Single axon fast inhibitory postsynaptic potentials elicited by a sparsely spiny interneuron in rat neocortex

Neuroscience Vol. 65, No. 4, pp. 935-942, 1995

Pergamon

0306-4522(95)00020-8

ElsevierScienceLid Copyright © 1995 IBRO Printed in Great Britain.All rights reserved 0306-4522/95 $9.50+ 0.00

Letter to Neuroscience SINGLE AXON FAST INHIBITORY POSTSYNAPTIC POTENTIALS ELICITED BY A SPARSELY SPINY INTERNEURON IN RAT NEOCORTEX J. DEUCHARS* and A. M. THOMSON Department of Physiology, Royal Free Hospital School of Medicine, Rowland Hill Street, London NW3 2PF, U.K.

Many 9f the different morphological types of interneurons in mammalian neocortex are presumed to be inhibitory, hut to date, conclusive functional data have been lacking. Using paired intracellular recordings in slices of adult rat somatosensory cortex, we present a sparsely spiny, burst fring interneuron that elicits in a simultaneously recorded pyramid a fast inhibitory postsynaptic potential, reversing at --78 mV. Neither inhibitory postsynaptic potential time course, nor paired pulse depression (inter-spike interval 5-120 ms), was affected by addition of the GABA B antagonist] partial agonist 2-OH-Saclofen (250 pM), but increasing extracellular [Ca~÷] enhanced inhibitory postsynaptic potential amplitude at low firing rates and increased paired pulse depression at higher rates. Light microscopic examination of the biocytin-filled neurons revealed the presynaptic cell to be a sparsely spiny interneuron and the postsynaptic to be a small pyramidal neuron, both in layer II. Ultrastructural examination of 16 terminals of the presynaptic interneuron revealed that they formed symmetric contacts with unlabelled neurons, four with neuronal somata, 10 with dendritic shafts and two with spine shafts. This, therefore, is the first report of the properties of a single axon inhibitory postsynaptic potential in neocortex resulting from action potentials in an electrophysiologically and morphologically identified interneuron. We propose that at least some of the sparsely spiny, burst firing interneurons inhibit pyramidal neurons via GABA A receptors. In 14 experiments, in which 897 neurons were recorded, 21 were identified electrophysically as burst firing interneurons. These were similar to the low threshold spiking interneurons previously described.14 Of these, five were successfully filled with dye and reconstructed. All had sparsely to medium spiny non-beaded dendrites, contrasting strongly with fast spiking interneurons (n -- 10 in a different *To whom correspondence should be addressed. IPSP, inhibitory postsynaptic potential.

Abbreviation:

series of experiments) whose dendrites were apsiny and were generally beaded. These burst firing interneurons differed from fast spiking interneurons in displaying higher input resistance 7 7 + 1 6 M ~ (compared with < 50 M~). In some of these neurons inward rectification in response to both hyperpolarizing and depolarizing current injection was apparent, in others, current voltage relations were relatively linear close to rest (see Fig. 1B). In contrast, fast spiking interneurons displayed strong and rapid outward rectification in response to depolarizing current and linear current voltage relations only at membrane potentials negative of rest. In addition, while aspiny, fast spiking interneurons displayed no spike accommodation or adaptation (Fig. I E), spiny interneurons generated a "rebound" depolarizing slow potential, which could elicit a burst of fast action potentials, in response to depolarization from membrane potentials more negative than - 7 0 to - 7 5 mV (Fig. 1A,B,C). This burst initiated a complex afterhyperpolarization, followed, if depolarization was maintained, by additional spikes. Action potentials were, however, curtailed by rapid and deep hyperpolarizations in both types of interneuron resulting in a narrow spike width compared with that of pyramidal neurons (see Fig. 1D). One hundred and nine dual recordings of these interneurons with simultaneously recorded pyramidal neurons yielded one pair in which the sparsely spiny, burst firing interneuron was demonstrably presynaptic for a pyramid. This low yield contrasts with six of 79 pairs in which the pyramidal neuron was presynaptic for the bursting interneuron and 12 of 471 in which fast spiking interneurons were demonstrably presynaptic for postsynaptic pyramids 2~ (data obtained in different experiments and not included in the numbers of experiments or recorded cells given here). The low yield could be due to: (i) a low incidence of such connections; (ii) a low synaptic efficacy easily obscured by noise; or (iii) inputs that were distal to the postsynaptic

935

936

J. Deuchars and A. M. Thomson

soma. This last is perhaps supported by the fact that the postsynaptic pyramid here was a superficial cell with short apical dendrites and was recorded intra-dendritically. The majority of bursting, spiny interneurons in this study were recorded in the deep layers and simultaneously recorded deep layer pyramids generally had long apical dendrites. Previous studies have indicated that bursting, or "low threshold spiking interneurons" can be calbindin positive ~4 and that calbindin positive terminals often innervate dendritic spines and shafts.6-8 In this case, the majority of the terminals of the filled interneuron studied at the ultrastructural

A

level (see below and Fig. 3) were in synaptic contact with dendrites. The morphology of the presynaptic interneuron and postsynaptic pyramid are illustrated in Figures 2 and 3. Both the dendritic morphology (Fig. 2) and the electrophysiological characteristics (see Fig. 1.) of this interneuron were similar to those of low threshold spiking interneurons described previously. ~4 This neuron had sparsely spiny dendrites and an unmyelinated axon that ramified locally and made symmetric synapses (see Fig. 3). Electron microscopic examination of 16 labelled terminals of the presynaptic interneuron revealed that they

D yramid

B 0.4 0.5

-0.5nA

2O mV

-70mV ~

-0.4 -0.6

Bursting~.~rInt E

L 2.0n~ ~ . . .

.. " ~ . ~ . . . ~ 8 6 m V

Fig. 1. Intracellular recordings from the presynaptic sparsely spiny interneuron (A-D) compared with recordings obtained from a pyramidal cell (D) and a fast spiking interneuron (E). In A, from a membrane potential of - 7 0 mV a depolarizing pulse elicited spike discharge exhibiting adaptation and accommodation. An afterhyperpolarization (AHP) followed the pulse. The termination of hyperpolarizing current initiated a slow depolarizing event. In B, from a membrane potential of - 8 6 m V , termination of hyperpolarizing current elicited no slow depolarization, but depolarizing pulse of sut~cient magnitude (> 0.3 nA), elicited a slow depolarization that itself initiated fast spikes. Spikes were followed by complex AHPs, but the longer lasting AHP apparent at the end of the pulse from membrane potentials around 70 mV (A) was not apparent on return to a membrane potential of - 86 mV. In C, depolarizing current pulses of increasing amplitude delivered from a membrane potential of - 86 mV elicited slow depolarizations that elicited bursts of 1-3 spikes. In D, the time course of action potentials in the bursting interneuron are compared with action potentials recorded from a typical pyramidal neuron. In both cases spikes elicited with just threshold depolarizations are illustrated. The spike in the interneuron is clearly of briefer time course and is terminated by a more rapidly developing and deeper afterhyperpolarization. In E, responses of a fast spiking interneuron to depolarizing current injection are illustrated for comparison. In these neurons, unlike the burst firing interneurons, there is little or no spike accommodation, or adaptation (owing to the very rapid time course of these spikes and the inadequate frequency response of the oscilloscope, some of these spikes have been retouched) and large depolarizing currents are frequently required to elicit firing. All traces are single sweep records except for those in D where each trace is an average of eight action potentials. Spikes in A-C are largely attenuated by the inadequate frequency response of the oscilloscope. The full amplitude of these spikes is apparent in D. Note that the same scale bars apply in A, B, C and E. -

-73

I A

/

possible /jf~fr~i~]recording site "" ,f ...........o....... /

~


/~/~

~i ~

.........

.......

........

- .

. . . . . . . . . . . . . .

Postsynaptic pyramid

,o// / /

II III

/ /

i

yt/ /

~

.

~

t

\,

Presynaptic interneuron

, f 7"

I

I

Fig. 2. A, Drawing tube reconstruction of a synaptically connected sparsely spiny presynaptic interneuron and postsynaptic pyramidal cell, in the superficial layers of rat somatosensory cortex. The presynaptic soma and dendrites are in black and axon in red and the postsynaptic soma and dendrites in green. The probable postsynaptic recording site was at an apical dendritic branch point as indicated by the arrow. Scale Bar = 100 Ftm. B, Photograph of interneuron dendrite (see double headed arrow) showing typical sparsely spiny appearance. Scale bar = 15/~m. This pair of synaptically connected neurons was recorded with a pair of conventional sharp electrodes filled with 2 M KMeSO4/2% Biocytin in a 500-/~m-thick slice (see Refs 19, 20). The slice was maintained at 35°C at the interface between 95%OJ5%CO 2 and artificial cerebrospinal fluid containing (in mM) 124NaC1, 25.5NaHCO 3, 3.3 KC1, 1.2KH2PO 4, 2.5CAC12, ! .0 MgSO 4 and 15 D-glucose. After recording the IPSP elicited in the pyramid by action potentials in the interneuron (see Figs 3, 4), both neurons were filled with biocytin (500 ms, 0.5 nA pulses at 1 Hz). The slice was fixed in 1.25% glutaraldehyde/2% paraformaldehyde/15% picric acid in 0.1 M phosphate buffer. This slice was sectioned at 60 # m and processed to localize biocytin with avidin-horseradish peroxidase (see Ref. 10 for further details). 937

938

J. Deuchars and A. M. Thomson

Fig. 3. Electron micrographs of terminals of the presynaptic sparsely spiny interneuron illustrated in Figure 1. These make symmetric type synapses, indicative o f inhibitory inputs and marked with double arrows. In A and C contacts were with unlabelled dendrites. In B a labelled terminal is in synaptic contact with a soma that did not appear to receive asymmetyric synapses and therefore is likely to be a pyramidal cell. A single arrow marks an unlabelled terminal (UT) which also synapses with the soma. In D a labelled terminal is in contact with a spine shaft which received an asymmetric input (UT) to the spine head. In subsequent serial sections the terminal illustrated in D was seen to be in synaptic contact with another spine shaft ($2). No clearly identifiable synaptic contacts with the labelled, recorded, postsynaptic pyramidal cell were observed despite extensive serial section electron microscopic examination of the postsynaptic cell. The failure to recover synaptic connections was probably due to the incomplete filling and therefore recovery of the presynaptic axon. Scale bar = 2 #m.

Single axon fast IPSPs formed symmetric synapses with unlabelled neurons, four with neuronal somata, 10 with dendritic shafts (eight of which could be identified as spiny) and two with spine shafts. N o clearly identifiable synaptic contacts with the labelled, recorded, postsynaptic pyramidal cell were observed despite extensive serial section electron microscopic examination of the postsynaptic cell. The inhibitory postsynaptic potential elicited by action potentials in the interneuron was relatively

939

brief in time course with a mean 10-90% rise time < 5 ms and width at half amplitude < 40 ms. There was, however, some variation in single sweep time course, the fastest events having rise times < 4 ms and the slowest > 6 ms. This and the rather negative reversal potential of the IPSP ( - 7 8 mV, Fig. 4C) might suggest that synapses located at disparate electrotonic distances from the recordings site were involved. Since the postsynaptic recording site appeared to be dendritic (see Fig. 2 for morphology

C

A

1.0

0.0

JL

-1.0

-2.0

, -100

I -90

,

i -80

Membrane

,

~ -70

Potential

,

~ -60

, -50

(mY)

B 1st

IPSP

2nd

IPSP

-56mY ~

3rd IPSP

I

O'0t f

-64mY -76mY ~ I lmV

i

-2 ~

,

.

' .

.

-t _

i

. ~

,

' ~

i

o

.

i

:2

. ,PSP

i

.

-'~

.

Amplitude

i

,

~

.

'

22

.

i

.

'1

,

L

' d

.

i

SOres

(mY)

Fig. 4. A. Electrophysiological recordings from the postsynaptic recording site (see also arrow Fig. IA). The inward rectification apparent with large hyperpolarizing current pulses and the complex afterhyperpolarization following the action potential (AP) are typical of a pyramidal neuron. However, the high (depolarized) spike threshold and the broad width of the AP are indicative of a dendritic recording site. An enlargement of the dendrite at the site indicated by the arrow in Figure 1A, also indicated that the postsynaptic recording site was dendritic. B, Inhibitory postsynaptic potential (IPSP) amplitude distributions are evenly distributed about the mean for IPSPs elicited by the first, second and third APs of short spike trains recorded in the presynaptic interneuron. IPSP mean amplitudes for second and third IPSPs are lower than for first IPSPs. The noise distribution is indicated below left. C, Voltage relation for the average IPSP elicited by 20-100 single presynaptic spikes at each holding potential and below, samples of averaged IPSPs recorded at three postsynaptic membrane potentials. The reversal potential for this single axon IPSP was approximately - 78 mV. Capacitance coupling artifacts have been removed from records. IPSPs were elicited in the postsynaptic pyramid by APs in the presynaptic interneuron. Analogue recordings of both pre- and post-synaptic neurons were analyzed off-line as described previously.~9'2°For analysis of the postsynaptic response, single sweeps were triggered from the rising phase of the presynaptic AP(s). For IPSP amplitude distributions (B), the peak amplitude of the IPSP in each sweep was measured from the baseline and is plotted as a probability. For these distributions, relatively long interval (I00-120) pairs of presynaptic spikes were used to obtain the second IPSP amplitudes plotted in B. Similarly, third IPSPs plotted in B, were elicited by third spikes occurring 100-120 ms after the first spike and following a short (< 30 ms) interval pair of spikes. Peak and baseline measurements were averages of the voltage over a time window that spanned the peak of the IPSP and a 10 ms period preceding the presynaptic spike respectively, for each sweep. Noise distributions were obtained from a similar pair of time windows at a similar interval occurring before the triggering spike.

940

J. Deuchars and A. M. Thomson

and Fig. 4 for the high threshold and long duration of postsynaptic spikes) this might suggest that synapses on several different dendrites were involved. Unfortunately, the presynaptic axon was not fully recovered and the locations of recorded contacts could not be confirmed. It should, perhaps be noted, however, that labelled presynaptic boutons making contact with the labelled postsynaptic soma were not observed despite serial section electron microscopic examination of the entire soma and despite the somatic contacts from this interneuron seen on unlabelled somata (Fig. 3). Paired pulse depression of the IPSP was apparent at inter-spike intervals between 5 and 120ms. The average second IPSP amplitude at shorter interspike intervals was 0.43 normalized relative to average first IPSP amplitude. Brief trains of short interval presynaptic spikes resulted in summing IPSPs that decayed rapidly with no indication of a later hyperpolarizing component (Fig. 5A). As is apparent from Figure 1, the electrophysiological characteristics of this class of interneurons, unlike those of fast spiking interneurons, largely preclude long bursts of very high frequency firing. Brief, high frequency bursts of up to five presynaptic spikes elicited no discernible late hyperpolarizing component. IPSPs reversed at around - 7 8 mV and were approximately 1 mV in amplitude at a postsynaptic membrane potential of - 6 0 m V (Figs 4C, 5). IPSP amplitudes fluctuated and amplitude distributions were evenly distributed about the mean with few apparent failures of transmission (see Fig. 4B). To determine whether GABAB receptors might have contributed to postsynaptic responses, 16 or to paired pulse depression, ~5,22 2-OH-Saclofen4'5 (250/~M) was added to the bathing medium after control data had been obtained (see Fig. 5A). There was a decrease in IPSP amplitude from 1.3 to 1.1 mV (average first IPSP amplitude) after 20 min, possibly due to this drug acting as a partial agonist at presynaptic GABA B receptors. 2a'22 However, 2-OH-Saclofen appeared not to affect significantly either the time course of the IPSP or the relative amplitudes of the second IPSPs in brief interval pairs (second IPSP average amplitude normalized to first IPSP average amplitude 0.43 cf. 0.43 in control). Paired pulse depression was still apparent at longer interspike intervals (second IPSP average amplitude 0.79 compared with first IPSP average amplitude, control data insufficient) and third IPSPs remained smaller than first IPSPs in the presence of 2-OHSaclofen (third IPSP amplitude normalized to first IPSP average amplitude 0.56 in control and 0.53 in 2-OH-Saclofen, respectively). A GABAn receptor mediated component, if it comprised part of the IPSP, might be expected to prolong IPSP time course. The lack of effect on IPSP time course might indicate that GABAB receptors did not contribute to the postsynaptic response. If 2-OH-Saclofen acted presynaptically, it acted to reduce, rather than

to enhance release, but did this without affecting paired pulse depression. That the drug had an effect in the preparation was indicated by the increase in spontaneous synaptic events that appeared after approximately 10 min. No other independent assessment of the efficacy or selectivity of this drug was made in this experiment, but the present finding supports the conclusion reached in a previous study employing minimal electrical stimulation in the hippocampal CA3 region and using a different GABAB antagonist, CGP 55845A, namely that the paired pulse depression observed when a single presynaptic interneuron is activated is independent of GABA B receptor mediation. 22'15a In the latter previous study, only when several presynaptic interneurons were activated was a GABAB receptor mediated component of paired pulse depression observed. Increasing extracellular [Ca2+ ] enhanced first IPSPs at low presynaptic firing rates (0.98-1.3 mV) and depressed second and third IPSP amplitudes relative to the first (see Fig. 5B). The second ISPS average amplitude declined from 0.79 to 0.65 of first IPSP amplitude with long interspike intervals and from 0.43 to 0.27 with shorter interspike intervals at 1 Hz after addition of high Ca 2+. Third IPSPs declined from 0.53 to 0.32 of first IPSP amplitude at 1 Hz in high Ca 2+. Interestingly, average first IPSP amplitude, which had been unaffected by changing firing rate from 0.33 to 1 Hz in control medium, or in 2-OH-Saclofen, declined when firing rate was raised in high Ca 2+, from 1.3 to 0.72 mV and increased again, to 1.0 mV, when firing rate was again slowed. Interneurons in neocortex can be divided into three broad classes by their immunoreactivity for calcium binding proteins. 3'6 9 Parvalbumin and calbindin containing interneurons are reported to exhibit distinct electrophysiological propertiesJ 4 Both are presumed to be inhibitory, but while there is evidence that fast-spiking, parvalbumin containing, aspiny interneurons in neocortex 19'2~ (and unpublished observations) and hippocampus 2 inhibit pyramidal neurons via G A B A A receptors, little information has till now been available about the role played by spiny interneurons with electrophysiological characteristics that distinguish them from fast spiking interneurons. Evidence is growing that the presynaptic neurons responsible for G A B A A and GABAB receptor mediated transmission are distinct. 1'13~17,18'23The present data clearly demonstrate that a sparsely spiny, burst firing interneuron inhibits a postsynaptic pyramid, probably via G A B A A receptors, lnterneurons belonging to this class may play a unique role in cortical circuits, since morphologically similar interneurons which are calbindin immuno-reactive, are a major target of GABAergic basal forebrain afferents in layer V H and in layer II of serotoninergic axons from median and dorsal rapheJ 2 They receive strong excitatory inputs from closely neighbouring pyramidal

A

pr~WnapU©

2-OH-Saclofen

Trigger Spike

OV,,

,|

I~

"

~

.

°

-1

.,

,.

b

-, ~ :-..

•

c--,.--.,--d

~

...-:.--;.-.-..-

, ,

,.

.-__'._.'.,.~.'.,~..:.

.....

. ._. O.~IHz

.

.

.:...

:

:.,

..o;~.

:':'':'

o

"~

'~':~"'

10

:

.

"~ ":":

20

:

.....I

....

""'~

.

30

Time (minutes) b

c

d

,0m.

W

-. 2-oH-Saclofen ~ "°B

•

]

,~ *

'IHz

~ -' i

R

I"

-z

._.......~

o!

a

.'

L

~ 1 o t" ..~,'~e . ." . . . .

pomynapu© ,psp

.

"

5mM Ca:" °

.

•

. .. ~.-.~.~:.. ,,

d-,.-

--.-- o -.-,-

"i

~.. **.

"

*t.

I

,

o

.

.

•

~:.,:. ,~,',

"':

.

.

,*

•

.&"

•

.,, .,.-

".--g

~.z .

.

.

.

~=.~

--

o.

.

-

.

'

~t "

:. o

.s.

".--f

:

o~ .

o

,:

. :

IO

•

21o

.

3(1

Time (minutes)

Fig. 5. Single sweep, baseline to peak IPSP amplitudes plotted against time. Typical presynaptic spike trains that elicited the IPSPs are shown to the left of each plot in A. The spike that triggered data collection and analysis in each case is indicated by the arrow head. Presynaptic action potentials were elicited by injection of long current pulses into the interneuron. Longer interval second and third spikes were elicited at a fixed interval following the first spike by an additional small current pulse superimposed on the longer pulse. Data collection and analysis was triggered from the rising phase of the triggering spike and averaged records illustrated (a-g) are composites of averages triggered by first, second and third APs. Averaged IPSPs illustrated were obtained from the time periods indicated by the letters on the plots. In A, 2-OH-Saclofen (250#m) reduced the amplitude of the elicited IPSPs (cf. b, c, d), but had no effect on their time course, nor on the relationship between first, second and third IPSPs. The control recordings in Aa were obtained at a presynaptic firing rate of one burst of 3 spikes/2 s approximately 15-20 rain before those illustrated in Ab and show that there was no reduction in the amplitude or change in the time course of elicited IPSPs over this period, or when the firing rate was increased to l Hz in Ab. The firing rate was then reduced to 1 burst/3 s (Be) with no significant change in amplitude until the extraccllular [Ca ~* ] was increased from 2.5 to 5 mM (e-f). First IPSPs then increased in amplitude. When the firing rate was raised again to l Hz, in high extracellular [Ca 2÷ ] (Bf), first IPSPs declined and second and third IPSPs were now reduced relative to first IPSPs (cf. f, a~l). The decline in the first IPSP was reversed when the firing rate was again slowed (g). All recordings illustrated were obtained at a postsynaptic MP within 2 mV of - 6 0 mV. To obtain single sweep measurements for shorter interval second IPSPs that summed with first IPSPs, the measurement equivalent to the baseline measure for first IPSPs was taken from a briefer time window of 4 ms just before the second presynaptic spike trigger. This is necessarily a slight underestimate of the second IPSP amplitude and for the average figures given in the text, an average of IPSPs elicited by single presynaptic spikes was subtracted from the averaged response to pairs of presynaptic spikes and the second IPSP average amplitude measured from the subtraction result. 941

942

J. Deuchars and A. M. Thomson

cells ( u n p u b l i s h e d observations) a n d their a x o n s generally aborize within a " c o l u m n " . 8 Their activation o f postsynaptic G A B A A receptors would act b o t h to s h u n t dendritic inputs a n d depress somatic excitability.

Acknowledgements--This work was supported by the Wellcome Trust and Medical Research Council. We would like to extend our thanks to Drs Peter Somogyi and Alan Larkman for their advice on aspects of the processing of biocytin-filled neurons.

REFERENCES

1. Benardo L. S. (1994) Separate activation of fast and slow inhibitory postsynaptic potentials in rat neocortex in vitro. J. Physiol. 476, 203-215. 2. Buhl E. H., Halasy K. and Somogyi P. (1994) Diverse sources of hippocampal unitary inhibitory postsynaptic potentials and the number of synaptic release sites. Nature 368, 823-828. 2a. Caddick S. J. and Chad J. E. (1994) Modulation of synaptic responses in CAI pyramidal neurones of the rat hippocampus by activation of presynaptic GABA B receptors. J. Physiol. 473, 174P. 3. Conde F., Lund J. S., Jacobowitz D. M., Baimbridge K. G. and Lewis D. A. (1994) Local circuit neurons immunoreactive for calretinin, calbindin D-28K or parvalbumin in monkey prefrontal cortex: distribution and morphology. J. comp. Neurol. 341, 95-116. 4. Curtis D. R., Gynther B. D., Beattie D. T., Kerr D. I. B. and Prager R. H. (1988) Baclofen antagonism by 2-OH-saclofen in the cat spinal cord. Neurosci. Lett. 92, 97 101. 5. Davies C. H., Davies S. N. and Collingridge G. L. (1990) Paired-pulse depression of monosynaptic GABA-mediated inhibitory postsynaptic responses in rat hippocampus. J. Physiol. 424, 513 531. 6. DeFelipe J. and Jones E. G. (1992) High-resolution light and electron microscopic immunocytochemistry of colocalized GABA and calbindin D-28K in somata and double bouquet cell axons of monkey somatosensory cortex. Eur. J. Neurosei. 4, 46 60. 7. DeFelipe J., Hendry S. H. and Jones E. G. (1989) Synapses of double bouquet cells in monkey cerebral cortex visualized by calbindin immunoreactivity. Brain Res. 503, 49-54. 8. DeFelipe J., Hendry S. H. C., Hashikawa T., Molinari M. and Jones E. G. (1990) A microcolumnar structure of monkey cerebral cortex revealed by immunocytochemical studies of double bouquet cell axons. Neuroscience 37, 655-673. 9. Del Rio M. R. and DeFelipe J. (1994) A study of SMI 32-stained pyramidal cells, parvalbumin-immunoreactive chandelier cells, and presumptive thalamocortical axons in the human temporal neocortex. J. comp. Neurol. 342, 389-408. 10. Deuchars J., West D. C. and Thomson A. M. (1994) Relationships between morphology and physiology of pyramid-pyramid single axon connections in rat neocortex in vitro. J. Physiol. 478, 423-435. 11. Freund T. F. and Gulyas A. I. (1991) GABAergic interneurons containing calbindin D28K or somatostatin are major targets of GABAergic basal forebrain afferents in the rat neocortex. J. comp. Neurol. 314, 187-199. 12. Hornung J.-P. and Celio M. R. (1992) The selective innervation by serotoninergic axons of calbindin-containing interneurons in the neocortex and hippocampus of the marmoset. J. comp. Neurol. 320, 457-467. 13. Kang Y., Kaneko T., Ohishi H., Endo K. and Araki T. (1994) Spatiotemporally differential inhibition of pyramidal cells in the cat motor cortex. J. Neurophysiol. 71, 280-293. 14. Kawaguchi Y. and Kubota Y. (1993) Correlation of physiological subgroupings of nonpyramidal cells with parvalbumin- and calbindin/D28k-immunoreactive neurons in layer V of rat frontal cortex. J. Neurophysiol. 70, 387-396. 15. Lambert N. A. and Wilson W. A. (1993) Discrimination of post- and presynaptic GABA Breceptor-mediated responses by tetrahydroaminoacridine in area CA3 of the rat hippocampus. J. Neurophysiol. 69, 630-635. 15a. Lambert N. A. and Wilson W. A. (1994) Temporally distinct mechanisms of use-dependent depression at inhibitory synapses in the rat hippocampus in vitro. J. Neurophysiol. 72, 121 130. 16. Otis T. S., De Koninek Y. and Mody I. (1993) Characterization of synaptically elicited GABAB responses using patch-clamp recordings in rat hippocampal slices. J. Physiol. 463, 391-407. 17. Samulack D. D. and Lacaille J.-C. (1993) Hyperpolarizing synaptic potentials evoked in CAI pyramidal cells by glutamate stimulation of interneurons from the oriens/alveus border of rat hippocampal slices. II. Sensitivity to GABA antagonists. Hippocampus 3, 345-358. 18. Samulack D. D., Williams S. and Lacaille J.-C. (1993) Hyperpolarizing synaptic potentials evoked in CA1 pyramidal cells by glutamate stimulation of interneurons from the oriens/alveus border of rat hippocampal slices. 1. Electrophysiological response properties. Hippocampus 3, 331-344. 19. Thomson A. M., Deuchars J. and West D. C. (1993) Single axon excitatory postsynaptic potentials in neocortical interneurons exhibit pronounced paired pulse facilitation. Neuroscience 54, 347 360. 20. Thomson A. M., Deuchars J. and West D. C. (1993) Large, deep layer pyramid-pyramid single axon EPSPs in slices of rat motor cortex display paired pulse and frequency-dependent depression, mediated presynaptically and self-facilitation, mediated postsynaptically. J. Neurophysiol. 70, 2354-2369. 21. Thomson A. M., West D. C. and Deuchars J. (1993) Single axon IPSPs elicited by morphologically identified interneurones in rat neocortex in vitro. J. Physiol. 473, 173P. 22. Wilcox K. S. and Diehter M. A. (1994) Paired pulse depression in cultured hippocampal neurons is due to a presynaptic mechanism independent of GABA B autoreceptor activation. J. Neurosci. 14, 1775-1788. 23. Williams S. and Lacaille J.-C. (1992) GABA Breceptor-mediated inhibitory postsynaptic potentials evoked by electrical stimulation and by glutamate stimulation of interneurons in stratum lacunosum-moleculare in hippocampal CAI pyramidal cells in vitro. Synapse 11, 249-258. (Accepted 14 December 1994)