Physiological evidence for two distinct GABAA responses in rat hippocampus

Physiological evidence for two distinct GABAA responses in rat hippocampus

Neuron, Vol. 10, 189-200, February, 1993, Copyright 0 1993 by Cell Press Physiological Evidence for Two Distinct GABA* Responses in Rat Hippocamp...

1MB Sizes 0 Downloads 5 Views

Neuron,

Vol. 10, 189-200,

February,

1993, Copyright

0 1993 by Cell Press

Physiological Evidence for Two Distinct GABA* Responses in Rat Hippocampus Robert A. Pearce Departments of Anesthesiology and Neuroscience Training University of Wisconsin Madison, Wisconsin 53706

Results and Program

Anatomy

Summary The y-aminobutyric acidA (GABAJ receptor is a ligandgated ionophore involved in synaptic inhibition. Biochemical and molecular biological studies indicate that considerable receptor heterogeneity exists, but physiological differences between inhibitory GABAA synaptic responses have not been identified in the brain. The present report describes two anatomically segregated CABAAmediated synaptic currents in the hippocampal CA1 region that have distinct physiological, pharmacological, and functional properties. CABAA,faSt enters at or near the cell body, decays rapidly (3-8 ms), is blocked by furosemide, and rapidly curtails the excitatory response. CABA,+I,, enters far from the cell body, decays slowly (38-78 ms), is not blocked by furosemide, and underlies the conventionally recognized early inhibitory postsynaptic potential. The receptors producing these responses may represent subtypes of the CABAA receptor. Introduction y-Aminobutyric acid (GABA), the major inhibitory transmitter in the vertebrate nervous system, exerts postsynaptic actions via two distinct classes of receptors, GABAA and GABAe (Dutar and Nicoll, 1988; Bormann, 1988; Sivilotti and Nistri, 1991). The GABAA receptor, a ligand-gated chloride ionophore, is the target of several therapeutically important classes of drugs, including sedative hypnotics, anxiolytics, anesthetics, and anticonvulsants (Sivilotti and Nistri, 1991). Molecular biological and receptor binding studies havedemonstrated that multipleGABAA receptor subtypes exist (Sieghart, 1989; Olsen and Tobin, 1990), and patterns of GABAA subunit isoform mRNA expression indicate that in some brain regions, including the hippocampus, multiple GABAA subtypes may coexist (Wisden et al., 1992). Electrophysiological characteristics of GABAA receptor-channel complexes expressed in vitro have been shown to depend upon subunit composition (Verdoorn et al., 1990), but physiological distinctions between GABAA synaptic responses corresponding to the demonstrated molecular heterogeneity have not been identified in the hippocampus or elsewhere in the central nervous system. The present report describes two GABAA-mediated synaptic responses in CA1 neurons of the rat hippocampus that are anatomically segregated and that have distinct physiology, pharmacology, and functional roles.

In the hippocampal slice preparation (Nicoll and Alger, 1981; Alger and Nicoll, 1982), electrical stimulation of stratum radiatum (SR) evokes a mixed excitatory-inhibitory synaptic response in CA1 pyramidal neurons via the Schaffer collaterallcommissural excitatory input and polysynaptic inhibitory pathways (Kandel et al., 1961; Alger and Nicoll, 1982; Newberry and Nicoll, 1984; Knowles and Schwartzkroin, 1981; Lacailleetal.,1987;LacailleandSchwartzkroin,1988b). An isolated monosynaptic GABA*-mediated chloride current may be evoked by using a combination of glutamate receptor antagonists to block excitatory inputs (Davies et al., 1990) and either the GABAe antagonist 2-hydroxy-saclofen in the extracellular medium or cesium ions and lidocaine N-ethyl bromide (QX314) in the recording pipette solution to block the GABAe component of the inhibitory postsynaptic current (IPSC) (Nathan et al., 1990; Kerr et al., 1988). Under these conditions, electrical stimulation of SR approximately 500 urn from the recording site evoked monosynaptic isolated GABAA-mediated IPSCs in voltageclamped rat hippocampal CA1 neurons (Figure IA). The decay phase of the IPSC was best fit by the sum of two exponentials (Figure IB) with time constants of 3-8 ms (fast component, Tfast) and 30-70 ms (slow component, ~~1~~) (Figure IC). Both components were blocked by bicuculline (Figure IAb and IAc), and both time constants were voltage dependent (Tfast 4.0 + 0.8 ms at -80 to -100 mV [n = 17],6.9 f 1.6 ms at -30 to -50 mV [n = 17; 7s~0w 39.2 f 5.8 ms at -80 to -100 mV [n = 151, 57.6 f 8.6 ms at -30 to -50 mV [n = 151; mean f SD). Current risetimewas independent of the relative amplitudes of the fast and slow components (Figure ID), and both components were present and clearly separated over a wide range of relative amplitudes (Figure IE). The relative amplitudes of the fast and slow components varied widely between cells in response to SR stimulation (Figures IAa and IAb). This variability was found to be associated with the location of the stimulating electrode. Stimuli delivered near to or directly onto the pyramidal cell body layer, stratum pyramidale (SP), evoked synaptic currents with predominantly fast decays, whereas stimuli to SR far from SP or into stratum lacunosum-moleculare (SL-M) evoked slowly decaying currents (Figure 2A). Responses to both stimuli were blocked by bicuculline (Figure 2A), and fast and slow components reversed at the same potential (Figure 2B). This response pattern was seen in seven of seven cells in which separate bipolar stimulating electrodes were placed in SP and SL-M. When a single bipolar stimulating electrode was moved in 50 urn steps across SR from SP to SL-M while maintaining the cell impalement, the fast component be-

.

Figure 1. Monosynaptic CABA,-Mediated IPSCs in CA1 Pyramidal Neurons Decay with &exponential Kinetics

Bicuculline 10pM C.

:-

1 nA

1 nA 1

1 nA c 30 ms

30 ms

C

Monoexponential

8or

. .

. .

. .

02

.

.

. l .

l

. .

l .

.

r’

. ..*

L 0.1

i . ..

..

0.3

0.5

.

2.

.

0.

9..

.

.

.

2

Amplitude (fast) I Amplitude

3

5

(slow)

10

0.2

.

*

.

l ,

.

l * .

l

c.,..:-l .- . . l ‘. . .

l

t .

:

1

.

l

. . .*

.

0.3

0.5

1

23

Amplitude (fast) /amplitude

came smaller, until only the slow component remained. Weak stimuli applied directlyto SP invariably resulted in predominantly fast decays, and in some cases no slow component was seen. However, as stimulus strength was increased, a slow component of current decay became apparent, the amplitude of which increased with increasing stimulus intensity, suggesting that it was a consequence of current spread to the dendritic region. Together these observations suggest that anatomically segregated populations of inhibitory interneurons or axonal processes that are being stimulated selectively give rise to the fast and slow current decay components. If this is the case, then the current components might enter the cell at anatomically segregated sites. This possibility was supported by the results of a series of experiments in which recordings were made with KCI or CsCl electrodes. As expected for neurons with an increased internal chloride ion concentration, the reversal potentials of the evoked currents were shifted to less negative potentials compared with measurements made with potassium acetate (KAc) or cesium acetate (CsAc) electrodes. However, reversal potentials of the fast and slow components were shifted by different amounts (Figures 2C and 20; Table 1). At hyperpolarized potentials, cur-

5

(slow)

10

(A) SR stimulation evoked bicucullinesensitive currents at -35 mV to -95 mV (top to bottom traces) that decayed at different rates in different neurons (a and b). Recordings were made with electrodes containing CsAc (3 M) and QX-314 (50 mM) to block the CABAB component of the IPSC and with 2-amino-5-phosphonovalerate (40 PM) and 6-cyano&nitroquinoxaline-2,3dione (20 uM) in the extracellular medium to block excitatory currents. Addition of bicuculline (IO PM, 12 min) blocked the evoked currents (c, same cell illustrated in b). (B) Current decays were best fit by the sum of two exponentials. Response at -45 mV (from b above),with superimposed best fits of mono- and biexponential decays. Monoexponential: T = 15.2 ms, amplitude = 0.87 nA. Biexponential: r(fast) = 5.6 ms, amplitude (fast) = 0.78 nA; ~(slow) = 56.7 ms, amplitude (slow) = 0.38 nA. (C) Data from nine cells show that the fast (open circles) and slow (closed circles) components of decay are clearly segregated at all voltage levels. For two traces only a single exponential process was evident, and data from traces at or near the reversal potential of the evoked currents were not included. (D and E) 20%-80% rise times and decay rates of evoked currents in response to SR stimulation plotted as a function of the relative amplitudes of fast and slow components; data are from same cells as illustrated in Figure IC.

rent decays appeared similar to those made with CsAc electrodes. However, at voltages between the reversal potentials of the two components, an early inward current was followed by a slowly decaying outward current (Figure 2C, top two traces). A larger shift in reversal potential with chloride loading would be expected for synaptic inputs onto or close to the cell body, where the recording electrode impales the cell, than at distant dendritic sites, due to the activity of a membrane chloride pump that maintains a low internal chloride ion concentration (Thompson et al., 1988). Therefore, from the results of these experiments itwouldappearthatthefastcurrentdecaycomponent enters at or near the cell body, and the slow decay component enters moredistally, i.e., in thedendrites. This possibility, that the synaptic current sites for the two components are anatomically segregated, was tested directly in experiments in which bicuculline was applied focally to the cell body layer or to the dendritic region while stimulating SPand SL-M electrically (Figure 3). The peak current in response to SP stimulation was greatly reduced by application of bicuculline to SP, but the response to SL-M stimulation was reduced only slightly (Figures 3A and 3C). Conversely, in the same cell, dendritic application of bicu-

Two 191

GABA*

A SL-M

Responses

in Rat Hippocampus

CsAc

stim.

electrode

Control

B

Bicuculline

CsAc

3.0

electrode

i

SP stim. .s., : c, : !

1 nA 20 ms

-100

-90

-90

-70

-60

-50

40

30

V WI

D 1.0

C SR stim

CsCl

CsCl

r

electrode

-80

:<.;-‘i :f !i 2. Anatomically

-70

-60

1 nA

.. ‘F

Figure

electrode

-50

-40

30

V (mv)

I 20 ms Segregated

(A) Electrical stimulation responses were completely mV. (B) Current-voltage 28-50 ms) current decay decay component (open by use of a CsCl/QX-314 artifact. (D) Current-voltage slow (closed circles, T =

Inputs

Give

Rise to the

Fast and

Slow

Components

of IPSC Decay

of SL-M evoked slowly decaying currents, whereas stimulation of blocked by bicuculline (IO PM). Recording conditions were as relation of IPSC components; data from (A). Fast (open circles, z components of SP response reversed at the same potential. SL-M squares, r = 33-60 ms). (C) Responses to SR stimulation at -35 to electrode. Holding currents have been supermposed to distinguish relation of IPSC components with CsCl electrode; data from 21-59 ms) components reversed at different potentials.

culline preferentially reduced the SL-M response (Figures 3B and 3D). For the experiment illustrated, a single brief puff of bicuculline to SP was sufficient to block the SP response rapidly and reversibly, and this protocol was repeated fourtimes to yield the averaged currents plotted in Figure 3C. However, no site was identified in SR or SL-M that resulted in a comparable block of the slow component with such a limited application of bicuculline. The 60% block of the SL-M response illustrated in Figure 3D was achieved by pressure ejecting bicuculline at multiple sites in rapid succession alongthe SL-M-SR border. In thiscase, the slow SL-M response was blocked to a greater extent and sooner than was the fast SP response, but the slow response was not blocked as selectively as was the fast response upon SP application. This experiment was repeated in three slices,with the same result in each: that the SP response could be blocked more selectively and with a more focally limited application of bicuculline than could the SL-M response. From these experiments, it appears that the fast component in response to SP stimulation is spatially restricted to the cell body region, but that the slow response enters

SP evoked currents with a fast decay. Both in Figure 1; holding potentials, -35 to -95 = 4.0-8.2 ms) and slow (closed circles, z = stimulation resulted in only a single slow -75 mV with elevation of internal chloride the onset of the response from stimulus (C). Fast (open circles, T = 2.3-4.1 ms) and

overmoredistributeddendriticsites.Indeed,thedendritic sites for the SL-M response may extend across thecell body layer into stratum oriens,for itwas noted in one experiment that application of bicuculline to stratum oriens resulted in a partial block of the slow, but not fast, response. However, this finding was not examined systematically in this or in other experiments. These two independent lines of evidence, i.e., different shifts of reversal potential with chloride loading and different sensitivities to focal application of bicuculline, support the notion that the fast and slow current decay components enter at anatomically segregated sites, the cell body and dendritic regions, respectively. Because the extended geometry of hippocampal pyramidal cells can result in significant distortion of synaptic currents due to inadequate space clamp, it was necessary to test whether the slow component results from the distortion of a current that is kinetically identical to the fast component but altered by dendritic filtering, rather than a kinetically distinct current.This possibilitywastested byelectricallystimulating SP and SL-M while holding acell at the reversal

Table

1. Effect

of Filling

Solution

on Reversal

Electrode

Filling

Solution

CsAclQX-314 Reversal potential fast component Reversal potential slow component

of (mV) of (mV)

tance was still active. By stepping to a new voltage at varying intervals following the synaptic activation, a time course of conductance of the two responses could thus be determined. Since the step to a new voltage itself resulted in a transient current, it was necessary to subtract this transient in the absence of synaptic stimulation from the step following synaptic activation, to determine the effective synaptic current. The results of this experiment are shown in Figure 4. The transient currents resulting from steps from -60 mV to -90 mV with and without prior synaptic stimulation of SP or SL-M are shown superimposed in each panel, and the resulting digital subtractions of the two traces are shown in the insets. Figure 4A illustrates synaptic stimulation at the same time as the voltage step (Figures 4Aa and 4Ab) and 20 ms prior to a voltage step (Figures 4Ac and 4Ad). It is apparent that 20 ms following the SL-M stimulus, a synaptic conductance is still active, as shown by the different current responses to thevoltage steps and bythe subtracted (synaptic) current (Figure 4Ac). However, 20 ms following the SP stimulus, the transient responses

Potential

CsCl/QX-314

-61.0

x!z 4.7 (16)

-26.0

+ 3.2 (4)

-62.8

+ 2.8 (15)

-49.5

+ 2.9 (4)

All values are expressed as mean + SD (r-r). Reversal potentials were significantly different following chloride loading, and after chloride loading the fast and slow components were significantly different from each other (Student-Newman-Keuls ANOVA).

potential of the synaptic currents, then after a variable time interval steppingtoadifferent potential. By holding the cell at the synaptic reversal potential, no current was allowed to flow and thus no charge to be builtupin thedendritestoredistributelatertothecell bodyasadistorted synapticcurrent.Upon steppingto a potential away from the reversal potential, synaptic current would then flow only if the synaptic conduc-

B SP stim.

SL-M stim.

SP stim.

SL-M

stim.

0.2 nA

0.2 nA

L50 ms

L-

50 ins

D 1.2

c 1.2 iz

SL-M RESPONSE

1 .o

g

1.0

-6 0.8

.tj 0.8

G $- 0.8 ?? 2 0.4 -2

0.8 0.4 0.2

2 A BICUCULLINE ~ICUCULLINE

0'

100 Time

Figure

3. Focal

Application

0.2 -

TO SP

50

0

L-M RESPONSE

I 150

0

BICUCULLINE TO SL-M

100

(s) of Bicuculline

Revealed

That

the Fast and

Slow

200 Time (s)

300

IPSC Components

400

Are Anatomically

(A and C) Bicuculline application to SP. SP stimulation followed by SL-M stimulation resulted in fast and slow ejection of bicuculline from an electrode positioned in SP selectively blocked the SP response, as seen in in the average of four successive applications (C). (B and D) Bicuculline application to SL-M&R. Pressure multiple sites along the SL-M/SR border led to an early block of the SL-M response, followed by a smaller Holding potential, -45 mV; all data are from the same cell.

Segregated

decays (control). Pressure the original record (A) and ejection of bicuculline at block of the SP response.

Two 193

GABA,

Responses

in Rat Hippocampus

SL-M stim. (-20 ms)

a

b

SL-M STIMULUS TIME (Ins)

SP STIMULUS TIME Ims)

HOLDI IOms

Figure

4. Time

Course

of Conductance

Changes

Matches

the Time

Course

of Current

Decays

for Both

IPSC Components

SP and SL-M stimuli were applied at a holding potential of -60 mV at a variety of intervals preceding a step to -90 mV. (A) Examples of the resulting currents at two intervals, 0 and 20 ms, are illustrated for each stimulus. The currents resulting from identical steps without prior synaptic stimulation are superimposed on each trace, and the results of digitally subtracting the traces are shown in the insets. Note that when the step followed the stimulus by 20 ms, only the SL-M response remained (c versus d). (B) Summary of results at several intervals between stimulus and step (numbers shown refer to time of stimulus compared with time of step, in milliseconds) and 2000 ms following a step to -90 mV (holding). The peak amplitudes and time course of currents reveal an envelope of conductance that matches the current decay for both components.

tothevoltagestepsareindistinguishable,andthesubtracted (synaptic) response is flat (Figure4Ad), indicating that the synaptic conductance is no longer active. in contrast, both the SP and the SL-M responses are clearly present when the synaptic stimulus is presented at the same time as the voltage step (Figures 4Aa and 4Ab). Figure 4B presents a summary of subtracted synaptic currents at a variety of intervals between synaptic stimulus and voltage step, aligned at the time of the stimulus, and the evoked synaptic current follotiing a 2 s holding period at -90 mV for comparison (holding). The time course of conductance clearly follows the time course of the synaptic currents for both the slow and the fast responses. This experiment was repeated in three slices with identical results. Even when the holding potential during synaptic stimulation was slightly depolarized relative to the synaptic current reversal potential and outward currents were elicited, they reversed to inward currents upon hyperpolarization. Thus, the results of this experiment demonstrate that the slower current decay in response to SL-M stimulation was not simply the result of dendritic filtering, but rather that the SP and SL-M cwrrents have different conductance time courses.

Additional evidence that the fast and slow current decay components are separate and distinct responses was provided bythe finding that furosemide, an inhibitor of chloride transport that is used experimentally to facilitate chloride loading of cells by preventing its rapid extrusion (Thompson et al., 1988; Misgeld et al., 1986), had the additional effect in these cells of selectively and reversibly blocking the fast but not the slow current decay component (Figure 5). This effectwas seen in seven of seven cellstested, of which two were held long enough to demonstrate reversibility after drug washout (e.g., Figure 5A). This effect of furosemide was not due to an alteration of chloride distribution, for a complete block of the fast component could be seen with little or no change in its reversal potential as the block was developing (data not shown), and the reversal potential of the slow component was changed little (Figure 5A) or not at all (Figure 5B). A rapid redistribution of chloride to its passive equilibrium potential in the presence of furosemide, confined to the cell body region, could result in the loss of driving potential for current flow and thus an apparent block of the fast current. However, this redistribution would beaccompanied byacurrent relaxation during the voltage steps, and no such alteration

Nl?UKHl 194

Figure 5. Furosemide the Fast Component

Furosemide

Selectively of the Evoked

Blocks IPSC

(A) Fast and slow inhibitory currents in a single cell voltage clamped at -30 mV to -90 mV (20 mV steps) were evoked by stimulating electrodes located in SP and SL-M. Recording conditions were as in Figure 1. Furosemide (600 PM, 15 min) reversibly blocked the fast SP response but not the slow SL-M response. (B) The effect of furosemide (600 PM, 11 min) on the fast IPSC component was not due to chloride ion redistribution.Currentsduringthestepspreceding synaptic stimulation were not altered by furosemide (a), yet the fast IPSC component was blocked (ends of traces in a, shown on an expanded time scale in b). Holding potential, -50 mV; steps to -120 mV to -20 mV in 20 mV increments. A transient inward calcium current is seen at the beginning of the most depolarized step.

SP stim.

SL-M Him.

Furosemide L

I-,

I==

SP stim. 4

SL-M stim. -1-

SP stim. -1

SL-M Him. 4

of the currents or relaxation was detected (Figure 5B). Therefore, it can be concluded that furosemide blocks the conductance underlying the fast but not slow component. In addition to blocking the fast IPSC component, furosemide caused the current rise time following SL-M stimulation to be prolonged (control 3.3 + 1.4 ms; furosemide 5.7 + 2.2 ms, mean + SD, p < 0.05, 7 traces from 3 cells). It is possible that this effect was due to a direct action of furosemide on the channels underlying the slow component, causing a change in its opening kinetic properties. Alternatively, a small fast component may have been present in control recordings, having a faster onset than the slow component, and upon blocking the fast IPSC component with furosemide, the slow onset of a more pure SL-M response was revealed. Although no correlation between rise time and relative amplitudes of fast and slow components in response to SR stimulation was observed (Figure ID), when rise times in response to selective SP and SL-M stimulation were compared, SP rise times were significantly faster than were SL-M rise times (SP 2.3 -f. 0.4 ms [I5 traces from 7cells], SL-M 3.8 f 0.9 ms[19tracesfrom8cells], mean + SD, p
No correlation between rise and decay times within each group was seen (data not shown). These results are consistent with the hypothesis that the effect of furosemide on rise time was due to block of a small fast component contaminating the SL-M response, but again a direct effect on the slow channel cannot be ruled out. The anatomical segregation and kinetic differences between the fast and slow components of IPSC decay, which will be referred to as GABAA,rast and GABAA,,l,,, suggest distinct functional roles for the two currents. Specifically, GABA,+,t at the cell body would be expected to influence primarily the early phase of an evoked response, quickly shunting depolarizing current from the spike-initiating zone and limiting the response to excitatory inputs. In contrast, the slower dendritic inhibition, GABA,+,,,, should result in a prolonged and perhaps pathway-specific shunting of excitatory inputs. The selective blockade of GABAA,r,,r by furosemide was used to test this hypothesis by recording intracellular and extracellular responses to electrical stimulation ofthe Schaffer collateral/commissural input pathway. Intracellular current-clamp recordings of the

Two 195

GABAA Responses

in Rat Hippocampus

Figure 6. Distinct Functional GAB&W and GABA.A,,I~~

IOOms

B

Control

b Control /-y

Wash

Furosemide

Furosemide

Roles

for

(A) Intracellular recording in current clamp mode with a KAc-filled electrode. SR stimulation evoked a single action potential under control conditions. Blocking GABA,r,,, by application of furosemide (600 RM, 17 min) resulted in two action potentials in response to the same stimulus. Two superimposed traces for each response; resting potential, -73 mV; action potentials truncated. (b) The later phases of the evoked IPSP were not affected by furosemide. Same cell as in a, averaged responses to four stimuli. (B) Extracellular population responses in Spin responsetoSRstimulation.(a)Furosemide (600 PM, 12 min) resulted in a double population spike in response to a single shock. The effect was reversible with washout of the drug. (b) A prior stimulus delivered to the alveus (data not shown) depressed the population response to SR stimulus at 20-70 ms interpulse intervals (compared with control, no prior stimulus). Furosemide (600 NM, 14 min) did not prevent paired-pulse depression (same interpulse intervals). A second population spike was present in the control response with no preceding stimulus, as in Figure 4Ba.

A

evoked response to SR stimulation (Figure 6A) revealed a pronounced effect of furosemide only during the early phase of the reponse, allowing a more fully expressed excitatory postsynaptic potential to evoke two action potentials. At a slower sweep speed, both the early (GABAJ and late (CABA& hyperpolarizing phases of the IPSP were unchanged (Figure 6Ab). These experiments demonstrate that GABAA,~,,~ operates on the time scale of the excitatory postsynaptic potential, whereas GABA,+,, underlies the conventionally recognized early, or fast, IPSP (Dutar and Nitoll, 1988; Davies et al., 1990; Nathan and Lambert, 1991). In extracellular recordings of population activity, supramaximal stimulation of SR evoked a single population spike; furosemide led to a double population spike in response to an unchanged stimulus (Figure 6Ba, n = 8 of 8 slices). When the SR stimulus was preceded by a strong electrical stimulus to the alveus, a long-lasting inhibition of the SR response was obtained, even in the presence of furosemide (Figure 6Bb). Weaker stimuli to the alveus, which resulted in a relatively selective antidromic activation of CA1 cells (seen as an abrupt downward deflection from baseline), failed to elicit a prolonged depression; only by increasing the stimulus to the point that an excitatory postsynaptic potential (a positive deflection) preceded the population spike could a long-lasting in-

hibition of the SR response be demonstrated. This modified paired-pulse protocol, which is the in vitro counterpart of in vivo paired-pulse depression, is a bicuculline-sensitive process that is prolonged by barbiturates and volatile anesthetics (Wolf and Haas, 1977; Rock and Taylor, 1986; Kapur et al., 1989; Dunwiddie et al., 1986; Pearce et al., 1989). The similarities in time course, sensitivity to bicuculline, and insensitivity to furosemide suggest that GABAA,,r,, may be the underlying basis for paired-pulse depression. The selective effects of furosemide on the early phase of the intracellular evoked response and on single and paired extracellular responses provide evidence for distinct functional roles for GABAA,tast and GABAA,,,,,. Discussion Two components of the GABAA IPSC have been described that can be differentiated by their anatomical, pharmacological, physiological, and functional properties. Each of these differences will be discussed in turn, but first the potential contribution of spaceclamp error to the present results will be considered. Space-Clamp The extended cells results

Error geometry in an imperfect

of

hippocampal space clamp,

pyramidal which can

Neuron 196

lead to loss of voltage control, particularly for fast, large, active, and electrotonically distant currents and which can alter the apparent time course of fast passive currents at electrotonically distant locations (Rail and Segev, 1985; Johnston and Brown, 1983). This raises the question whether the observed fast and slow components of current decay represent trulydistinct types of GABA-gated currents with different channel kinetics, or are simply an artifact of an imperfect space clamp caused by filtering of high frequency components of identical currents entering the cell at different locations. Several lines of evidence suggest thatthetwocomponentsaretrulydistinctand notdue to imperfect space clamp. First, reversal potentials of the two components were the same, independent of stimulus location (Figures2A and 2B; Table 1). Second, imperfect space clamp attenuates high frequency components in a continuously variable manner as a function of frequency and electrotonic distance, leading to a continuous variation in (erroneous) estimated decay time constants at large electrotonic distances (Johnston and Brown; 7983). However, fast and slow decay components were clearly segregated regardless of the relative amplitudes of the two components (Figure IE). Third, current decay rates should be influenced less than rise times by inadequate space clamp, and IPSC rise times were fast and not correlated with the relative amplitudes of GABA,+,t and GABAA,,I,, (Figure ID). Fourth, even at the most electrotonically distant sites a 3 ms decay rate would be prolonged only to 10 ms (Johnston and Brown, 1983), much less than the 30-70 ms time constant of the slow component. Fifth, recordings made with KAc or KCI electrodes and 2-hydroxy-saclofen in the bathing medium yielded two decay phases with time constants similar to those made with CsAclQX-314 electrode:, despite lower input resistances (32.0 f 7.3 MD, n = 8, versus 59.9 + 16.8 MQ, mean f SD, n = 17), indicating a less electrotonically compact structure. Sixth, the selective action of furosemide on GABAA,rast was reflected by its effects on extracellular population responses and on intracellular current-clamp recordings, neither of which should be influenced by space-clamp artifact. The final and strongest evidence that space-clamp artifact cannot account for the different time courses of GABAA,r,,, and GABAA,s~,, came from experiments in which voltage steps followed synaptic activation by a variable time interval (Figure 4). These experiments demonstrated a difference in the time courses of conductance of the two components, not simply in the time courses of the evoked currents. By activating the synaptic conductance at the GABAA reversal potential, no current was evoked. In this way an excess unclamped capacitative charge was prevented from building up in electrotonically distant locations, i.e., the dendrites, that could then discharge back into the cell body recording site and appear as an artifactually prolonged synaptic current. Thus, upon stepping to a hyperpolarized potential, synaptic current would

flow only if the synaptic conductance remained. A potential problem with this method would arise in the case in which dendritic synaptic current would flow even when holding the cell body at the synaptic current reversal potential, due to a difference between membrane potential at the cell body and in the dendrites. However, the finding that GABAA,r,,t and have identical reversal potentials demonGABAA,,I,, strates that this has not occurred. Also, by holding the cell slightly depolarized to the GABAA reversal potential, an initially outward current in response to SL-M stimulation became a large and long-lasting inward current upon stepping to a hyperpolarized potential (Figure 4B). Therefore, the time course of decay of the synaptic currents GABAA,~,,~ and GABA,+l,, reflects accurately the time course of conductance underlying the two separate currents. It is possible that inadequate space clamp did influencesomeof the present results. In particular, it is likely that the slower onset of CABAA,,~,, than GABAA,tast is due at least in part to dendritic filtering, given its relatively fast rise time, even in the presence of furosemide (Figure 5), and its electrotonically distant location. There may be in addition a true difference in the onset kinetics of the two currents. Also, the time course of GABA,+,r may be faster than reported here, for its rise and decay times are in the appropriate range to be influenced by inadequate space clamp (Johnston and Brown, 1983). However, since the current is located near the cell body, it should be less susceptible to this effect. Anatomical Segregation of CABAA,fast and GABAA,,I,, Two findings indicate that GABAA,r,,r is spatially restricted to the cell body, whereas GABAA,,l,, occurs in the dendrites. First, there is a greater shift of the reversal potential with chloride loading for GABAA,raSt than GABAA,,r,, (Figures 2C and 2D; Table 1). Since an active transport process that maintains a low internal chloride ion concentration would be overcome more easily at the cell body, where the recording electrode penetrates the cell and where there is a low surface area/volume, this suggests a cell body localization of the fast component. Second, focal application of bicuculline to the cell body layer produced a selective and nearly complete blockade of GABAA,tast (Figures 3A and 3C). A less selective block of GABAA,,r,,, did result from dendritic application of bicuculline (Figures 3B and 3D), though it proved impossible to achieve a selective or large degree of block of GABAA,,I,, when bicuculline was applied to a single site in the apical dendritic layer. Either there is significant contribution to GABAA,,,,, from the cell body region and/or basal dendrites, or CABAA,s~,,,, occurs over such a large spatially distributed areawithin the apical dendrites that diffusion of bicuculline away from that site could not be avoided. The two IPSC components could arise from different presynaptic neurons or from different populations of terminals of a single type of interneuron. The

Two 197

GABAA Responses

in Rat Hippocampus

former appears more likely, as several inhibitory interneuron types have been described, having clear anatomical and physiological distinctions. In particular, basket cells or other interneurons located within orclosetoSPthatmakeinhibitorysynapsesontopyramidal cells (Knowles and Schwartzkroin, 1981; Somogyi et al., 1983; Miles, 1990) are likely candidates for the presynaptic source of the GABAA,t,,, current. Likewise, SL-M interneurons (Lacaille and Schwartzkroin, 1988b), with cell bodies located within SL-M and axons that project along the SL-M layer and descend into SR, into and along SP, and occasionally enter stratum oriens, are likely candidates for the source of the current.These interneurons havealso been GABAA,,I,, proposed as presynaptic elements of the CABAe response (Lacaille and Schwartzkroin, 1988a; Williams and Lacaille, 1992).

Furosemide Effects There may be several reasons that a distinction between GABAA,rast and GABA,+l,, and the ability of furosemide to block GABAA,rast selectively have not been noted previously. Only recently have techniques been described to elicit monosynaptic GABAergic IPSCs (Davies et al., 1990) and to block the GABAe component of the evoked IPSC (Nathan et al., 1990; Kerr et al., 1987, 1988; Olpe et al., 1990), so that GABA,+,t is not obscured by temporal overlap with the glutamate-mediated excitatory postsynaptic current and GABA,+,,, by overlap with the slow GABAe current. Because furosemide is known to block chloride transport (Misgeld et al., 1986; Thompson et al., 1988), the observation that its application results in multiple population spikes may havebeen ascribedtothisproperty,aswasthefinding that afterdischarges could be produced by furosemide in the CA3 region in organotypic culture (Thompson and Gahwiler, 1989). Indeed, it was reported that an outwardly directed chloride transport that is blocked byfurosemide maintains a low intracellular chloride concentration in hippocampal CA3 pyramidal cells (Misgeld et al., 1986;Thompson and Gahwiler, 1989) and neocortical neurons (Thompson et al., 1988), resulting in a hyperpolarizing IPSP. Accordingly, in the presence of furosemide the chloride reversal potential Ecr may become less negative, as was noted in some of the present experiments (e.g., Figure 5A). However, this effect on Ecr was small and often not present (e.g., Figure 5B; Figure 6A), probably because the concentration of furosemide used (0.6 mM) was at the lower limit of a detectable effect (0.5 mM [Misgeld et al., 19861) and 2 mM furosemide shifts Ecr byonly7mVinCA3cells(Misgeldetal.,1986;Thompson and Gahwiler, 1989). An uneven distribution of chloride and chloride transport systems has been suggested as the basis for differences in IPSP reversal potentials in different hippocampal neurons and for the depolarizing response to GABA application (Misgeld et al., 1986). Using a chloride-sensitive fluorescent dye to measure

chloride activity, an uneven distribution of chloride has recently been demonstrated in cultured hippocampal neurons (Hara et al., 1992). Furosemide (I mM) was found to decrease the resting concentration of chloride in the peripheral region of the perikaryon, but to have no effect on chloride concentration in neuronal processes (Hara et al., 1992). This regional difference in furosemide sensitivity, presumably imparted by an uneven distribution of an underlying Na+/K+/2CIcotransporter, parallels the selective block of GABAA,raSt at the cell body. Whether there is a common basis for this regional sensitivity to furosemide is not known, but it is clear that the major effect of furosemide in the present experiments is a reduction of the GABAA,rast conductance rather than a shift in reversal potential, as would result from an altered distribution of chloride (Figure 5). Furosemide was also reported to block the GABAgated chloridecurrent in frog spinal cord (Nicoll, 1978) and in isolated frog sensory neurons (Inomata et al., 1988), but not in hippocampal CA3 cells (Misgeld et al., 1986; Thompson and Gahwiler, 1989) or neocortex (Thompson et al., 1988). In the voltage-clamp study of furosemideeffectson CA3 pyramidal cells (Thompson and Gahwiler, 1989), as well as in the other currentclamp studies (Misgeld et al., 1986; Nicoll, 1978), IPSCs were contaminated by excitatory postsynaptic currents. It was found that excitatory postsynaptic current amplitude was greatly increased in the presence of furosemide (Thompson and Gahwiler, 1989); however, it is possible that this effect on the excitatory postsynaptic current, attributed to generalized disinhibition caused by a decrease in IPSP driving force throughout the neuronal population, was contributed to or caused by a blockade of a temporally overlapping and unrecognized GABAA,ra,tcurrent. Indeed,the residual GABAA current evident in the records in the presence of furosemide lasts up to 100 ms (Figure 7A in Thompson and Gahwiler, 1989), so it likely represents a GABA,+,, current. The mechanism by which furosemide alters GABAgated chloride conductance is unknown. In frog isolated sensory neurons (Inomata et al., 1988) it was found to reduce a GABA-activated chloride current in a noncompetitive manner with an I& of 0.8 mM and to facilitate the inactivation phase of the response, without altering the chloride reversal potential. Itwas suggested that it may do so by open-channel blockade. This drug and other “loop diuretics” have also been found to prevent N-methyl-o-aspartate channel activation, possiblyviadirect interaction with a modulatory polyamine-binding site (Lerma and Martin del Rio, 1992). The lCsO for this effect was 1.2 mM. It is possiblethatfurosemide blocks GABAA,rast by a similar mechanism, by binding to a modulatory site on the receptor/channel complex. Although a polyamine binding-site has not been demonstrated for the GABA channel, there are several modulatory sites on the GABA receptor/channel complex (Sivilotti and Nistri, 1991).

NWi-Oil 198

CABAA Kinetics Markedly different time courses have been observed between spontaneous IPSCs in CA1 neurons, which decay with a monoexponential time constant of approximately 8 ms (Collingridge et al., 1984; Mody et al., 1991), and evoked monosynaptic IPSCs, which are much slower to decay (approximately 50 ms by inspection of published data[Davies et al., 1990,199l; Nathan and Lambert, 19911, though the authors did not include an analysis of current decay). The present results suggest that different proportions of CABAA,cst may account for this unexplained and GABAA,,I,, difference. One might expect that two populations of spontaneous IPSCs having different decay rates could be detected, corresponding to CABAA,fast and CABA,+,,, currents. However, only a single population with a fast decay rate has been reported by others (Collingridge et al., 1984; Mody et al., 1991). Also, during the course of the present experiments, only fast spontaneous IPSCs were observed, although a formal analysis of spontaneous currents was not performed, and it is possible that a thorough search would reveal a second population of slow spontaneous IPSCs. The absence of slow spontaneous IPSCs may reflect a low level of spontaneous activity of presynaptic cells underlying the response. Alternatively, the current resulting from the discharge of a single presynaptic cell may be small for GABAA+~,,.,, so that a large number of synchronous synaptic inputs, as may be achieved by electrical stimulation, are required for its detection. IPSC decay kinetics with approximately the same time constants as GABA,+,r and CABAA,,~,,,,~~ the present report have been described for dentate gyrus granule cells (Edwards et al., 1990). In contrast to findings in CA1 pyramidal cells, these patch-clamp recordings revealed that decays were the same for spontaneous and evoked IPSCs. Single-channel recordings from these cells revealed evidence for two GABAA receptor subtypes, distinguishable by conductance levels, coexisting on the somata. It was suggested that the two channel types might mediate synaptic events at different synapses and hypothesized that their colocalization gives rise to the biexponential decay. However, the relationship between single-channel conductance and kinetics was not described, and it is not known whether the two time constants observed correspond to the two channel types. In a study of cultured tectal neurons (Kraszewski and Grantyn, 1992), it was found that evoked unitary IPSCs from synaptically connected pairs of neurons had decay time constants significantly longer than miniature IPSCs arising at the soma (34.4 f 15.6 ms versus 16.2 f 2.5 ms at 23°C-260C). The authors suggested that the difference was caused by distortion of IPSCs of distal origin, and the possibilitythat different postsynaptic mechanisms exist was not investigated. It seems likely that GABAA,f,,, and CABA,+,,, arise from activation of different GABAA receptor subtypes. It is clear that multiple GABAA subtypes exist in the brain, that they have different physiological and phar-

macological properties, and that different subtypes coexist in some brain regions (Sieghart, 1989; Olsen and Tobin, 1990; Verdoorn et al., 1990; Wisden et al., 1992). However, other possible explanations for the two components may be considered. The receptors underlying GABAA,rast and GABAQI,, may be identical in structure, but subject to different modulatory influences at the soma and in the periphery, for example, due to different internal calcium ion levels (Mody et al., 1991) or phosphorylation states (Chen et al., 1990). A second alternative to different receptor subtypes is that the slow decay component results from a longer channel open time of a doubly liganded receptor (Mathers and Wang, 1988; Macdonald et al., 1989; Busch and Sakmann, 1990). If this were the case, factors governing quanta1 content and transmitter reuptake and clearance from the synaptic cleft would need to differ at anatomically segregated sites. Also, this would imply that only singly liganded receptors are susceptible to block byfurosemide. These alternative explanations seem less likely, but may be tested. Extracellular Responses The finding that furosemide causes multiple evoked population spikes in response to stimulation of the Schaffer collateral pathway in SR supports the hypothesis that GABAA,rast acts to limit the response to excitatory input to a single action potential. The low concentration of furosemide employed makes it unlikely that its effect was due to a depolarizing shift of E. (see discussion above), but some contribution from this mechanism cannot be ruled out. Even in the presence of furosemide, a long-lasting inhibition of the SR response followed stimulation of the alveus (Figure 6Bb). To interpret this result it is necessary to know what was activated by stimulation of the alveus. At low stimulus intensities, resulting in a relatively selective antidromic activation of CA1 cells, feedback inhibitory circuitry would also have been activated. The absence of long-lasting inhibition at low stimulus intensities is consistent with the suggestion that feedback inhibition is directed primarily to the cell body (Nicoll et al., 1990). Only by increasing the stimulus to the point that a positive deflection preceded the population spike could a long-lasting inhibition of the SR response be demonstrated. This positive deflection, an extracellular excitatory postsynaptic potential, represents the passive source associated with excitatory synaptic input in the dendritic layers. Thus, at higher stimulus intensities additional neuronal elements were recruited. Axons projecting from CA3 to CA1 travel through SR (Schaffer collaterals), but also through stratum oriens (Ishizuka et al., 1990). These latter axons may be the source of the excitatory input that was recruited and of a feedforward inhibitory response that produced the longlasting paired-pulse depression. Direct activation of other pyramidal cells and inhibitory interneurons may also have occurred; these are less likely to have contributed tothe inhibition, as monosynaptic responses

Two 199

GABAA Responses

in Rat Hippocampus

are obtained recording

site

only within approximately (Davies et al., 1990).

Functional

Implications

500 Km of the

Based upon the segregation of inputs, different kinetic properties, and evidence from experiments with furosemide suggesting distinct functional roles for it would appear that two GABAA,t,,t and GABAA,,I,,, GABAA-mediated inhibitory systems coexist within the hippocampus. These results may have implications in the study of anesthetic mechanisms, epilepsy, learning and memory, and perhaps other aspects of cortical function, given the ubiquity of GABAergic inhibition in the brain. Volatile anesthetics prolong paired-pulse depression in viva from approximately 70 ms to 200 ms (Pearce et al., 1989), an effect that is clearly too long for GABAA,fast to explain, but which may be mediated by effects on GABAA,,~,,. The existence of convulsant anesthetics has been difficult to reconcile with the observations that GABAA inhibition is enhanced by both anesthetics and anticonvulsants (Sivilotti and Nistri, 1991; Collingridge et al., 1984; Nicoll, 1972; Proctor et al., 1986) and perhaps will be expained by different effects of these agents on GABAA,f,,, and The presence of multiple population GABAA,,I,,. spikes, a hallmark of epileptic tissue seen in the present experiments with furosemide, suggests that GABAA,t,,t plays an important role in this disease process. Theoretical and empirical evidence for a distinction between proximal and distal neuronal inhibition in the integrative function of neurons (Vu and Krasne, 1992) and the present evidence that GABAA,~,,~ and GABAA,,I,, are anatomically segregated suggest that they playfundamentallydifferent roles in information processing in the hippocampus and perhaps in other cortical structures. Experimental

Procedures

Hippocampal slices (400 pm thick) from adult male SpragueDawley rats (200-300 g) were cut on a Vibratome and maintained submerged at room temperature prior to transfer to a submersion style recording chamber at 36OC. They were continuously perfused at 3 cc/min with artificial cerebrospinal fluid containing the following: 127 mM NaCI, 1.9 mM KCI, 1.2 mM KH,PO+ 2.2 mM CaCl*, 1.4 mM MgS04, 26 mM NaHCO% and IO mM glucose and bubbled with 95% 02/5% CO*. Drugswere added to the perfusate by syringe pump to achieve the desired concentrations. For intracellular current-clamp and voltage-clamp experiments, standard glass recording microelectrodes (30-60 Ma) were filled with 3 M CsAc, 3 M CsCI, 3 M KAc, or 3 M KCI and, for most experiments, 50 mM QX-314 and coated near the tip with Sylgard 184 to reduce electrode capacitance. Recordings from CA1 cells were made using an Axoclamp 2A amplifier (switching frequency4-8 kHz), and data were acquired and analyzed using pClamp (Axon Instruments). Cells were held at -50 to -60 mV, stepped to a test potential 2 s before an IPSC was evoked, and held at the test potential for an additional 1100 ms. Stimuli (60-600 PA) were delivered every 10 s using bipolar tungsten microelectrodes (tip separation approximately 50 pm), and averaged responses to four series of voltage steps in increments of 10 or 20 mV were analyzed. Current decay phases were fit using an automated least squares method (Clampfit, Axon Instruments) to mono- and biexponential equations. Adequacy of fit was judged by eye, and the biexponential results were used

only if the fit was superior to monoexponential analysis (e.g., Figure IB). Bicuculline was applied to restricted regions of the slice in some experiments using pressure ejection (Picospritzer, 20 psi, IO-50 ms)of 1 mM bicucullinefrom a microelectrode with its tip broken to a diameter of approximately 3 pm. For extracellular experiments, slices were cut to 400-800 pm thick, and recordings were made at 32OC with electrodes filled with NaCl(2-4 M, tip broken to 5-10 Ma). Thicker slices allowed evoked responses to be obtained at agreater separation between stimulating and recording electrodes and resulted in enhanced paired-pulse depression. The extracellular recording electrode was placed in SP.by visual guidance, and stimulating electrodes were placed in the alveus and SR. For experiments in which paired-pulse depression was tested (Figure 4B), a stimulus was delivered to the alveus to evoke a mixed excitatory-inhibitory response, and this was followed at a variable interval by a stimulus to SR. The response to SR stimulation was compared with an identical stimulus in the absence of prior alveus stimulation. Acknowledgments The author thanks Lew Haberly, Larry Trussell, Meyer Jackson, and Phil Smith for helpful discussions and Sarah Bromley and Lee Faucher for technical assistance. This work was supported by National Institutes of Health grant NS01548. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received

June

4,1992;

revised

November

12, 1992.

References Alger, B. E., and Nicoll, R. A. (1982). tion in rat hippocampal pyramidal iol. (Land.) 328, 105-123. Bormann, J. (1988). Electrophysiology ceptor subtypes. Trends Neurosci.

Feed-forward cells studied of GABAa 71, 112-116.

dendritic in vitro.

inhibiJ. Phys-

and GABAb

Busch, C., and Sakmann, B. (1990). Synaptic transmission pocampal neurons: numerical reconstruction of quanta1 Cold Spring Harbor Symp. Quant. Biol. 55, 69-80.

re-

in hipIPSCs.

Chen, Q. X., Stelzer, A., Kay, A. R., and Wong, R. K. (1990). CABAA receptor function is regulated by phosphorylation in acutelydissociated guinea-pig hippocampal neurones. J. Physiol. (Land.) 420, 207-221. Collingridge, G. L., Gage, P. W., and Robertson, tory post-synaptic currents in rat hippocampal Physiol. (Land.) 356, 551-564.

B. (1984). InhibiCA1 neurones. J.

Davies, C. H., Davies, S. N., and Collingridge, G. L. (1990). Pairedpulse depression of monosynaptic CABA-mediated inhibitory postsynaptic responses in rat hippocampus. J. Physiol. (Lond.) 424, 513-531. Davies, C. H., Starkey, (1991). GABAautoreceptors 349, 609-611.

S. J., Pozza, M. F., and Collingridge, G. L. regulate the induction of LTP. Nature

Dunwiddie, T. V., Worth, T. S., and Olsen, R. W. (1986). Facilitation of recurrent inhibition iri rat hippocampus by barbiturate and related nonbarbiturate depressant drugs. J. Pharmacol. Exp. Ther. 238, 564-575. Dutar, P., and Nicoll, R. A. (1988). A physiological receptors in the central nervous system. Nature

role for GABAB 332, 156-158.

Edwards, F. A., Konnerth, A., and Sakmann, B. (1990). Quanta1 analysis of inhibitory synaptic transmission in the dentate gyrus of rat hippocampal slices: a patch-clamp study. J. Physiol. (Land.) 430, 213-249. Hara, M., Inoue, M., Yasukura, T., Ohnishi, S., Mikami, Y., and Inagaki, C. (1992). Unveven distribution of intracellular Cl- in rat hippocampal neurons. Neurosci. Lett. 143, 135-138. Inomata,N.,Ishihara,T.,andAkaike,N.(1988).Effectsofdiuretics

Neuron 200

on GABA-gated chloride rones. Br. J. Pharmacol,

current in frog 93, 679-683.

isolated

sensory

neu-

Ishizuka, N., Weber, J., and Amaral, D. G. (1990). Organization of intrahippocampal projectionsoriginatingfrom CA3 pyramidal cells in the rat. J. Comp. Neurol. 295, 580-623. Johnston, D., and Brown, T. H. (1983). Interpretation of voltageclamp measurements in hippocampal neurons. J. Neurophysiol. 50, 464-486. Kandel, E. R., Spencer, W.A., and Brinley, F. J. (1961). Electrophysiology of hippocampal neurons. I. Sequential invasion and synaptic organization. J. Neurophysiol. 24, 225-242. :, Kapur, J., Stringer, J. L., and Lothman, E. W. (1989). Evidence that repetitive seizures in the hippocampus cause a lasting reduction of GABAergic inhibition. J. Neurophysiol. 61, 417-426. Kerr, D. I., Ong, J., Prager, R. H., Cynther, B. D., and Curtis, D. R. (1987). Phaclofen: a peripheral and central baclofen antagonist. Brain Res. 405, 150-154. Kerr, D. I., Ong, J., Johnston, G. A., Abbenante, J., and Prager, R. H. (1988). 2-Hydroxy-saclofen: an improved antagonist at central and peripheral CABAR receptors. Neurosci. Lett. 92, 92-96. Knowles, synaptic 318-322.

W. D., and Schwartskroin, interactions in hippocampal

P. A. (1981). Local circuit brain slices. J. Neurosci. 7,

Kraszewski, K., and Crantyn; R. (1992). Unitary, quanta1 and miniature GABA-activated synaptic chloride currents in cultured neurons from the rat superior coiliculus. Neuroscience 47, 555-570. .&, Lacaille, J. C., and Schwartzkroin, P. A. (1988a). Stratum lacunosum-moleculare interneurons of hippocampal CA1 region. II. Intrasomaticand intradendritic recordingsof local circuit synaptic interactions. J. Neurosci. 8, 1411-1424. Lacaille, J. C., and Schwartzkroin, P. A. (1988b). Stratum lacunosum-moleculare interneurons of hippocampal CA1 region. I. Intracellular response characteristics, synaptic responses, and morphology. J. Neurosci. 8, 1400-1410. Lacaille, J. C., Mueller, A. L.,, Kunkel, D. D., and Schwartzkroin, P. A. (1987). Local circuit interactions between oriens/alveus interneurons and CA1 pyramidal cells in hippocampal slices: electrophysiology and morphology. J. Neurosci. 7, 1979-1993. Lerma, J., and Martin del Rio, R. (1992). Chloride ers prevent N-methyl-o-aspartate receptor-channel vation. Mol. Pharmacol. 47, 217-222.

transport blockcomplexacti-

Nicoll, R. A. (1972). The effects of anaesthetics on synaptic excitation and inhibition in the olfactory bulb. J. Physiol. (Land.) 223, 803-814. Nicoll, R. A. (1978). The blockade of GABA mediated responses in the frog spinal cord by ammonium ions and furosemide. J. Physiol. (Land.) 283, 121-132. Nicoll, R. A., and Alger, cording from submerged 153-156.

B. E. (1981). A simple chamber for brain slices. J. Neurosci. Meth.

re4,

Nicoll, R. A., Malenka, R. C., and Kauer, J. A. (1990). Functional comparison of neurotransmitter receptor subtypes in mammalian central nervous system. Physiol. Rev. 70, 513-565. Olpe, H. R., Karlsson, C., Pozza, M. F., Brugger, F., Steinmann, M., Van Riezen, H., Fagg, G., Hall, R. C., Froestl, W., and Bittiger, H. (1990). CGP 35348: a centrally active blocker of GABAB receptors. Eur. J. Pharmacol. 787, 27-38. Olsen, R. W., and Tobin, A. J. (1990). Molecular receptors. FASEB J. 4, 1469-1480. Pearce, volatile campus.

biology

of GABAA

R. A., Stringer, J. L., and Lothman, E. W. (1989). Effect of anesthetics on synaptic transmission in the rat hippoAnesthesiology 77, 591-598.

Proctor, W. R., Mynlieff, tory action of etomidate tion in rat hippocampal 3161-3168.

M., and Dunwiddie, T. V. (1986). Facilitaand pentobarbital on recurrent inhibipyramidal neurons. J. Neurosci. 6,

Rail, W., and Segev, I. (1985). Space-clamp problems when voltage clamping branched neurons with intracellular microelectrodes. In Voltage and Patch Clamping with Microelectrodes, T. G. Smith, H. Lecar, S. J. Redman, and P. W. Gage, eds. (Bethesda, Maryland: American Physiological Society), pp. 191-215. Rock, D. M., and Taylor, C. P. (1986). Effects of diazepam, pentobarbital, phenytoin and pentylenetetrazol on hippocampal paired-pulse inhibition in vivo. Neurosci. Lett. 65, 265-270. Sieghart, ceptors.

W. (1989). Multiplicity of GABAA-benzodiazepine Trends Pharmacol. Sci. 70,407-411.

Sivilotti, L., and Nistri, A. (1991). GABA receptor mechanisms the central nervous system. Prog. Neurobiol. 36, 35-92.

rein

Somogyi, P., Nunzi, M. G., Gorio, A., and Smith, A. D. (1983). A new type of specific interneuron in the monkey hippocampus forming synapses exclusively with the axon initial segments of pyramidal cells. Brain Res. 259, 137-142.

Macdonald, R. L., Rogers, C. J., and Twyman, R. E. (1989). Kinetic properties of the CABAA receptor main conductance state of mouse spinal cord neurones in culture. J. Physiol. (Lond.) 470, 479-499.

Thompson, S. M., and Glhwiler, B. H. (1989). Activity-dependent disinhibition. II. Effects of extracellular potassium, furosemide, and membrane potential on ECI- in hippocampal CA3 neurons. J. Neurophysiol. 67, 512-523.

Mathers, D. A., and Wang, Y. H. (1988). Effect of agonist concentration on the lifetime of GABA-activated membrane channels in spinal cord neurons. Synapse 2, 627-632.

Thompson, S. M., Deisz, R. A., and Prince, D. A. (1988). Relative contributions of passive equilibrium and active transport to the distribution of chloride in mammalian cortical neurons. J. Neurophysiol. 60, 105-124.

Miles, R. (1990). Synaptic excitation CA3 hippocampal pyramidal cells Physiol. (Land.) 428, 61-77.

of inhibitory cells by single of the guinea-pig in vitro. J.

Misgeld, U., Deisz, R. A., Dodt, H. U., and Lux, H. D. (1986). The role of chloride transport in postsynaptic inhibition of hippocampal neurons. Science 232, 1413-1415. Mody, I., Tanelian, D. L., and Maciver, M. B. (1991). Halothane enhances tonic neuronal inhibition by elevating intracellular calcium. Brain Res. 538, 319-323. Nathan, underlies campus.

T., and Lambert, J. D. (1991). Depression of the fast IPSP paired-pulse facilitation in area CA1 of the rat hippoJ. Neurophysiol. 66, 1704-1715.

Verdoorn, T. A., Draguhn, mann, B. (1990). Functional receptors depend upon 919-928.

A., Ymer, properties subunit

S., Seeburg, P. H., and Sakof recombinant rat GABAA composition. Neuron 4,

Vu, E. T., and Krasne, F. B. (1992). Evidence for a computational distinction between proximal and distal inhibition. Science 255, 1710-1712. Williams, S., and Lacaille, J. C. (1992). GABAK receptor-mediated inhibitory postsynaptic potentials evoked by electrical stimulation and by glutamate stimulation of interneurons in stratum lacunosum-moleculare in hippocampal CA1 pyramidal cells in vitro. Synapse 71, 249-258.

Nathan, T., Jensen, M. S., and Lambert, J. D. (1990). The slow inhibitory postsynaptic potential in rat hippocampal CA1 neurones is blocked by intracellular injection of QX-314. Neurosci. Lett. 770, 309-313.

Wisden, W., Laurie, D. J., Monyer, H., and Seeburg, P. H. (1992). The distribution of 13 GABA-A receptor subunit mRNAs in the rat brain. I. Telencephalon, diencephalon, mesencephalon. J. Neurosci. 72, 1040-1062.

Newberry, N. R., and Nicoll, R. A. (1984). A bicuculline-resistant inhibitory post-synaptic potential in rat hippocampal pyramidal cells in vitro. J. Physiol. (Land.) 348, 239-254.

Wolf, P., and Haas, H. L. (1977). Effects of diazepines and barbiturates on hippocampal recurrent inhibition. Naunyn Schmiedebergs Arch. Pharmacol. 299, 211-218.