpected, given the diffuse organization of noradrenergic afferents in the brain and particularly in the hippocampus (see reviews by Moore and Bloom, 1979; Cooper et al., 1986). Norepinephrine-containing fibers from the locus ceruleus distribute throughout the hippocampal formation in a diffuse manner quite distinct from the highly laminated nature of many other hippocampal inputs (Loy et al., 1980). Furthermore, it may be the case that transmitter-releasing varicosities on these noradrenergic axons may not always form synapses onto postsynaptic specializations, but may release NE in a “hormonal”fashion. The localization of the noradrenergic disinhibitory action to the excitatory synapses innervating a subset of interneurons suggests that despite the diffuse properties of the noradrenergic inputs, specificity of noradrenergic action may still be maintained because of extreme localization of the target receptors on very discrete anatomical sites.

NWWXl 898

Experimental

Procedures

Slice Preparation Experiments were performed on in vitro hippocampal slices prepared from young adult male Sprague-Dawley rats (150-250 g). Slices were prepared by standard procedures (see Madison and Nicoll, 1986a). Preparation of the tissue is performed in strict accordance with a protocol approved by the Stanford University Animal Use and Care Committee. Briefly, hippocampi are dissectedfrom thebrainandcut into500pm slices usingacommercial sectioning apparatus. Slices are then placed in an incubation chamber, where they rest on a piece of filter paper moistened with ACSF in an atmosphere of 95% 02, 5% CO*. After at least an hour in the incubation chamber, an individual slice is transferred to the recording chamber, where it is submerged beneath a continuously flowing stream of ACSF that has been pre-gassed with 95% 02, 5% CO,. The composition of the ACSF was 119 m M NaCI, 5 m M KCI, 1.3 m M MgSO+ 2.5 m M CaC12, 1.0 m M NaH*PO+ 26.2 m M NaHC03, 11.0 m M glucose. The temperature of the medium was maintained at 22°C. Electrophysiological Recording Synaptic potentials were evoked by delivering constant current electrical stimuli of 100 ps duration through bipolar platinum/ iridium stimulating electrodes placed in stratum radiatum or stratum oriens/alveus of the slice. Extracellular field potentials were recorded with glass micropipettes filled with 3 M NaCL Field EPSPs were recorded by placing the electrode in stratum radiatum approximately midway between stratum pyramidale and the hippocampal fissure. Population spikes were recorded by placing the recording electrode in stratum pyramidale. Intracellular electrodes were pulled from thin-wall alumina silicate glass tubing on a Flaming/Brown Micropipette Puller Model P87 (Sutter Instruments, Novato, CA). These electrodes were filled with either 2 M KMeS04 (ICN Pharmaceuticals, Irvine, CA) or 3 M KCI (J. T. Baker, Phillipsburg, NJ) as a recording electrolyte. Intracellular recording electrodes were driven through the slice using a Newport CorporaGon (Fountain Valley, CA) series 860 motorizer that replaced the fine Z-axis micrometer on a Marzhauzer MM-33 micromanipulator (Sutter Instruments). Impalement was achieved by briefly overcompensating the capacitance compensation circuit on the electrometer (Axoclamp IIA). Experiments recording spontaneous IPSPs had the intracellular sodium channel blocker QX-314 (100 mM), a quaternary lidoCaine derivative, present in the electrode to suppress the generation of action potentials, which can obscure the spontaneous IPSPs. During drug applications, the pyramidal cell membrane potential was held constant with D.C. injected through the recording electrode. Data Analysis Data were digitized on a Scientific Solutions 125 kHz labmaster analog to digital converter (Axon Instruments TL-125 interface) and were stored and analyzed using Axobasic-based software (Axon Instruments, Foster City, CA) running on an 80386 computer. Using this system, data were analyzed on-line and then stored for further analysis at a later time. Each sweep of data illustrated consisted of 1024 individual points, collected at sample rates ranging from 0.5 to 20 kHz, depending on the desired tempoial resolution. IPSP amplitudes were measured by the computer relative to the pre-stimulus baseline. Spontaneous IPSP frequency was determined by a software window discriminator and rate meter running under Axobasic and written by us specifically for this purpose. Each sweep of data (1024 points) was differentiated over a 4 point span. The computer scanned the differential, marking each position having a large dvldt as a potential location of the rising phase of a spontaneous IPSP. The raw record was then re-analyzed, with the computer examining each location with a large dv/dt, looking for an event above a threshold level. The threshold was set just above the peak of the noise on a trace that had no spontaneous IPSPs. Examining the record for large dv/dt and also for threshold crossing eliminates the most common analysis errors. If only the differential was used, significant number of noise events, with a fast rate of

change, were counted as IPSPs. If only threshold crossing was used, then slow shifts in membrane potential or current were counted and multiple superimposed events were counted as a single event. By demanding that an event both have a high dvi dt and cross a threshold, we have found that the program gave approximately 98% accuracy in correctly identifying synaptic events. Data are presented as mean values + standard deviations and were analyzed for statistical significance using the paired t test. P values of