HIPPOCAMPUS 6:294-305 (1996)

Physiological Properties of Anatomically Identified Basket and Bistratified Cells in the CAl Area of the Rat Hippocampus In Vitro

Eberhard H. BuhV Tibor Szilagyi,l,2 Katalin Halasy,1,3 and Peter Somogyi1

IMRC Anatomical Neuropharmacology Unit, Oxford University, Oxford, England; 2Department of Physiology, University of Medicine and Pharmacy, Tirgu Mures, Romania; 3Department of Zoology and Cell Biolog1j, f6zsef Attila University, Szeged, Hungary

ABSTRACT: Basket and bistratified cells form two anatomically distinct classes of GABAergic local-circuit neurons in the CA 1 region of the rat hippocampus. A physiological comparison was made of intracellularly recorded basket (n = 13) and bistratified neurons (n = 6), all of which had been anatomically defined by their efferent target profile (Halasy et al., 1996).

Basket cells had an average resting membrane potential of -64.2 ± 7.2 vs. -69.2 ± 4.6 mV in bistratified cells. The latter had considerably higher mean input resistances (60.2 ± 42.1 vs. 31.3 ± 10.9 MU) and longer membrane time constants (18.6 ± 8.1 vs. 9.8 ± 4.5 ms) than bas­ket cells. Differences were also apparent in the duration of action po­tentials, those of basket cells being 364 ± 77 and those of bistratified cells being 527 ± 138 p.s at half-amplitude. Action potentials were generally followed by prominent, fast afterhyperpolarizing potentia Is which in bas­ket cells were 13.5 ± 6.7 mV in amplitude vs. 10.5 ± 5.1 in bistratified cells. The differences in membrane time constant, resting membrane po­tential, and action potential duration reached statistical significance (P < 0.05).

Extracellular stimulation of Schaffer collateral/commissural afferents elicited short-latency ex citatory postsynaptic potentials (EPSPs) in both cell types. The average 10-90% rise time and duration (at half-ampli­tude) of subthreshold EPSPs in basket cells were 1.9 ± 0.5 and 10.7 ± 5.6 ms, compared to 3.3 ± 1.3 and 20.1 ± 9.7 ms in bistratified cells, the difference in EPSP rise times being statistically significant. Basket and bi­stratified EPSPs were highly sensitive to a bath applied antagonist of non-N-methyl-D-aspartate (NMDA) receptors, whereas the remaining slow-rise EPSP could be abolished by an NMDA receptor antagonist. Increasing stimulation intensity elicited biphasic inhibitory postsynaptic potentials (IPSPs) in both basket and bistratified cells.

In conclusion, basket and bistratified cells in the CA 1 area show promi­nent differences in several of their membrane and firing properties. Both cell classes are activated by Schaffer collateral/commissural axons in a feedforward manner and receive inhibitory input from other, as yet unidentified, local-circuit neurons. © 1996 Wiley-Liss, Inc.

KEY WORDS: GABA, interneuron, inhibition, postsynaptic, feed for-ward

Accepted for pub li cation May 1, 1996. Address correspondence and reprint requests to E.H. Buhl, MRC Anatomica l Neuropharmaco logy Un it, Oxford Un iversi ty, Ma nsfield Rd ., Oxford OX1 3TH, UK.

© 1996 WlLEY-LISS, [Ne.

INTRODUCTION

Hippocampal interneurons, or non-principal cells, share two common properties. First, they have a dense local axonal arbor which may either target principal cells, other local-ci tcuit neurons, or a mixture of both (Somogyi et aI., 1983.; Schw.rrLkroin and Kunkel , 1985; Gulyas et al. , 1993b; Halasy and Somogyi, 1993; H an et al., 1993; Buhl et al., 1994a,b). Second , it ap­pears that most interneurons investigated to date store the inhibitory neurotransmittet l'-aminoburytic acid (GABA) in their tetminals (Somogyi et al ., 1983b, 1984, 1985; Sotiano and Frotschet, 1989; Halasy and Somogyi, 1993; Halasy et al., 1996). Apart from these common characteristics, however, interneurons must be tegatded as a conglomerate of heterogeneous, albeit dis­tinct, cell classes. With respect to their anato mical prop­erties, non-ptincipal cells may be segregated into several categories, differing with respect ro their content of pep­tide neurotransmittets (Somogyi et al. , 1984; Kosab et al ., 1985; Sloviter and Nilavet, 1987) , calcium-binding proteins (Nitsch et al ., 1990; G ulyas et al ., 1991), and/or efferent target profile (Somogyi et al ., 1983a, 1985; Gulyas et al., 1993b; Halasy and Somogyi , 1993; Buh/ et al., 1994a,b; Sik et al., 1995). However, these features are not necessarily mutually exclusive; on the contrary, for example, neurons sllch as the dentate gyrus hilat per­forant pathway-associated (HIPP) cells (H an et al ., 1993) ptesumably correspond to the class of somato­statin-positive hilar neurons.

With respect to their physiological properties, how­ever, distinctions berween diffetent classes of GABAergic neuron are either blurred or are yet to be established. As yet, there is some degtee of general consensus that in­terneurons per se may differ from ptincipal cells in many of theit inttinsic properties, such as short-duration ac-

INTERNEURON PROPERTIES IN RAT HIPPOCAMPUS 295

Don potentials. the absence of spike frequency adaptation. a large­ampli tude fast afrerhyperpolarizing potential (fAHP). or prominent outward rectification (Schwarrzkroin and Mathers. 1978; Ashwood et al .. 1984; Schwarrzkroin and Kunkel, 1985; Misgeld and

Frotseher. 1986; Lacaille and Williams, 1990; Scharfman et al .. 1990; Scharfman 1992). Moreover, (some) interneurons have also been

hown to differ from principal cells with respect to their glutamate =eptor expression (Baude et al., 1993; McBain et al., 1994; Poncer et al ., 1995) and the kinetic properties of their glutamate receptors (uvsey et al. , 1993; McBain and Dingledine, 1993; Perouansky and

Yaari, 1993) as well as their synaptic currents (Livsey and Vicini,

1992). T here are, however. considerably fewer data available for the correlation of anatomical classes of interneurons with their physio­

logical properties. because of the requirement for a stringent mor­phological or immunoeyrochemical identification of recorded neu­rons. By usi ng such a muhidisciplinary approach, it was, for example. possible to establish that fast-spiking basket cells contain the calcium­

binding protein parvalbumin (Kawaguchi et al., 1987; Srk et al. ,

1995), whereas a bistratified cell has been shown to be calbindin im­munoreactive (Srk et al. , 1995). The physiological properties of in­terneurons have also been com pared with respect to their laminar

position (Kawaguchi and Hama, 1988; Lac.1.i.lle and Schwartzkroin . 1988; Lacaille and W illiams, 1990; Lacaille, 1991). Although these

studies provided compelling ~"Vidence for the physiological hetero­

geneiry of hippoc.1mpal non-principal cells. the relationship between me variabiliry in physiological properties and the patterns of synap­tic connections of interneurons remained elusive, largely because the hippocampal laminae are fur from homogeneous with regard to

meir complement of GABAergic interneurons (Lac.1.i.lIe and Sch\varrzkroin, 1988; Kawaguchi and H ama. 1988; G ulyas et al .• 1993b; Han et al., 1993; Buhl et al. . 1994.1; McBain et al ., 1994) .

With hippocampal layering being a relatively poor indicator of interneuronal diversiry, the co ncept of grouping non-principal cells with respect to their efferent con nectiviry appears to be a more suit­

able approach toward ca tegorizing GABAergic neurons (Somogyi

et al., 1983a.b; Gulyas et al. , 1993b; Halasy and Somogyi, 1993; Buhl er aI. , 1994a.b). Moreover, the advent of routinely combin­

mg in vitro intracellular recording techniques with correlated light and electron microscopy has now provided the opportuniry to de­

fine the physiological properties and synaptic effects of interneu­rons which have been characterized wi rh respect to their target se­

kcciviry (Buhl et al. , 1994a.b; Buh l et al. . 1995). T his study will provide such a physiological comparison between twO classes of GABAergic hippocampal interneurons in the pyramidal cell layer of the CAI area. The data presented in the accompanying article (Halasy et al. , 1996) demonstrate nOt only the GABAergic nature of basket and bistratified cells. but also distinct differences in the pattern of their afferent and efferent con nections.

MATERIALS AND METHODS

lice Preparation

Young adult female Wistar rats (> 150 g) were deeply anes­metized with intramuscularly inj ected ketamine (lOO mg/kg) and

xyl3""l ine (1O mg/kg). Then the animals were perfused with ~30 ml of chilled artificial cerebrospinal fluid (ACS F). and their brains were removed and immersed in chilled ACSF. Using a Vibrosli ce

(Cam pden Instruments, Loughborough. UK) . 400-/Lm-thick

slices were cut in the horizontal plane. The sli ces were transferred co a recording chamber and maintained at 34-35°C at the inter­

face between oxygenated ACSF and a humidified atmosphere sat­urated with 95% 0 2/5% C O 2. The flow rate was adjusted co 1.5 rnI /min. and the slices were allowed co recover for > 1 h . During the initial part of the experiment (perfusion, cutting, 30-45 min incubation). the ACSF waS composed of (in mM) 256 sucrose,

3 .0 KCI. ] .25 NH2P04. 24 NaHC04, 2.0 MgS04• 2.0 CaClz and 10 gl ucose. Subsequently, the sucrose in the ACSF was replaced

by equiosmolar NaCI (I26 mM). All drugs were kept as concen­trated scocks. which were diluted in ACSF and then bath applied. T he excitatory amino acid blockers 6-cyano-7-nitroquinoxal inc-2,3-dione (CNQX) and DL-2-am ino-5-phosphonopentanoic acid

(AP5) were obtained from Tocris Cookson (Bristol , UK) and bicu­

culline hydrochloride was purchased from Sigma, (Poole, UK).

Intracellular Recordings and Data Analysis

Micropipettes were pulled from standard wal l borosilicate cap­

illaries, filled with 2% bioeyrin in 1. 5 M KCH3S04• and usually beveled co a D C resistance of 80-150 MO. Neurons were im­

paled in or close co the cell body layer of the hippocampal CA 1 area. Putative interneurons were identified by their distinct phys­iological properties (see below). Recordings were made in bridge mode using Axoclamp or modified Axoprobe (both Axon

Instruments, Foster Ciry. CA) intracellular amplifiers. Postsynap­tic potentials were elicited with a fine-tipped bipolar tungsten stimulation electrode with a tip separation of ~50 /Lm. Data were

acquired using PCM instrumentatio n recorders and stored on videotapes. Experimental data were redigitized at 10-20 kH1. with a 12 Bit NO board (RC Electronics Computerscope, Santa

Barbara, CA) and analyzed with commercially available sofrware

(Axograph. Axon Instrum ents. Foster C ity. CA) . All data are ex­

pressed as mean :!: SO. Statistical analysis was performed using a non-parametric statistic.-u test (Mann-Whirney U-test).

Resting membrane potentials were taken as the difference be­tween surFace DC potential following electrode withd rawal and the steady-state membrane potential without the injection of bias

current. M embrane time constants were determined by fitting sin­

gle exponentials co the decay of small-anlplitude «5 m V) hy­perpolarizing current pulses. Simi larly, input resistances were ca lculated from the maximum voltage deflection of small hyper­polarizing current pulses. Using rheobasic current pulses. spike amplitudes were taken from the baseline to the peak of the ac­tion potentials, whi le spike duration was measured at half-anlpli­tude. The ampl itudes of fAH Ps were measured from the shoul­der of the preceding action potentials. Whenever late AHPs were

seen to fo ll ow trains of action potentials, single exponential func­tions were fitted to determine their decay characteristics. Variations in the rate of firing were calculated From rhe respec­

t ive interspike intervals and numericall y expressed as percent changes of the initial discharge rate. The rise time o f excitarory

296 RUHL ET A L.

postsynaptic potentials (EPSPs) was determi ned as the interval between 10 and 90% of their respective peak ampli tudes, and EPSP duration was taken at half-amplitude, excluding events that were apparencly curtailed by inhibitory postsynaptic potentials (IPSPs). T he time-to-peak of IPSPs was determined at depolar­ized membrane potentials and extrapolated from the preceding stimulus artifac t.

RESULTS

All cells included in the analys is had resting membrane po­tentials of min imally - 55 m V without requiring hyperpolarizing bias current. Moreover, the cells remained stable for at least 30 min reco rding time and , upon depolarization, could fire sustained trains of high-frequency action potentials. Furthermore, only those cells were included for analysis which could be recovered for light microscopic analysis, followed by the post hoc c1ecrron microscopic scrutiny of their synaptic target profi le (for full anatomical description, see accompan ying article). In this regard, it has been previously established that, with respect to their ef­ferent output, CA l pyramidal cell layer interneurons predomi­nancly faU into three distinct categories: axo-axonic, basket, and bistrat ifled cells (Buhl et al ., 1994a). Below, we summarize and compare the physiological properties of two types of stringently identified in terneuro ns, basket cells (n = 13) and bistratifled cel ls (n = 6).

Membrane and Firing Properties

In comparison, basket cells (n = 13) had significantly more de­polarized resting membrane potenti als (- 64.2 ::':: 7.2 m V) than bistratifled cells (n = 6; - 69.2 ::'::4 .6 mY; P < 0.05). Likewise, basket cells had significantly fas ter membrane time constants (9 .9::':: 4.6 ms) when compared to bistratifled cells (18.6 ::':: 8 .1 ms). However, although basket cells had, on average, consider­ably lower input impedances (3 1.3 ::':: 10.9 vs. 60.2 ::':: 42 Mo' in bistratified cells), this difference was not staristicall y signifi cant (P = 0.056) . O nly basket cells at membrane potentials more de-

A, B: Basket cell

A B

~

polarized than approximately - 60 m V tended to fi re spontaneous action potentials, although it should be emphasized that the sam­pling of neurones with extensive labeling and therefore prolonged recording periods may have biased these data toward the inclu­sion of cells with relatively hyperpolarized membrane potentials and, accordingly, little spontaneous activity.

Both types of interneuron had non-overshooting action po­tentials, the means for basket and bimatifled cells being 63.6 ( ::':: 12.7) mV and 69.8 ::'::5.0) mV (P > 0.05), respectively. Action potentials were generally brief due to a rapid rate of fall (Fig. 1) and were found to be significantly faster in basket cells (0.4 1 ::':: 0.09 vs. 0.53 ::':: 0.14 ms in bistratifled cells).

In response to depolarizing current pulses, both groups of ncu­ron could display a variery of different firing patterns. O nly a mi­no riry of cells fi red non-accommodating trains of action poten­tials (Fig. 2A, D ; n = I each; adaptation rate < 10%; note: the widely used terms spike frequency acco mmodation and adapta­tion are here regarded as synonymous) , although this firing pat­tern has commonly been associated with GABAergic interneurons (Connors and G utnick, 1990; Scharfman, 1992). Although the majori ty of cells assumed a tegular firing pattern, they revealed vary­ing degrees of spike frequency adaptation, the respective changes in the rate of firing ranging from 14 to 88% (Fig. 2B; mean for all cells 42 ::':: 31 %; n = 12) in basket cells and 20-88% (Fig. 2E; mean for all cells 54 ::':: 34%; n = 6) in bistratified cells. Finally, three basket (Fig. 2C) and twO bimatified cells (Fig. 2F) showed a burst-like fi ring pattern in response to a depolarizing current pulse. Interestingly, different firing patterns may not be mutually exclu­sive, as they could be observed within individual cells. Accordingly, it was possible to switch cells from bursting into regular firing mode by means of a constant depolarizing current injection (Fig. 2E,F), although this property was not systematically investigated.

W hen the firing frequency of basket (n = 4) and bistratifled (n = 3) neurons was determined from the fi rst in terspike in terval, all cells revealed a linear current-frequency relationship (Fig. 3A,D). Differences, however, were apparent with respect to the slope of the individual curves; i.e., across cells, the same increment in current intensity could affect the respective fi ring rates to a vari­able degree. When current- frequency plots were determined from

C, 0 : Bistratlfied cell

c D

A, C: 1 ms S, D: 30ms

FIGURE 1. Action potential. in both basket (A) and bistratified cells (C) are characterized by their short duration which, when mea­sured at half-amplitude, is generally less than 600 /LS. In both types of interneurons action potentials have a rapid rate of fall , which is

in the same range as their rate of rise. Action potentials of basket (8) and bisteatified (D) ceUs are foUowed by a short-latency fast af­terhyperpolacization.

I

INTERNEURON PROPERTIES IN RAT HIPPOCAMPUS 297

A-C Basket cells

A

0.4 nA

B

0.5nA

c

0.5 nA

FIGURE 2. Intrinsic firing patterns of three basket (A-C) and two bistratificd cells (D-F). Within the same group of cells the fir­ing pattern and the degree of spike frequency adaptation could vary significandy. Both basket (A) and bistratified cells (D) could dis­charge with a train of non-accommodating action potentials in re­sponse to a depolarizing current injection, thus corresponding to the firing pattern which is generally assumed to be associated with fast­spiking interneurons in cortical areas. Interestingly, a large propor­tion of both basket (B) and bistratified cells (E) showed a marked

me first and last interspike interval of 200 ms depolarizing cur­rent pulses, differences were apparent in the respective slopes of cells which exhib ited marked spike frequency adaptation (Fig. 3B,E; solid symbols). Towards the end of a current pulse, these cells showed a linear, but disproportionally smaller increase in their firi ng rates in response to regular increases in current intensity (F ig. 3B,E; solid symbols connected by broken lines).

The regularity of neuronal discharge was assessed by plotting me respective fi ring frequencies, again taken as me recip rocal val ue of successive interspike intervals, during the injection of a constant depolarizing current pulse (Fig. 3C,F). It was clear that cells with no apparent accom modation of meir firing rate maintained re­markably constant interspike intervals or discharge rates, respec­ri,'e1y (Fig. 3C,F; open symbols). Both basket and bimatified cells which revealed marked reductions of their firing rate had relatively uniform patterns of adaptation. Thus, a sharp drop of firing rate occurred during the first 50-70 ms of the pulse, and then cells re­sumed discharging at a rela tively constant frequency during the re-

D·E Bistratified cells

D

0.5 nA

E

0.1 nA

F

0.1 nA

degree of spike frequency adaptation. Moreover some basket (C) and bi-stratified cells (F) were also capable of firing a burst of action po­tentials, which were riding on a depolarizing envelope. Note, how­ever, that in one instance d,e firing mode of a bistratified cell could be changed from bursting (F) to a regular firing/adapting behavior (E) hy depolarizing the cell from - 62 mY (F) to - 57 mY (E) . The cells in A and C are also illustrated in Figures 4 and 3 of the ac­companying article.

mainder of the pu lse. Occasional ly the regular firing pattern of basket (n = I) and bimatifled cells (n = \) could be d isru pted by sp ike doublets or triplets (Figs. 3C, 4A,O), here defined as an ac­tion potential following anomer with very short latency.

Afterpotentials

W ithout exception, action potentia Is in basket and bistrat ified cells were followed by a short-latency fAHP (F igs. \ , 2, 4A,O). Although, on average, basket ceUs had more pronounced fAHPs than bistratified cells (13.5 :t 6.7 [n = 12] vs. 10.5:t 5. 1 mV [n = 6]) , this difference did not reach statistical signifi cance. Both basket and bistratified cells could reveal prom inenr depolarizing afterpotentials (OAPs; Fig. 4B,E; arrows). T hese were most promi nent during the period of strongest spike frequency adap­tation and could be of sufficient amplitude to trigger spike dou­blets and triplets (for more derail, see Buhl et al. , 1994b) . Only a fraction of basket cells (5113; 39%) displayed OAPs, thus dif-

298 BUHL ET AL.

A 'so

B 2SO

~ >-u c ~ , I '" c §

so

C :lOO

2SO

i;' c ~ , :;r ,. f

01

A~C : Basket cells

current increment [nAI

. 0

• e - · - ...

0.2 0.3 0.4 0.5 06 0 7

so

current Increment [nAI

100

time (ms] ISO 200

FIGURE 3. Firing characteristics of basket (A- C; open and solid circles) and bistratified cells (D-F; open and solid squares). The in­stantaneous firing frequency (expressed as the inverse of the first in­terspike interval [ISI]) of both basket (A) and bistratihed cells (D) increased linearly with the amount of injected depolarizing current (in A,B,D,E expressed as current increment above threshold inten­sity) . The respective slopes, i.e., the rate of frequency increase per current step, of individual cells showed great variability. In basket and bist.ratified cells which exhibited spike frequency adaptation (B,E: solid symbols), the respective cnerent/frequency slopes differed markedly when taken from the first and last ISI (first ISIs connected by continuous line, last ISTs linked by broken line) of a 200-ms de­polarizing current pulse. In contrast, the slopes of first and last in­terspike intervals in non-adapting or weakly adapting ceUs (B: open circles) remained relatively constant. T he regularity and pattern of firing in basket (C) and bistratihed ceUs (F) was assessed by plot-

D 300 D-F: Bistratified cells

02 0.4 O.B 0.8

current increment [nl\)

E 200

~ >-u ~ ,

100

I '" c ·c ~

SO •.. -.. • • -tI _ .... .

. ... --0.2 0.'1 0.6 0.8 1.2 1.4

current increment (nAI

F 200

1902922

o,~~--~--~--~--~,,~~~ o SO 100

time [ms]

ISO 200

ri ng the inverse of successive intcrspike intervals (as a measure of firing frequency) against time. It is apparent that non-adapting cells (open symbols; two sweeps in each celJ at sanle intensity) assume a regular firing pattern, whereas adapting cells (solid symbols; in C four sweeps at variable intensities, in F two sweeps with identical Clurent intensity) decelerate markedly during the first 50- to 70-ms interva1 of the current puJse and then resume firing at near-steady­state levels. Intersweep variability was relatively minor. Only infre­quently occnering doublets (C) preseoted a minor disruption in the overall regularity of firing patterns. When comparing two sweeps which contained doublets with two successive sweeps which were without doublets, it is, however, obvious that despite the dramatic increase in firing frequency (i.e., a very short inlerspike interval) , the overall firing pattern remains remarkably unaffected. Cell codes correspond to Table I of the accompanying article.

INTERNEURON PROPERTIES IN RAT HIPPOCAMPUS 299

A·C: Basket cells

A

1 B

~w c

1.0 nA

-=J A: 40ms D:1OOms

\~

~ 20 ms

~ 200 ms

FIGURE 4. Both basket (A-C) and bisttatificd cells (D-F) could sbow complex afterpotentials resembling those of hippocampal prin­cipal cells. In these instances a fast afterhyperpolarizing potential (see also Fig. IB,D) was followed by a depolarizing afterpotential (OAP; B,E arrows). Occasionally DAPs were of sufficient amplitude lO trigger spike doublets (D,E) or even triplets (A,B). Prominent DAPs were frequently observed in conjunction with marked fre-

feri ng from bistratified cells where OAPs appear to be a more reg­ular feature (5/6; 83%). Similarly, following occasionally single (Fig. I D), but more frequently bursts of action potentials, both basket and bisrratified cells could reveal prominent late afterhy­perpolarizing potentials (IAHPs; Fig. 4C,F), although this feature was considerably less prominent in basket cells (n = 2/13; 15%) when compared to bisrratified cells (5/6; 83%). Generally, the amplitude of iAHPs was positively correlated with the burst du­ratio n and number of action potentials. A1IIAHPs decayed slowly (> 1 s) back to baseline, and their decay could be well fitted with a single exponential function. With respect to their respective mean time constants of decay, basket (1.24 :!: 0.20 s) and bi­stratified cells (0.96 :!: 0.97) appeared to be relatively similar. Final ly, it is noteworthy that, with the exception of one bistrati­fied cell, the occurrence of IAHPs predominantly coincided with prominent DAPs (617 cells).

D·F: Bistratified cells

o

1.0 nA

E

F

0.7 nA

quency adaptation (A,D). Moreover, in these cells, trains of action potentials were often followed by a long-lasting afterhyperpolariza­tion (C,F; clipping of depolarizing response indicated by stippled line; the cell in F is different from that shown in D and E). Note the clipping of action potentials in B and E. Broken lines in A in­dicate off-line adjustment of bridge balance. The cell in D and E is also iIIusttated in Figure 9 of the accompanying article.

Postsynaptic Responses

Using low stimulation intensities, which were invariably sub­threshold for concomitantly recorded pyramidal cells (data not shown), basket as well as bisrratified cells could be orthodromi­cally activated at very short latencies «4 ms), thus minimizing the likelihood of polysynaptic pathways contributing to the early part of EPSPs. Although synaptic responses could be elicited from several stimulation sites, such as stratum lacunosum-moleculare, the data reported here resulted from placing the stimulation elec­trode into stratum radiatum, usually at the CA1/CA3 border re­gion, thus presumably stimulating predominantly Schaffer col­lateral/commissural input fibers .

Afferent stimulation at the threshold for synaptic responses (Fig. SA,G) invariably elicited EPSPs in basket as well as bistratified cells. The analysis of single sweeps revealed that small-amplitude

300 BUHL ET AL.

EPSPs in both types of interneurons frequently had a fragmented,

mul t i-peaked appearance (Fig. 5G,H ), the latter feature generally smoothing out in averaged traces (Fig. 5A- E). Subthreshold EPSPs

in basket cel ls (n = 9) had an average rise-ti me of 1.9 ± 0.5 ms, whereas that of bistratified cells (n = 4) was determined to be 3.3 ± 1.3 ms. T his d ifference was statistically significant (P < 0.05) . Likewise, the mea n duration of EPSPs in basket cells

A-F: Basket cell

4 Volts ~ , ----------~------------------

B,~ __ --J ~ 7 Volts

C, ~ __ --~~ 12 Volts

D

36 Volts

E

A-D superimposed

F

11 Volts

FIGURE 5. Synaptic responses in basket (A-F) and bistratifled ceUs (G-L) foUowing extraceUnlar stimnlation of the Schaffer col­lateral/commissural pathway (arrowheads dellote stimnlus artifacts). Low stimulus intensities (A,G) usually resulted in excitatory postsy­naptic potentials (EPSPs) without apparent inhibitory postsynaptic potentials (IPSPs) _ Increasing the stimnlation intensity frequendy recruited short-latency IPSPs (G-£, I-K; asterisks), which curtailed the duration of the EPSPs_ Late IPSPs (D,1; solid squares) were only

(10.7 ± 5.7 ms) was considerably shorter than that of bistratifiecl

cells (20.1 ± 9.7 ms), although this difference did nOt reach sta­tisti cal significance (P = 0.09). When stimulating Schaffer collat­eral /comm issural afferents with maximal intensity, al l cells reached fi ring threshold . Suprathreshold EPSPs general ly elicited one (Fig. 5E) and occasionally two action potentials.

W ithout exception, stimulation intensities had to be adjusted

G G-L: Bistratified cell

threshold

H ---.JI'\. low intenSity -, "---____ ---------.J---.r----'"

medium intensity

~.....:.~ ________ hi9_h in_ten_Sity

K

G-J : superimposed

L JlOmv 20 ms

-67 mV

apparent at higher stimnlation intensities. Traces A-F and L repre­sent averages of four to 30 trials, G-K represent single sweeps. Note the characteristic "fragmented" appearance of EPSPs (G,H) . In all tested basket (F) and bistratifled cells (L) the EPSP amplitude in­creased with membrane hyperpolarization. Fast IPSPs (asterisks) re­versed in the range of the presumed chloride equilibrium potential. The cell in A-F is also illustrated in Figure 5 of the accompanying article.

TNTERNEURON PROPERTIES IN RAT HIPPOCAMPUS 301

A A, B: Basket cell c C, 0 : Bistratified cell

Slim. Schalfer coIL

B D

~ 20 m,

0.9 nA

control

FIGURE 6. Inhibitory responses in inhibitory cells. Schaffer col­Iatuallcommissural pathway- evoked IPSPs suppress repetitive fir­ing in CAI basket (A,B) and bistratified cells (C,D) . Strong stimuli rttntited short-latency IPSPs (asterisks) in both basket (A) and bi­str.l1ified (C) cells. The resulting hyperpolarization was effective i11 suppressing current-induced firing for periods greater than 50 ms

considerably higher rhan response threshold (0 evoke fast IPSPs (Fig. 56-0, H- J). Increasing rhe srimularion strength resulted in an amplirude increase of rhe fasr IPSP and eventually, ar high in­u:nsiries, the appearance of a lare IPSP (Fig. 50-E, J-K). The

-69mV

and frequently reducing the rate of firing during the remainder of the pulse (data not shown). The control traces (B,D) ilhL"rate that in the absence of the synaptic stimulus, the cells fired throughout the current pulse. Note that some reduction of the firing rate of the cell shown in C aod 0 is due to spike frequency adaptation. Arrowheads denote stimulation artifacts.

peak latency of the fast JPSP in basker cells was 2 1.4 :t 4.2 ms (n = 6) , which did nor d iffer significantly from thar of bisrrari­fied cells (25.7 :t 7.8 ms; n = 4). Likewise, rhe response reversal of the fasr IPSPs in basker and bisrrarified cells was very similar

A-D: Basket cell 5mv I 20 ms

E-H: Bistratified cell

A control

B -----1-----------------------------------

16V

c 10jiM CNOX

~ 1

48V

D r-.

1 48V

FIGURE 7. Excitatory amino acid receptor pharmacology of Schaffer collateral/commissural pathway-evoked EPSPs in a CAl bask", (A-D) and a bistratified cell (E-H) . At lower stimulation in­tcnsities, bath application of the non-NMDA receptor antagonist CNQX (1 0 I'M) resulted in a dramatic reduction of Schaffer col­Ia.uallcommissural EPSPs (A,B,E,F). In the presence of CNQX, higher stimulation intensities uncovered a CNQX-resistant EPSP with slower rise and decay kinetics (C,G). This CNQX-insensitive

E - --I

F

G

7V

1 7V

control

10jiM CNOX

-----j~------------------------

H

1 20V

~--------------------------------20V

EPSP could be largely blocked by bath application of the NMDA receptor antagonist AP5 (D,H; 30 f'M)- Note that all traces in A-H were obtained in a low concentration of bicuculJine (l I'M) to re­duce polysynaptic IPSPs_ Membrane potential in A -71 mY, in B -64 mY, in C and 0 -67 mY, in E and F -88 mY, and in G and H -77 mY. Arrowheads denote stimulation artifacts. The cell in A-D is also illustrated in Figure 3 of the accompanying article.

302 BUHL ET AL.

A c

control

·62 mV

/ 2 ~M bicuculline

" control

FIGURE 8. Short-latency IPSPs in basket cells are mediated by GABAA receptors. In this basket cell the Schaffer collateral/com­missural compound postsynaptic potential comprised a short-latency EPSP, which was followed by a fast IPSP (asterisks) and a late IPSP (solid square). The fast IPSP reversed around -66 mV (A), sug­gesting that chloride is the major charge carrier. (B) Addition of ini­tially 1 I'M (middle trace) , followed by 2 I'M of the GABAA recep-

(Fig. 5F,L; - 69.0 ::':: 2.S vs. -72.0 ::':: 2.0 mY). T he peak latency of the late component of the compound rpsp was 135 ::':: 29 ms in basket cells and 137 ::':: 17 ms in bistratified cells. T he reversal of the late rpsp appeared to be at rather hyperpolarized mem­brane potentials (approx. - 90 m y), although this parameter was difficul t to determine unless its amplitude was enhanced fo llow­ing the pharmacological blockade of GABAA receptors (Fig. SC).

The efficacy of rps ps to inhibit firing of inhibitory neurons wa., tested by delivering a shock to the Schaffer/collatera l com­missural pathway during depolarization-induced repetitive firi ng in both basket and bistratified cells (F ig. 6A,C). Initially all cells responded with one to twO short-latency action potentials, fol­lowed by a hyperpolarizing IPSP, wh ich was effective in sup­press ing fir ing for periods exceeding 100 ms. Control responses at the same current intensity (Fig. 6B,0) were also a.nalyzed to ascertain that the quiescent period was not due to a prolonged AHP followi ng a burst-li ke discharge.

Amino Acid Receptor Pharmacology

Scha.ffer collateral/commissural pathway-evoked EpSPs in both cell classes invariably showed a conventional voltage depen­dence (Fig. 5F,L), with their amplitudes increasing at more hy­perpolarized membrane potentials. Such EpSPs were pharmaco­logically explored in th ree basket and three bistratifi ed cells using subthreshold stimulatio n intensities (Fig. 7 A and E). After mon­itoring the control responses for > I 0 min, the slices were super­fused with 10 JLM of the non-N-methyl-D-aspartate (NMDA) receptor a.ntagonist CNQX, which could either substantially re-

2 ~M bicuculline ~10mv

100 ms

tor antagonist bicuculline methochloride to the bath, resulted in a reduction of the early IPSP and a concomitant increase of the late IPSP amplitude. Subsequently, the voltage dependency of the pre­dominating late LPSP (C; solid square) was explored. Whereas the reversal of the small remaining early IPSP remained unchanged (C; asterisk) , the late IPSP reversed around -99 mY, indicating the in­volvement of GABAs receptors.

duce or completely elimi nate the evoked EPSp (Fig. 7 B,F). In the presence of CNQX, a substantial increase of the stimulation in­tensity in conjunction with membrane depolarization could serve to uncover a residual EPSp with slower kinetics (Fig. 7C,G) . T his residual component proved ro be sens itive to bath application of the NMDA receptor a.ntagonist M 5 (30 JLM ; Fig. 70 ,H).

Both the time course a.nd reversal potential of the early phase of Scha.ffer collateral /commissural pathway-evoked rpsps suggest that at least part of the compound postsynaptic response is medi­ated by GABAA reccprors (Fig. SA). This notion, however, could only be adequately verified in basket cells (n = 2) . In the illus­trated example, bath application of initially I JLM of the GABAA

receptor a.ntagonist bicuculline resulted in a substantial reduction of the fast IPSP, whereas 2 JLM diminished mOSt of the early hy­perpolarization (Fig. SB). Interestingly, the late IPSP nOt only re­mained, but also grew in amplitude concomitant with increasing GABAA receptor blockade. Further hyperpolarization revealed that the bicucuIli ne-resistant IpSP reversed around - 99 mV (Fig. 8C).

DISCUSSION

Comparative Physiology of Hippocampal and Cortical Interneurons

T here is general agreement that, in principle, it is possible to physiologically d istinguish a substantial fraction of hi ppocampal interneurons from principal cells by virtue of their distinct mem­brane and firing properties. Differences from principal cells were

INTERNEURON PROPERTIES IN RAT HiPPOCAMPUS 303

reported with respect to membrane time constants, membrane rectification. action potential duration, spike frequency adapta­tion , as well as afterpotentials (Ashwood et al .. 1984; Schwartzkroin and Kunkcl , 1985; Misge1d and Frotscher, 1986; Lacaille and Williams. 1990; Scharfman et a1., 1990; Buhl et al., 1994a,b; 1995). Moreover. it was also noted that interneurons as st/ch may be physiologically heterogeneous (Kawaguchi and Hama. 1988; Lacaille and Schwarrzkroin, 1988). However, since the sample in these studies presumably originated from an anatomically heterogeneous pool of local-circuit neurons. it has remained unresolved whether the noted differences were due to variabiliry within or across classes of interneurons. Indeed, recent observations on hippocampal axo-axonic cells, defined by their efferent connectiviry, indicated a surprising variery of firing pat­terns and afterpotentials in response to depolarizing current pulses (Buhl et al. , 1994b) . T he present study corroborated this finding by revealing a similar degree of inter-individual heterogeneiry within the basket and bistratifled cell classes. The majoriry of neu­rons do not correspond to the rather firm ly established concept of most interneurons being non-adapting, wide-band transfor­mators of incoming excitation, generally lacking complex after­potentials (Connors and Gutnick, 1990; Hamill et al. . 1991; Scharfman, 1992). Only a fraction of basket and bimatifled cells in our sample displayed such properties. With regard to the re­mainder, the app31'ent discrepancies to previous studies may be resolved, since, in the absence of detailed morphological verifica­tion, the sanlpling of putative local-circuit neurons in earlier stud­ies could have been heavil y biased toward those displaying prop­erties traditionall y associated with interneurons.

Despite the aforementioned variabili ry of physiological prop­erties within given classes of hippocampal interneurons, a statis­tica l comparison berween basket and bistratified cells revealed a number of differences, among which spike duration and mem­brane time constant reached the level of statistical signifi cance. In this respect, very similar parameters were recently found to seg­regate rwo populations of rat neocortical layer V interneurons, re­ferred to as "fast-spiking" and "low-threshold spike (LTS)" cells, showing prominent differences in their membr311e time constants, action potential duration , and input resist311ce (Kawaguchi, 1993; Deuchars and T homson, 1995). In a similar way to cortical "fast­spiking" cell s, their hippocampal equivalents have been also shown to contain the calcium-binding protein parvalbumin (Kawaguchi et .1.1., 1987; Kawaguchi and Kubota, 1993). the lat­ter being a marker for hippocampal basket 311d axo-axonic cells (Kosaka et al., 1987) . Regarding their membr311e and firing prop­erties. the group of bistratified cells may correspond to cortical "L TS cells. " This notion is further supported by the observation that, like the bisrratified cell illustrated in Figure 2E 311d F, L TS cells consistently fire low-threshold spikes at hyperpolarized mem­br311e potentials (Kawagucl1i, 1993) and contain the calcium­binding protein calbindinD28k (Kawaguchi and Kubota. 1993), the latter being a marker for cortical interneurons with an effer­ent target profile enriched in dendritic shafts 311d spines (DeFelipe et al. . 1989). Finally, further similarities berween co rtical LTS cells and hippocampal bisrratifled cells were noted by S,k et al. (1995), who demonstrated calbindin D28k immunoreacriviry in a bistratifled cell labeled in vivo. However. the degree of homology

berween these rwo classes of local-circuit neurons requires further assessment, owing to the recent discovery of other cortical in­terneurons. such as regular spiking non-pyramidal cells. with an overlapping spectrum of physiological properties (Kawaguchi, 1995).

Schaffer Collateral/Commissural Activation of Basket and Bistratified Cells

Previous studies predicted that certain hippocampal interneu­rons, such as the dentate MaPI' cell (Han et al ., 1993), are pre­dominantly activated in a feedforward manner, because of the complete spatial segregation of their dendrites from the recurrent axon collaterals of principal cells. In contrast, another distinct rype of hippocampal interneuron , irnmunopositive for mGluR 1 Cl' as well as somatostatin and projecting to the stratum lacunosum­moleculare in rhe CA1 area (Baude et al., 1993; McBain et al., 1994). has been shown to receive mainly recurrent pyramidal cell input (Blasco-Ibanez and Freund. 1995; Maccaferri and McBain, 1995). Although basket cells in the CA3 and CA 1 regions of Ammon 's horn are known to be involved in tecurrent microcir­cuits (Gulyas et aI. , 1993a.b; Buhl et aI., 1994.1.). data presented above provide clear evidence that CA1 basket cells are also acti­vated by at least one of the major extrinsic excitatory inputs. the Schaffer collateral/commissural pathway. Likewise. bistratifled and axo-axonic cells (Buhl et aI., 1994b) are also involved in fced­forward microcircuits. Although the latter classes of pyra midal cell layer interneurons possess a relatively l31'ge proportion of their dendrites deeply invading the oriens-alveus border region. where they mingle with the bulk of pyramidal cell axon collaterals (Buhl et aI., 1994b; Halasy et al .• 1996), unequi vocal evidence is as yet missing as to whether these neurons also participate in recurrent

. . . micrOCirCUits.

Both the voltage dependence and rhe fast kinetics of com­pound EPSPs resulting from the bulk stimularion of Schaffer col­lateral /co mmissural afferents indicate that these responses are largely mediated by AMI'A-rype glutamate receptors. And indeed, orthodromically elicited EI'SPs in both basket and bistratified cells were sensitive to the action of a bath applied non-NMDA gluta­mate receptOr antagonist. Moreover. ir is reasonable to assume that much of the early AMI'A receptor-mediated EPSP was of monosynaptic origin, owing to the considerably longer latency of disynaptically (recurrent) elicited EI'SPS in CA1 interneurons (Maccaferri 311d McBain. 1995). However, it also appears that some of the response may have been mediated by NMDA re­ceptors. as indicated by the slow kinetics as well as the sensitiv­iry of the residual EI'SP to a bath-applied NMDA receptor an­tagonist. T hese results thus complement previous work providing pharmacological evidence for the presence of NMDA receptors on anatomically identified CA 1 axo-axo nic cells (Buhl et al. . 1994b). Moreover, data on hippocampal interneurons in general have repeatedly prompted very simi lar conclusions (Sa], et al. , 1990; Lacai lle, 1991 ; McBain and Dingledine, 1993; Perouansky and Yaari, 1993).

Interestingly, basket and bistratifled cells showed a significant difference in the rise time of Schaffer collateral/co mmissurally elicited EPSPs. Quite conceivably this EPSP par31Jlerer may have

304 BUHL ET AL.

been, at least in part, determined by the equally prominent di f­

ference in the membrane time constants of basket and bistrati­

fl ed cells. Several other factors may have shaped the £PSP kinet­

ics. among them variabili ty in the properties of AMPA receptors

(Li vsey and Vicini , 1992; Livsey et al ., 1993) and a differential

d egree of electrotoni c attenuation (Thurbon et al ., 1994; reviewed

in Spruston et al. . 1994) . With respect to the fas t rise time ofbas­

ket cell £ PSPs. the jimctional consequence, following the activa­

tion of excitatory synaptic affetents. will be that basket cell action

potentials will generally precede those of bistratifled cells. H ence

basket cell-evoked IPSPs will presumably precede those mediated

by bistratified cells in postsynaptic pyramidal cells, and it is per­

haps no coincidence that basket cell- mediated unitary [PS Ps also

have significantly faster kinetics when compared to bisttatifled cell

effects (Buhl et aI. , 1994a). It therefote appeats that two inde­

pendent factots favor basket cells to be rather effi cient in rapidly

prov iding a perisomatic increase in membrane conductance. In

contrast. the tempo ral properties o fbistratifled cell-evoked lPSPs

may render them mo re effective in shunting N MDA recep­tor-mediated conductances (Staley and Mody, 1992).

GABAergic Control of Basket and Bistratified Cells

Previous studies, using either dual recording techniques o r ex­

tracellular bulk stimulation of synaptic afferents, demonstrated

t1ut physiologically and/or anatomicalIy identified interneurons

receive inhibitory input fro m other, as yet unidentified , local-cir­

cui t neurons (Misgeld and Frotscher, 1986; Lacaille et al ., 1987;

Scharfman et aI. , 1990 ; LacaiUe, 199 1; Buhl et al. 1994 b). In this

respect. basket and bistratifled cells corres po nded well to this gen­

eral pattern . since Sch.ffer coll ateral /commissural flber activatio n

elicited biphasic [PSPs which were pres um ably mediated by both

GABAA and GABA~ teceptors. Because basket (D eller et aI. ,

1994) and bistratified cells are directly targeted by excitatory ex­

trinsic afferents and have been shown to target other intern eu­

ro ns synaptically (Buhl et aI. , 1994a; Sfk et al ., 1995; Halasy et

aI. , 1996), they may not only have a key role in the feedfo rward

co nrro l of principal cells (Buzsaki , 1984; Cobb et al ., 1995), but

also parti cipate in shaping neuronal activity across the GABAergic network (Buzsa ki and C hrobak, 1995) .

Acknowledgments

The authors thank P. Jays, F.D . Kenn edy, D . Latawiec, and

J .D .B. Roberts for expert techni cal support. L. Lyford for secre­

tarial assistance, and S.c. C obb. S.V . Karnup . and V.V. Stezhka

fo r contributing some of the experimental data. K.H. was sup­po rted by the W ellcome T rust.

REFERENCES

Ashwood T] , Lancaster B. Wheal HV (1984) In vivo and in vitro stud­ies on putative inrcrncurom:,s in me rat hippocam pus: possible me­diators of feed- forward inhibition. Brai n Res 293:279- 29 1.

BaudcA. Nusser Z, Roberts ] DB. Mulvi hill E. Mcllhinney RA] , Somogyi

P (1993) T he metabotropic glutamate receptor (mGluR 1 (X) is con­centrated at pcrisynapti c membrane of neuronal subpopulations as detected by immunogold reaction. Neuron 11 :77 1-787.

Bl asco-I ba ncz JM. heund TF (1995) Synaptic input of horizontal in­terneurons in stratum oriens of rhe hippocampal CA l subfield : struc­tural basis of feed-back activation. Eu r] Neurosci 7:2 170-2 180.

Buhl EH. Halasy K. Somogyi P (l994a) Diverse sou rces of hippocam­pal unitary inhibitory posrsynaptic potentials and the Ilumber of synaptic rclease sites. Nature 368:823- 828.

Buhl EH, Han Z-S. Lorinczi Z. Srczhk. VV, Karnllp SV, Somogyi P (1994b) Physiological properties of anatomically idenrified axo­axon ic cells in rhe rat hippocampus. J Neurophysiol 71: 1289- 1307.

Buhl EH. Cobb SR, Halasy K, Somogyi P (1995) Properties of uni tary IPSPs evoked by anatomically identified basket cells in the ra t hip­pocampus. Eur ] Neurosci 7: 1989-2004.

lluzsaki G (1984) Feed-forward inhibition in the hippoca mpal forma­rion . Prog Neurobiol 22: 13 1- 153.

Buzsaki G. Chrobak JJ (1995) Temporal structure in spacially organized neuronal t:nsembles: a role for imerneufonal networks. Curr D pin Nellrobiol 5:504-5 10.

Cobb SR. Buhl EH. Halasy K. Paulsen 0 , Somogyi P (1995) Synchron­i7--a rion of neuronal aC[iviry in hippocampus by individual GABAergic interneurons. Nature 378:75-78.

Connors BW. Gurnick MJ (1990) Intrinsic fi ring patterns of diverse neo­co rtic.1l neurons. Trends Neurosci 13:99- 104.

DeFelipe ] . Hendry SHC. ]ones EG (1989) Synapses of doublc bouquet cells in monkey ccrcbra1 cortex visualized by calbindin immunoreac­rivity. Brain Res 503:49- 54.

Deller T . Nirsch R. Frorscher M (1994) Associational and com missural afferenrs of parvalbumin-immunoreacrivc neurons in the rat hip­pocampus: a combined immunocytochemical and PHA-L srudy. J Comp Neurol 350:61 2-622.

Deuchars J, Thomson AM (1995) Single axon f."r IPSPs el icited by a sparsely spiny interneuron in rat neocortex. Neuroscience 65:935-942.

Gulyas AI. T 6th K. Danos p. Freu nd T F (1 99 1) Subpopuiations of GABAergic neurons containing parvalbumin , calbi ndin D28k, and cholecystokin in in the rat hippocampus. J Camp Ncurol 3 12:371-378.

Gulyas AI , Miles R. Slk A. T6rh K. Tamamaki N. Freund TF (l993a) Hippocampal pyramidal cells excite inhibi tory neurons through a sin­gle release si te. Narure 366:683- 687.

Gulyas AI, Miles R, Hajos N. Freund TF (1993b) Precision and vari­abiliry in postsynapric t. rger selection of inhibitory cells in the hip­pocampal CA3 region. Eur] Neurosci 5: 1729-175 1.

Halasy K. Somogyi P (1993) Subdivisions in the multiple GABAergic innervation of granule cel ls in the den rate gyrus of rhc rat hip­pocampus. Eur J Neurosci 5:4 11 -429.

Halasy K, Buhl EH. LorinC7.i Z. Tam;!s G. Somogyi I' (1 996) Synaptic targer selectivity and inpur of GABAergic basker and bimatified in­terncurons in the CA I area of the rat hippocampus. Hippocampus 6:306-329.

Hami ll 0 1', Huguenard ]R. Prince DA (199 1) Patch-damp studies of voltage-gatccl curn::nts in identified neurons o f the rat cerebral cor­tex. Cerebral COrtex 1 :48-61 .

Han ZS. Buhl EH. Larinczi Z. Somogyi P ( 1993) A high degree of spa­tial selectivity in the axonal and dendri tic domains of physiologically identified local-circuit neurons in the dentate gyrus of the rat hip­pocam pus. Eur ] Neurosci 5:395-4 10.

Kawaguchi Y (I993) Groupings of non pyramidal and pyramidal cells with specific physiological and morphological characteristics in rat frontal cortex. ] Neurophysiol 69:4 16-43 1.

Kawaguchi Y. Ham" K (I988) Physiological heterogeneity of nonpyra­midal cells in rat hippocampal CA 1 region. Exl' Brain Res 72:494-502.

Kawaguchi Y (1995) Physiologica l subgroups of non pyramidal cells with specific morphological characteristics in layer 111111 of rar fro ntal cor­tex. ] Neurosci 15:2638-2655.

INTERNEURON PROPERTIES IN RAT HIPPOCAMPUS 305

Kawaguchi Y, Kubota Y (1993) Correlation of physiological subgroup­ings of nonpyram idal cel ls with parvalbumin- and calbindin D28k-im­munoreactive neurons in layer V of rat frontal Cortex. J Neurophysiol 70:387-396.

K1waguchi Y, Karsumaru H, Kosaka T , Heizmann CW, Hama K (1987) FaSt spiking cells in rat hippocampus (CAI region) contain the cal­cium-binding protein parvalbumin. Brain Res 4 16:369-374.

Kosaka T , Kosaka K, Tateishi K, H amaoka Y, Yanaihara N , Wu J-Y, Hama K (1985) GABAergic neurons conraining CCK-8-like and/or VIP-like im,rnunoreacri vi ries in the rat hippocampus and dentate gyrus. J Comp Neurol 239:420--430.

Kosaka T , Katsumaru H, Hama K, Wu J-Y, Heizmann CW (1987) GABAergic neurons containing the Ca2+ -binding protein parvalbu­min in the rat hippocampus and dentate gytus. Brain Res 4 19: 119-130.

Lacaille J-C (1991 ) Postsynaptic potentials mediated by excitatory and inhibimry amino acids in inrerneurons of stratum pyramidale of the CAI region of rat hippocampal slices in vitro. J Neurophysiol 66: 1441-1454.

b c.1 ille J-C, Schwarrzkroin PA (1988) Srratum lacunosum-molecularc interneurons of hippocampal CAI region. I. Intracellular response characteristics, synapric responses, and morphology. ] New·osci 8: 1400-1410.

LacaiHe J-C, Williams S (1990) Membrane properties of interneurons in stratum o riens-alveus of the CA ! region of rat hippocampus in vitro. Neuroscience 36:349-359.

Lacai lle J-C, MueHer AL, Kunkel DD, Schwartzkroin PA (1987) Local circuit interactions between oriens/alveus inrerneurons and CA I pyra­midal ceHs in hippocampal slices: electrophysiology and morphology. J NellfOsci 7:1979-1993.

Livsey cr, Vicini S (1992) Slower spontaneous excitatory postsynaptic currents in spiny versus aspiny hilar neurons. Neuron 8:745-755.

Livsey CT, Costa E, Vicini S (1993) Gluramare-acriv>fed currents in outside-out patches from spiny versus aspiny hilar neurons of rat hip­pocampal slices. J Neurosci 13:5324--5333.

Maccaferri G, McBain Cj (1995) Passive propagation of LTD to stra­tum oriens-alveus inh ibitory neurons modulates the temporoam­monic input to the hippocampal CAI region. Neuron) 5: 137-145.

McBain Cj , Dingledine R (1993) Heterogeneity of synaptic glutamate receprors on CA3 stratum-radiatum interneurones of rat hippocam­pus. J Physiol (Lond) 462:373-392

McBain Cj , DiChiara TJ , Kauer JA (1994) Activation of metabotropic glutamate receptors differentially affects twO classes of hippocampal interneurons and potentiates excitawry synaptic transmission. ] Neurosci 14:4433--4445.

Misgeld U, Frotscher M (1986) Postsynaptic-GABAergic inhibition of non-pyramidaJ neu rons in the gu inea-pig hippocampus. Neurosci­ence 19: 193- 206.

Nitsch R, Soriano E, Frotscher M (1990) The parvalbumin-containing non pyramidal neurons in rhe rat hippocampus. Anat Embryol (Berl) 181:413--4 25.

Perouansky M, Yaari Y (1993) Kinetic properties of NMDA r«eptor­mediated synaptic currents in rat hippocampal pyramidal cells versus interneurones. J Physiol (Lond) 465:223- 244.

Poncer J-C, Shinozaki H, Miles R (1995) Dual modulation of synaptic

inhibition by distinct metabotropic glutamate receptors in the rat hip­pocampus. J Physiol (Lond) 485:121 - 134.

Sah P, H esrrin S, Nicoll RA (1990) Properties of excitatory postsynap­tic currents recorded in vitro from rat hippoc1mpal interneurones. J Physiol (Lond) 430:605--616.

Scharfman HE (1992) Differentiation of rat dentate neurons by mor­phology and electrophysiology in hippocampal slices: granule cells, spiny hilar cel ls and aspiny 'f.1sr-spiking' cells. In: The dentate gyrus and its role in seizures (Ribak CE, Gall CM, Mody 1, eds), pp 93- 109. Amsterdam: Elsevier.

Scharfman HE, Kunkel DD, Schwarrzkroin PA (1990) Syna ptic con­nections of dentate granule cells and hilar neurons: results of paired intracellu lar recordings and intracellu lar horseradish peroxida.'\e in­jections. Neuroscience 37:693- 707.

Schwartzkroin PA, Kunkel DD (1985) Morphology of identified in­terneurons in the CAI regions of guinea pig hippocam pus. J Com p NeuroI232:205- 218.

Schwarrzkroin PA, Mathers LH (1978) Physiological and morphologi­cal identification of a non pyramidal hippocampal cell rype. Brain Res 157: 1- 10.

Sfk A, Penttonen M, Ylinen A, Buzs:iki G (1995) Hippocam pal CA ) in­terneurons: an 111 vivo intracellular labeling study. ] Neurosci 15: 6651--6665.

Sloviter RS, N ilaver G (1987) Immunocytochemical loc.,lization of GABA-, cholecystokinin, vasoactive intestinal polypeptide-, and so­marostarin-like immunoreactivity in the area denrat3 and hippocam­pus of the rat. J Comp Neurol 256:42--60.

Somogyi P, Nu nzi MG, Gorio A, Smith AD (1983a) A new type of spe­cific interneuron in the monkey hippocampus forming synapses ex­clusively with the axon initial segments of pyramidal cel ls. Brain Res 259:137- 142.

Somogyi P, Smith AD, N unzi MG, Gorio A, Takagi H , W u J-Y (l983b) Glutamate dccarboxyla.l\c immunoreactivity in the hippocampus of the cat: distribution of immunoreactive synaptic terminals with spe­cial reference to the axon initial segment of pyramidal neurons. J Neurosci 3: 1450-1468.

Somogyi P, Hodgson Al , Smith AD, N unzi MG, Gorio A, Wu J-Y (1984) Different populations ofGABAergic neurons in the visual cor­tex and hippocampus of cat contain somatostatin- or cholecystokinin­immunoreactive material. J Neurosci 4:2590-2603.

Somogyi P, Freund TF, Hodgson AJ , Somogyi J , Berollkas D, Chllbb IW (1985) Identified axo-axonic cells ate immunoteactive for GABA in the hippoc.'mpus and visual cortex of the C.1t. Brain Res 332:1 43-149.

Soriano E, Frotscher M (1989) A GABAetgic axo-axonic cell in the fas­cia denrara controls rhe main excirarory hippocampal parhway. Brain Res 503: 170- 174.

Spruston N , Jaffe DB, Johnsron D (1994) Dendritic attenuation of synaptic potentials and currents: the role of passive membrane prop­erties. Trends Neurosci 17:16 1- 166.

Staley K), Mody I (J 992) Shunting of excit.rory input to dentate gyrus granule cells by a depolarizing GABAA receptor-mediated postsynap­tic conductance. J Neurophysiol 68: 197-21 2.

Thurbon D, Field A, Redman S (994) Electrotonic ptofiles of in­terneurons in stratum pyramidale of the CA 1 region of rat hippo­campus. J Neurophysiol 71: 1948-1958.