Exp Brain Res (1981) 43:429-438

Springer-Verlag 1981

Commissural and Perforant Path Interactions in the Rat Field Potentials and Unitary Activity

G. Buzs~iki 1 and G. Cz6h 2 ~Department of Physiologyand 2Department of Biophysics, University Medical School, H-7643 P6cs, Hungary

Summary. The interaction of the commissural and path which traverses the hippocampal fissure and perforant path systems was studied by recording makes synaptic contacts with the granule cells of the extracellular field potentials and single unit activity in as well as with the pyramidal cells of the dentate gyrus in urethane-anesthetized rats. Con- CA 3-4 region (Ram6n y Cajal 1911; Lorente de N6 ditioning commissural volleys suppressed extracellu- 1934; Steward 1976). The axons of the granule cells lar synaptic potentials, population spikes and single (mossy fibers) travel through the hilus of the dentate unit discharges evoked by perforant path stimulation. fascia and form giant synapses with the pyramidal Commissural stimulation (single or repetitive) failed neurons of CA 3-4 region. The latter cells give rise to to induce a population spike, however strong the axons which leave the hippocampus, and in addition, stimulation. About half of the cells fired monosynap- terminate in the stratum radiatum of the ipsilateral tically to perforant path volleys and 20% to commis- CA i region by way of the system. sural volleys. Half of the commissurally driven units The major physiological characteristics of this fired before or coincided with field potential onset. have already been elucidated The antidromic mechanism of these short latency (Andersen et al. 1966a, b, 1971a; L0mo 1971a, bi unitary spikes was shown by the collision test. Winson and Abzug 1977). Much less research has Commisural activation reduced spontaneous cell fir- been devoted to the physiological properties of the ing without previous excitation in 25% of the commissural pathways (Deadwyler et al. 1975a; neurons. Less than 6% of the cells responded to Segal 1977, 1978; Steward et al. 1977; Buzs~iki 1980). stimulation of both inputs, indicating little conver- The commissural fibers originate in the CA 3-4 gence between the two pathways. We contend that a region of the hippocampus and terminate in the inner simple form of recurrent inhibition fails to explain third of the molecular layer of the dentate gyms, in the above findings, and the possibility of feed- the stratum radiatum, and, to a lesser extent, in the forward inhibition by commissural activation has stratum oriens of fields CA 1 and CA 3 of the been raised. contralateral hippocampus (Blackstad 1956; Raisman et al. 1965; Laatsch and Cowan 1967; Gottlieb and Key words: Hippocampus - Feed-forward inhibition Cowan 1973; Hjorth-Simonsen and Laurberg 1977; - Commissural and perforant path inputs - Field potentials - Unit activity Fricke and Cowan 1978; Swanson et al. 1978). The purpose of the present investigation was to study the functional characteristics of the commis- It is apparent that a thorough understanding of the sural system and its interaction with the perforant physiological properties and interactions of the path system in the dentate gyrus. A preliminary numerous hippocampal afferent systems (Chronister report has appeared elsewhere (Buzsgtki et al. 1979). and White 1975; MacLean 1975) is a prerequisite for our understanding of hippocampal functioning. The Methods two major afferent systems to the hippocampus are the perforant and commissural pathways. Axons Surgical Procedure from cells of the entorhinal area form the perforant Thirty male, mature hooded rats (250-350 g) were anesthetized Offprint requests to: Dr. Gy6rgyBuzsfiki (address see above) with urethane (1.5 g/kg, i.p.). Teflon-coatedstainless-steel, twisted

0014-4819/81/0043/0429/$ 2.00 430 G. Buzs~iki and G. Czrh: Interactions of Hippocampal Inputs

bipolar electrodes (100 ~tm in diameter, with intertip distances of Results less than 0.5 mm) were implanted in the CA 3-4 region of the left dorsal hippocampus (3.0 mm posterior to bregma, 3.2 mm lateral to midline and 3.6 mm below the brain surface). An additional pair Field Responses of electrodes was placed in the right angular bundle (AP = 7.0 ram, L = 4.5 ram, V = -4.5 mm) to stimulate perforant path Responses to perforant path and commissural stimu- fibers. The electrodes were fixed with dental cement. A hole was lations were found similar in all major respects as drilled over the right hippocampus and the dura mater was cut previously reported by others (Andersen et al. 1966a, under a dissection microscope. b; 1971a; LOmo 1971a, b; Deadwyler et al. 1975a; McNaughton and Barnes 1977; Steward et al. 1977), and will not be reiterated here. Some comments are Recording and Stimulation Procedures in order, however. Stimulation of the contralateral CA 3-4 field of Recording was made with either glass micropipettes (2 M NaCI, the hippocampus elicited short latency evoked poten- 2-20 M~) or tungsten microelectrodes (2-5 ~tm tip, 5-15 M~?). Pilot experiments indicated that high impedance of the recording tials from both field CA 1 and dentate gyrus. When electrode was a prerequisite for isolating neurons in the granular strong commissural volleys were used, a large popu- layer. An indifferent electrode was placed in the neck muscles. lation spike appeared on the initial deflection of the Amplification, oscillographic display and photography were con- extracellular synaptic wave in the CA 1 pyramidal ventional. The amplifier had a flat frequency response from d.c. to layer. On further penetration by the recording elec- 5,000 Hz. Occasionally the bioelectric data were recorded on FM tape for subsequent analysis. The stimulating pulses were mono- trode, the amplitude of the population spike progres- phasic square waves (0.1 ms 2-100 V), delivered at a rate of sively decreased and fell to zero as the electrode 0.1-0.2 Hz. For certain tests the rate was increased from 2 to traversed the dentate gyms. 100 Hz. Stimulation of the contralateral CA 3-4 region failed to evoke population spikes in the dentate area even at intensities which produced afterdischarges. Histology Repetitive impulses were also ineffective. Because it has been thought that this lack of commissurally Following each experiment, d.c. current was passed through the evoked population spikes in the dentate area may be stimulating electrodes. The glass microelectrode was cut just due to a decreased effectiveness in synchronously below the shank and left in the brain. In experiments utilizing tungsten electrodes, tip locations were marked by passing a d.c. exciting and discharging a large number of granule current (5-10 ~A, 5 s) through the electrodes. The subjects were cells (see Discussion), the flollowing experiment was perfused with saline, 3% potassium ferrocyanide (blue spot carried out. The strength of the perforant path reaction) and formalin. The brains were excised and imbedded in volleys was adjusted slightly above population spike paraffin. Coronal sections were stained with cresyl violet. threshold and then commissural stimulation was increased to supramaximal values. With this proce- dure, in some instances, the commissurally evoked Data Analysis field response was higher in amplitude than the perforant path evoked synaptic potential at any point Bioelectric activity of the hippocampus was sampled at 25-100 ~m along the recording track. This experiment showed steps to a depth of 4-5 mm below the surface of the cortex, 2.7 mm that commissural stimulation was not capable of lateral from the midsaggital suture and 3.0 mm posterior to the bregma. Field potentials and unitary activity were averaged by a discharging large numbers of neurons, although the Neurolog Averager (Digitimer). Unitary activity was collected in synaptic field potential was of greater amplitude than eight animals. Units were classified as theta ceils, complex spike that of the perforant path evoked response, which cells or other cells (Ranck 1973; Fox and Ranck 1981). Theta cells did produce a population spike (Fig. 1). were recognized by their short duration (< 0.5 ms), a rhythmic High frequency stimulation of the commissural firing which was phase-locked to the negative component of the EEG theta waves, and their invariable increase in frequency in path (50 Hz, 5 s) reduced the amplitude of the association with the appearance of theta activity. Theta activity dentate response to single volleys after the train. was induced by tail pinching or by eserin injection (1 mg/kg, i.v.). Recovery of the evoked field response seemed to A unit was classified as a complex spike cell if it produced a firing parallel the recovery of the EEG activity. During pattern as shown in Fig. 5A at least once. Cells which did not meet these criteria were classified as "other" cells. This third category is afterdischarges, the dentate field response was overestimated by the fact that not all units were held long enough reduced to zero or even reversed its polarity. On the to classify them as theta or complex spike cells. Latencies were other hand, high frequency stimulation of the com- determined by measuring the time interval defined by the onset of missural path resulted in long-term potentiation in shock artifact and some special aspect of the response. Short latency commissural units (see Results) were tested for antidromic the CA 1 region (Buzs~iki 1980). activation by measuring the refractory period, and attempting Strong commissural volleys sometimes generated collision (Fuller and Schlag 1976). a later evoked response following the primary G. Buzsfiki and G. Cz6h: Interactions of Hippocampal Inputs 431

A A 4 ms c PP ;: IL _ !! a W c pp V

B

B (.-

C Pig. 1A, B. Comparison of commissural and perforant path evoked responses at different depths of the dentate gyrus. A molecular layer. B granular layer. Note the blocking of the perforant path evoked population spike with short interpulse intervals (lower trace in A). Dotted line in A indicates the algebraic sum of the responses to stimulation of either site alone. Note also the similar size of the synaptic response to comrnissural (c) and perforant path (pp) stimulation and the absence of population spike to commis- sural volleys. Positivity in this and all subsequent figures is up. Calibrations: 4 ms and 2 mV in A and B Pig. 2. A Superimposed responses in the granular cell layer to stimulation of the contralateral dentate gyms. Stimulation rate: 0.2 Hz. B Stimulation at a rate of 3 Hz produced a gradually augmented late response. Note the appearance of unitary spikes dentate field potential. Its latency of onset varied only in the late response. C Superimposed sweeps recorded from the moIecular layer. Stimulation rate: 3 Hz. Note the lack of between 18-25 ms. In some cases at stimulus fre- potentiation of the early field response quencies __ 2 Hz, a population spike developed from this late potential. Earlier studies have provided evidence that the late dentate potential reflects the activation of the pathways described by Hjorth- response and was more effective than a supramaxi- Simonsen (1971) from CA 3 to the ipsilateral entorhi- mal conditioning perforant path volley. With strong nal cortex and back to the dentate (Deadwyler et al. commissural volleys the perforant path evoked popu- 1975b). It was frequently observed that dentate units lation spike was completely abolished. Suppression were driven by the late field response while the early was also obtained with weak commissural volleys, response itself was not capable of discharging the which caused facilitation when paired with a same unit (Fig. 2). homonymous stimulus (commissural-commissural, Fig. 4). In another series of experiments, convergence of Interactions the commissural and perforant pathways was tested by activating the two inputs simultaneously, and this Commissural conditioning volleys decreased the response was compared with responses to stimulation amplitude of the perforant path elicited synaptic of either site alone. When both pathways were response and the population spike in the dentate area activated simultaneously the amplitude of the (Fig. 3). The suppressive effect of commissural response was less than the arithmetic sum of the stimulation outlasted the duration of the field independent responses or the perforant path 432 G. Bnzs~ki and G. Cz6h: Interactions of Hippocampal Inputs

r C

pp-

11

!It! I! Fig. 4. Field responses to homonymous and heteronymous double pulse stimulations in the dentate gyrus. Paired stimuli of equal intensity and duration to the contralateral hippocampus (c) resulted in an increase of the test response (uppertrace). The same conditioning volley, however, depressed the perforant path (pp) evoked test response (lower trace). Middle trace: paired pulse stimulation of the perforant path. Each trace is an average of eight responses G c I Unitary Activity

A total of 204 units were held long enough to test their electrophysiological characteristics. One hundred and fifty-six units maintained spontaneous PP i I !, activity. Thirty units were silent in the resting condi- tion, but could be fired by commissural or perforant path stimulation. Finally, there were some units which fired a few spikes spontaneously and then Fig. 3A-C. Paired commissural (c) - perforant path (pp) pulse remained silent for long periods. Six theta cells and series; recording from the dentate gyrus in the upper molecular 36 complex spike cells were observed (Fig. 5). With layer (A) and in the lower granular layer (B). Several superim- respect to the evoked responses and interactions posed sweeps. Note partial suppression of the synaptic response studied here, these ceils did not appreciably differ and virtually complete inhibition of the population spike at short interpulse intervals. C Test for convergence. (1) Simultaneous from the rest of the units in our sample. activation of the dentate gyrus by contralateral dentate and ipsilateral perforant path stimulation (latency delay compensated). Average of eight responses. (2) Algebraic sum of the responses Perforant Path Responses (average of eight) to stimulation of either path alone. Calibrations: 4 ms, 2 mV One hundred and eight units fired with short latency to perforant path volleys. To obtain the histogram in Fig. 6, units were counted according to their shortest firing latency. With short trains of stimuli, unitary response alone (Figs. 1A and 3C). These findings spikes could regularly follow each pulse up to 50 Hz were consistent in all animals. or so, and unitary spikes occurred after the first two With commissural volleys of moderate intensity, the perforant path elicited population spike was or three pulses of the train up to 100 Hz. sometimes increased after the initial suppression. Facilitation occurred with commissural-perforant Commissural Responses path intervals of 15-25 ms, a latency which corre- sponded to the CA 3--dentate fascia We found 45 units to be activated with a short latency late commissural response (see above). by commissural volleys. Several units fired twice with G. Buzsfiki and G. Cz6h: Interactions of Hippocampal Inputs 433

PP corn unlt 2O A m

,o

10 ] field onset

8 "1 pr ot held

E .11I_lll_lll _111ILIII IAi IIAIIL. C 8 peak of pop B Ill- Irl'- ~ ll'r"r, ~ rl"l'! ,rl"r [

I 2 ~ ~ . , 0 , 2 1. , . r i ~" ; " ; 11; ,i "lk 1'; m S. c I i" Fig. 6. Comparison of latencies of various components of the _ LI_JlJJlI_JI_I I_ responses evoked by perforant path (pp) and commissural (corn) -- "p n- -vr'qr- n'r r~- volleys. In the uppermost histograms, each cell is represented by its action potential response of shortest latency. The field onset and peak of field histograms indicate the latency of beginning and peak of the field potential responses, respectively. The bottom histograms show the latency of the peak of population spike. No population spike could be evoked in the dentate area by stimula- tion of the contralateral hippocampus. Time scales at bottom apply to all histograms

different thresholds. These units helped to define two distinct classes of spikes: a short latency group and a longer latency group. Spikes counted in the former group occurred before (six units) or coincided with the onset of the field potential (17 units) at a constant latency. Those in the latter group started from the rising phase of the field potential and had a more variable latency (e.g., Fig. 9A). Since the rising phase of the commissural-evoked field potential has been interpreted as an extracellular sign of monosyn- aptic excitatory currents in the dentate gyrus (Dead- Fig. 5. A Illustrations of complex spike firing patterns induced by wyler et al. 1975a), any unitary spike with a latency commissural stimulation. B Two examples of spontaneous activity shorter than that of the field potential should be phase-locked to the EEG theta waves ("theta cells"). Lower trace: considered as originating from presynaptic axons or firing activity and theta waves are depressed by a single commis- from antidromic invasion of postsynaptic elements. sural volley (arrow). Spikes are retouched. C Spike-triggered The possibility of a presynaptic axon response could averages of theta waves. Each trace is an average of 256 individual epochs. The averager was triggered by a "theta cell". Calibrations: be rejected in some cases when the same unit could A and B: 4 mV, 100 ms; C: 700 lxV, sweep: 256 ms also be activated by perforant path volleys (Fig. 9A). 434 G, Buzsfikiand G. Cz6h: Interactions of ttippocampal Inputs

The absolute refractory period measured with paired commissural volleys was less than 3 ms, followed by a period of 3-5 ms relative refractoriness when the test spike was partially blocked. After spontaneously occurring discharges, the short latency commissural-evoked spikes were completely blocked up to 10 ms (Fig. 7). We attributed such complete blocks to collision and the phenomenon was thus felt to prove the antidromic mechanism of the evoked spikes. In other units commissural evoked spikes could also be collided with discharges induced by perforant path stimulation. Activation was followed by depression of firing activity for 100200 ms. In another 52 units inhibition occurred without previous excitation (Fig. 8). In the remaining neurons, spontaneous activity was too low to determine any changes in the firing pattern. Complex spike firing patterns were never observed on the field potentials. There were cases, however, when complex patterns were reliably induced by commissural stimulation, firing several 100 ms after the stimulus (Fig. 5A).

Fig. 7a-c. Properties of the short latency unitary spikes to Interactions commissural volleys, a Threshold intensity stimulation, b Partial blockage of the second of two evoked discharges during the relative refractoryperiod, e In one of the two superimposed traces Not only spontaneous activity, but also the perforant the evoked spike fails after a spontaneous one (collision). Filled path evoked spikes could be depressed by commis- squares indicate stimulus artifacts. Calibrations: 2 mV, 1 ms in a sural volleys. High intensity stimulation of the con- and b, and 2 mV, 4 ms in e tralateral hippocampus was not required for this depression. Furthermore, since firing of the unit in response to the conditioning commissural volley was Lack of Population Spike not a prerequisite for the depression, it could be to Commissural Input Activation distinguished from simple refractoriness. The depres- sion lasted for approximately 20 ms and was accom- In the present study, moderate perforant path volleys panied by a diminution of the field potential evoked a negative-going population spike riding on responses to the test volley (Fig. 9). Only 12 units of the initial portion of the positive-going field potential the surveyed population responded with short in the granular layer of the dentate gyrus. Commis- latency to both perforant path and commissural input sural stimulation, on the other hand, failed to evoke stimulation. Four of them were theta cells. Five population spikes in this region even at very high neurons of this overlapping category could be anti- intensities which were capable of inducing afterdis- dromically activated from the contralateral hip- charges. The possibility that the recording electrode pocampus. Two of the five were theta cells. was "off-line" (Andersen et al. 1971b) for the focus of maximal commissural activation is unlikely for several reasons. First, the stereotaxic coordinates of Discussion the stimulating and recording electrodes were identi- cal. Second, single units were driven both ortho- Both entorhinal and commissural afferents to the dromically and antidromically by commissural stimu- dentate gyrus have been believed to be excitatory lation. Third, withdrawal of the recording pipette (Andersen 1960; Deadwyler et al. 1975a; Steward et into the pyramidal layer of the CA 1 region resulted al. 1977), and the only source of inhibition in the in the occurrence of large population spikes. "On- hippocampus is presumed to be of the recurrent type beam" properties of perforant path stimulation were (Kandel and Spencer 1961; Eccles 1969; Andersen et demonstrated by the finding that the threshold of al. 1969). The present findings cast doubt on these unit driving was below the threshold of the popula- suggestions. tion spike. G. Buzsfiki and G. Cz6h: Interactions of Hippocampal Inputs 435

by perforant path activation were two or three times ._.i higher in amplitude than those evoked by commis- sural path stimulation is consistent with this interpre- tation. On the other hand, several other results of the present experiment show that this explanation is not satisfactory. For example, the observation that the width of latency histograms of the units driven by the perforant path input and commissural input, respec- tively, were nearly the same indicates that the degree of synchrony is comparable for the two inputs. Further, in some experiments where the stimulus currents could be adjusted so as to increase the size of the commissural field response above that of the perforant path elicited synaptic potential, commis- sural activation still failed to induce a population spike. A significant proportion of the commissuraUy driven neurons fired before or coincided with the onset of the extracellular field potential. Approxi- mately half also fired at a longer latency on the rising slope of the field potential. Collision tests revealed that the early unit responses antidromically fired by -150 0 t50 300 450 600 750ms stimulation of the contralateral hippocampus. This Fig. 8. Peristimulus time histograms of three different units in the finding was not unexpected in light of the anatomical granular layer of the dentate gyrus. The stimulus (single pulse) was connections known in the area explored. Swanson et applied to the contralateral hippocampus at 0 ms. There were 32 repetitions in each case. Calibration: 100 ms and five spikes/4 ms al. (1978) recently found that horseradish peroxidase bin. Note depression of firing immediately after the stimtdus and injected into the hilus resulted in retrograde filling of rebound of activity at about 200 ms a variety of neurons in the contralateral dentate gyrus just below the granule cell layer, in Cajal's "limiting subzone". The lack of a population spike in the dentate Even if we exclude those cells which responded gyrus was also noted by Deadwyler et al. (1975a). only at an early latency, one fifth of the surveyed These authors suggested that the commissural termi- neuronal population was still responsive to commis- nals in the dentate area were too sparse to synchron- sural stimulation. In about one quarter of all cells ously activate large numbers of granule cells. Our only suppression was seen without previous excita- finding that extracellular synaptic responses elicited tion.

A B

Fig. 9A-C. Depression of responses to volleysin the perforant path (arrows) by commissural (filled squares) impulses. A-C Three different units. Upper frames show control responses to perforant path volleys, and bottom ones, their depression. In C, responses to paired stimulation were obtained by stepwise increases of the conditioning-testinterval. Calibrations: 2 mV; 1 ms 436 G. Buzs~iki and G. Cz6h: Interactions of Hippocampal Inputs

Lack of Convergence of the Perforant Path interneurons in turn have ascending axons that form and Commissural lnputs "basket" complexes which cover the somata of many granule cells (Ram6n y Cajal 1911; Lorente de N6 We noted 12 units which responded with short 1934). This arrangement has been shown to provide latency to both perforant path and commissural input inhibitory feedback in other systems (Andersen et al. stimulations. This number represents approximately 1964; Eccles 1969), and it has been postulated to 6% of our sample. In another comparison, only 10% serve this purpose in the dentate gyrus as well of the perforant path driven units (n = 108) were also (Andersen et al. 1966; L0mo 1971b). fired by the commissural input. Four cells of this The hypothesis of recurrent inhibition requires category were theta cells. If one accepts the proposi- that the population spike produced by the test tion that theta cells correspond to interneurons (Fox stimulus fails to show inhibition until the strength of and Ranck 1975), the degree of convergence seems the conditioning response passes the threshold of even less. Lack of spatio-temporal summation by population spike generation (Andersen 1975). How- simultaneous activation of the two inputs confirmed ever, the results of both Deadwyler et al. (1975a) and our unit data. Had both pathways converged onto the ourselves failed to support a very important step in same cells, and were both excitatory, simultaneous this mechanism, namely, that commissural volleys submaximal stimulation ought to have resulted in a activated a large population of granule cells. Com- greater population spike than expected by linear missural stimulation, without itself inducing such an summation of the response to each pathway alone. In activation, resulted in a more powerful suppression our experiments, however, the population spike was than post-activation induced perforant path inhibi- partially reduced or completely abolished by such a tion (see also McNaughton et al. 1978). procedure. The small degree of overlap would ex- If we accept recurrent inhibition as the sole plain why others failed to obtain heterosynaptic source of inhibition in the dentate gyrus, we have to potentiation in the dentate gyrus (Steward et al. make additional assumptions about the target 1977). Interestingly, inhibition of the perforant path granule cells in order to explain the present findings. evoked response by a prior conditioning commissural First, only a few granule neurons, requiring very little volley is evident in their Fig. 3B, at'an interstimulus "integration" of synaptic volleys for discharge, will interval which produced potentiation with homony- be fired by commissural stimulation. Second, at least mous double pulse stimulation. some portion of the inhibitory interneurons are more easily driven by these low threshold granule cells than by others. Thus, excitation of a small number of Inhibition from the Contralateral Hippocampus granule cells with these features will result in a powerful recurrent inhibition through their basket Even if the low synaptic density hypothesis (Deadwy- neurons. Recent neuroanatomical data are in favor ler et al. 1975a) helps to explain the lack of a of such an interpretation (Laatsch and Cowan 1966; commissural population spike, it fails to explain the Hoff et al. 1980). Another possibility is that commis- powerful inhibitory effects of the commissural vol- sural stimulation activated hilar cells, which in turn leys. Commissural path activation exerted a suppres- inhibited a great number of granule cells in a sive effect on the perforant path elicited population collateral fashion through their basket neurons. It spike (see also McNaughton et al. 1978) and the remains to be answered, however, why this mecha- spontaneous and evoked unit activity without itself nism was not activated by stimulation of the perfo- discharging a large population of neurons. Since rant path input. firing to commissural volleys was not a prerequisite Neither alternative of the recurrent mechanisms for inhibition of the perforant path evoked neuronal would explain why simultaneous activation of the discharges, postactivation refractoriness or shunting commissural and perforant path inputs resulted in of the synaptic current did not explain the suppres- smaller population spikes as compared to perforant sion of unitary activities. Several other mechanisms path stimulation alone. Another possibility is offered could be suggested for this inhibition, the most by the finding of Ranck et al. (Ranck 1973; Fox and conservative of which being the well known recurrent Ranck 1975, 1981). According to their suggestion, inhibition in the hippocampus (Kandel and Spencer complex spike cells and projection neurons on the 1961; Andersen et al. 1964; Eccles 1969). one hand, and theta cells and interneurons on the The only known source of inhibition in the other, are identical. Since in our experiments theta hippocampus are the basket cells. The granule cell cells were fired with a short latency on the rising axons emit recurrent collaterals that innervate inter- slope of field potentials, this would indicate direct neurons located beneath the granular layer. These activation .of dentate interneurons. Such a feedfor- G. Buzs~iki and G. Czth: Interactions of Hippocampal Inputs 437 ward mechanism would explain the strong inhibitory References effect of commissural stimulation without preceding Amaral DG (1978) A Golgi study of cell types in the hilar region of excitation of a great number of granule cells. This the hippocampus of the rat. J Comp Neurol 182:851-914 mechanism, however, must be regarded as tentative Andersen P (1960) Interhippocampal impulses. II. Apical dendri- until a direct connection of commissural afferents to tic activation of CA 1 neurons. Acta Physiol Scand 48: inhibitory interneurons is discovered. The fact that 178-208 Andersen P (1975) Organization of hippocampal neurons and their several basket and non-granule cells project their interconnections. In: Isaacson RL, Pribram KH (eds) The dendrites into the molecular layer (Amaral 1978; hippocampus, vol 1. Plenum Press, New York, pp t55-175 Ribak et al. 1978; Seress 1978), together with the Andersen P. Bliss TVP, Skrede KK (1971a) Unit analysis of present physiological findings, suggest that dentate hippocampal population spikes. Exp Brain Res 13:208-221 interneurons do receive commissural afferent fibers. Andersen P, Bliss TVP, Skrede KK (1971b) LameUar organization of hippocampal excitatory pathways. Exp Brain Res 13: 222-238 Andersen P, Eccles JC, L0yning Y (1964) Pathway of postsynaptic Identification of Dentate Units inhibition in the hippocampus. J Neurophysiol 27:608-619 Andersen P, Holmquist B, Voorhoeve PE (1966a) Entorhinal activation of dentate granule cells, Acta Physiol Scand 66: The method of unitary recording might select indi- 448--460 viduals from a population of cells which may not Andersen P, Holmquist B, Voorhoeve PE (1966b) Excitatory characteristically fire to a given volley. There are two synapses on hippocampal apical dendrites activated by purely electrophysiological approaches to classifying entorhinal stimulation. Acta Physiol Scand 66:461-472 units. One of them considers spike morphology, Andersen P, Gross GN, LCmo T, Sveen O (1969) Participation of inhibitory and excitatory interueurons in the control of firing rate and pattern, and the relationship between hippocampal cortical output. In: Brazier MAB (ed) The unitary firing and slow waves, to distinguish theta interneuron. University of California Press, Berkeley, pp cells from complex spike cells (Ranck 197'3; Fox, and 415-465" Ranck 1975). The other approach uses evoked Blackstad T (1956) Commissural connections of the hippocampal responses and classifies cells according to the sources region in the rat, with special reference to their mode of termination. J Comp Neurol 105:417-538 from which they are fired. This approach is widely Bland BH, Andersen P, Ganes T, Sveen O (1980) Automated used to recognize granule cells, pyramidal cells and analysis of rhythmicity of physiologically identified hippocam- basket cells (Andersen et al. 1964, 1966; Bland et al. pal formation neurons. Exp Brain Res 38:205-219 1980). Buzs~ki G (1980) Long-term potentiation of the commissural path- CA 1 pyramidal synapse in the freely moving rat. Neurosci As we have demonstrated, at least some of the Lett 19:293-296 units driven monosynaptically by perforant path Buzs~iki G, Cz6h G, Grasty~n E, Kell6nyi L, Czopf J (1979) volleys could be fired antidromically by stimulating Commissural, perforant path and septal interactions in the the commissural path. Obviously, these neurons dentate gyrus of the rat. Neurosci Lett [Suppl] 3:$67 were not granule cells. Since many non-granule Chronister RB, White LE, Jr (1975) Fiber architecture of the : Anatomy, projections and structural neurons are scattered in and just below the granular significance. In: Isaacson RL, Probram KH (eds) The hip- layer (e.g., basket neurons, "mossy cells" of Amaral pocampus, vol 1. Plenum Press, New York, pp 9-39 1978), and many more send their thick dendrites Deadwyler SA, West JR, Cotman CW, Lynch G (1975a) A through the granule cell layer, localization of the neurophysiological analysis of commissural projection to recording electrode tip in the granular layer, either dentate gyrus of the rat. J Neurophysiol 38:16%184 Deadwyler SA, West JR, Cotman CW, Lynch G (1975) Physiolog- by field potentials or histology, does not guarantee ical studies of the reciprocal connections between the hip- that the recorded unit activity reflects granule cell pocampus and entorhinal cortex. Exp Neurol 49:35-67 activity. To our knowledge there are no electrophy- Eccles JC (1969) The inhibitory pathways of the central nervous siological characteristics which conclusively identify a system. Thomas, Springfield, IL Fox SE, Ranck JB, Jr (1975) Localization and anatomical identifi- neuron as a granule cell. In conclusion, it will be cation of theta and complex spike cells in dorsal hippocampal necessary in future work to use more refined formation of rats. Exp Neurol 49:299-313 neurophysiological criteria combined with anatomi- Fox SE, Ranck JB, Jr (1981) Electrophysiological characteristics cal marking techniques for classification of dentate of hippocampal complex-spike cells and theta cells. Exp Brain Res 41:399-410 units. Fricke R, Cowan VM (1978) An autoradiographic study of the Acknowledgements. We are grateful to Drs. E. Grasty~n, B.L. commissural and ipsilateral hippocampo-dentate projections McNaughton, C.A. Barnes, L.S. Leung, and J. Winson for their in the adult rat. J Comp Neurol 181:253-270 comments on an earlier version of this paper, and to Drs. J.B. Fuller FH, Schlag JD (1976) Determination of antidromic excita- Ranck, S.E. Fox, E. Eidelberg, R.M. Douglas and G. Berluchi tion by the collision test: problems of interpretation. Brain for discussions. We also wish to acknowledge technical contribu- Res 112:283-298 tions by Dr. L. KeU6nyi, Zsuzsa Herzfeld and Zita Cz6h. We also Gottlieb DI, Cowan WM (1973) Autoradiographic studies of thank Eszter Aggrdy and Becky Ramirez for typing the manu- connections of the hippocampus and dentate gyrus of the rat. script. I. The commissural connections. J Comp Neurol t49:393-422 438 O. Buzsfiki and G. Cz6h: Interactions of Hippocampal Inputs

Hjorth-Simonsen A (1971) Hippocampal efferents to the ipsilat- Ralsman G, Cowan W, Powell T (1965) The extrinsic afferent, eral entorhinal area. An experimental study in the rat. J commissural and association fibers of the hippocampus. Brain Comp Neurol 142:417-438 88:963-996 Hjorth-Simonsen A, Laurberg S (1977) Commissural connections Ram6n y Cajal (1911) Histologie du syst6me nerveux de l'homme of the dentate area in the rat. J Comp Neurol 174:591-606 et des vert6br6s, vol 2. Maloine, Paris Hoff SF, Scheff SW, Cotman CW (1980) A bimodal population of Ranck FB, Jr (1973) Studies on single neurons in dorsal hippocam- dentate gyrus granule cells demonstrating age-related pal formation and septum in unrestrained rats. I. Behavioral changes. Neurosci Abstr No. 247.30 correlates and firing repertoires. Exp Neurol 42:461-531 Kandel ER, Spencer WA (1961) Electrophysiology of hippocam- Ribak, CE, Vaugh JE, Saito K (1978) Immunocytochemical pal neurons. II. Afterpotentials and repetitive firing. J localization of glutamic acid decarboxilase in neuronal somata Neurophysiol 24:243-259 following colchicine inhibition of axonal transport. Brain Res Laatsch RH, Cowan WM (1966) Electron microscopic studies of 140:315-332 the dentate gyms of the rat. I. Normal structure with special Segal M (1977) Excitability changes in rat hippocampus during reference to synaptic organization. J Comp Neurol 128: conditioning. Exp Neurol 55:67-73 35%396 Segal M (1978) A correlation between hippocampal responses to Laatsch RH, Cowan WM (1967) Electron microscopic studies of interhemispheric stimulation, hippocampal slow rhythmic the dentate gyms of the rat. II. Degeneration of commissural activity and behaviour. Electroencephalogr Clin Neurophy- afferents. J Comp Neurol 130:241-262 siol 45:409-411 L0mo T (1971a) Patterns of activation in a monosynaptic cortical Seress L (1978) Pyramid-like basket cells in the granule layer of pathway. The perforant path input to the dentate area of the the dentate gyms in the rat. J Anat 127:163-168 hippocampal formation. Exp Brain Res 12:18-45 Steward O (1976) Topographic organization of the projections L0mo T (1971b) Potentiation of monosynaptic EPSP's in the from the entorhinal area to the hippocampal formation of the perforant pathdentate granule cell synapse. Exp Brain Res 12: rat. J Comp Neurol 167:285-314 46-63 Steward O, White WF, Cotman CW (1977) Potentiation of the Lorente de N6 R (1934) Studies on the structure of the cerebral excitatory synaptic action of commissural, associational and cortex. II. Continuation of the study of the Ammonic system. entorhinal afferents to dentate granule cells. Brain Res 134: J Psychol Nem-ol 46:113-177 554-560 MacLean PD (1975) An ongoing analysis of hippocampal inputs Swanson LW, Wyss FM, Cowan WM (1978) An autoradiographic and outputs: Microelectrode and neuroanatomical findings in study of the organization of the intrahippocampal association squirrel monkeys. In: Isaacson RL, Pribram KH (ed) The pathways in the rat. J Comp Neurol 181:681-716 hippocampus, vol 1. Plenum Press, New York, pp 177-211 Winson J, Abzug C (1977) Gating of neuronal transmission in the McNaughton BL, Barnes CA (1977) Physiological identification hippocampus. Efficacy of transmission varies with behavioral and analysis of dentate granule cell responses to stimulation of state. Science 196:1223-1225 the medial and lateral perforant pathways in the rat. J Comp Neurol 175:43%454 McNaughton BL, Douglas RM, Goddard GV (1978) Synaptic enhancement in fascia dentate. Cooperativity among coactive afferents. Brain Res 157:277-293 Received September 12, 1980