The effects of brainstem peribrachial stimulation on perigeniculate neurons: The blockage of spindle waves

Akuroscience Vol. 31, No. 1, pp. 1-12, 1989 Printed in Great Britain

0306-4522/89

$3.00 + 0.00

Maxwell Pergamon Macmillan plc 0 1989IBRO

THE EFFECTS OF BRAINSTEM PERIBRACHIAL STIMULATION ON PERIGENICULATE NEURONS: THE BLOCKAGE OF SPINDLE WAVES B. Hu, M. STBRIADEAND M. DFSC~~~NIZS Laboratoire de Neurophysiologie,

Dtpartement de Physiologie, Faculte de Mtdecine, Universitt Lava], Quebec, Canada GlK 7P4

Ahstraet-The mode of action of afferents arising from the brainstem peribrachial region at the midbrain-pontine junction on neurons recorded from the reticular thalamic sector adjacent to the lateral geniculate nucleus (perigeniculate cells) was investigated at the intracellular level in the cat. Experiments were performed in cats under barbiturate or urethane anaesthesia and in non-anaesthetized deafferented animals. Most cats were pretreated with reserpine (l-2 mg/kg) and were also acutely deprived of their retinal and cortical visual inputs. It was found that peribrachial stimulation produced a short train of fast-rising depolarizations followed by a long-lasting period of hyperpolarization in all perigeniculate neurons. Although the latest part of the early depolarizations preceding the hyperpolarization resulted from a parallel activation of lateral geniculate relay neurons by peribrachial afferents, those occurring at shortest latencies appear to result from a direct excitation produced by peribrachial afferents. Furthermore, these early excitatory postsynaptic potentials persisted under deep barbiturate anaesthesia, a condition that prevents activation of thalamic relay neurons by peribrachial stimulation. The evoked hyperpolarization decreased with membrane hyperpolarization, was associated with a 4&50% increase in membrane conductance and was insensitive to Cl injections. It was no longer observed within one hour after iv. injection of scopolamine. However, the depolarizing responses were not depressed by this muscarinic antagonist. Iontophoretic applications of scopolamine also removed peribrachial-evoked inhibition of synaptic responses triggered by optic chiasma stimulation. The peribrachial input exerted a powerful control on the oscillatory behavior of perigeniculate neurons. Spindle oscillations which are generated within the reticular thalamic complex were readily blocked by peribrachial stimulation. It is then concluded that the transition from an oscillatory to a relay mode of operation in the thalamus is controlled at least in part by a muscarinic inhibition of reticular thalamic neurons. The synaptic mechanism responsible for the early depolarization remains to be elucidated.

geniculate (LG) nucleus and its reticular thalamic (RE) sector, the perigeniculate (PG) nucleus, that they receive an important bilateral cholinergic input from neurons located around the brachium conjunctivum at the pontomesencephalic level.*’ In the present series of papers, we shall refer to this region as to the peribrachial (PB) region. Moreover, a less numerous population of retrogradely labeled cells that were not stained for choline acetyltransferase (ChAT) were found in the dorsal raphe, locus coeruleus and caudal part of the PB area. A recent study”’ concerned with the brainstem projections to the LG nucleus in cats has revealed that these ChATnegative retrogradely labelled neurons are either serotonergic or catecholaminergic. While a similar study remains to be done in the PG nucleus, the presence of noradrenergic and serotonergic fibres in other sectors of the RE nucleus*’ suggests that these results may also apply to the PG nucleus. This new panorama of connections raises problems concerning the specificity of the effects obtained at the thalamic level following MRF stimulation. The problem of specificity is still more troublesome when one considers that these brainstem regions may be interconnected by a rich network of axonal collaterals and that they may influence indirectly the activity of thalamic cells via a cortical loop.

The present series of experiments were undertaken in order to determine how the mesencephalic reticular formation (MRF) controls the activity of thalamic neurons. It is well known that MRF stimulation desynchronizes the electroencephalogram (EEG) and changes the firing mode of thalamic and cortical neurons. Among the effects previously observed at the thalamic level, MRF stimulation was found to block spindle oscillations, to induce tonic neuronal firing, to increase cellular excitability and, under certain conditions, to trigger sharp spiky waves known as ponto-geniculo-occipital waves.26*29These various effects were all ascribed to the activation of a reticulothalamic pathway presumed to be cholinergic. Until very recently the neuronal populations of the brainstem projecting to the thalamus remained ill-defined, both anatomically and chemically. Recent studies combining retrograde transport with enzymatic markers have specified the source and nature of these connections in cats’0,23,27and rats.28*32It has been shown, for instance, in the case of the lateral

ACh, acetylcholine; ChAT, choline acetyltransferase; LG, lateral geniculate; MRF, mesencephalic reticular formation; OX, optic chiasma; PB, peribrachial; PG, perigeniculate; PGO, ponto-geniculooccipital; RE, reticularis thalami.

Abbreviations:

1

B. Hu et al.

2

The action of mesencephalic reticular formation on reticularis thalami neurons

The visual sector of the RE nucleus adjacent to the LG nucleus is usually referred to as the PG nucleus. The RE nucleus is a network of GABAergic neurons acting as an interface between the thalamus and the cortex. Recent studies have disclosed the essential role of this structure in the genesis of oscillations in thalamocortical systems and it has been demonstrated that RE neurons act as generators of EEG spindle waves. 31 Under natural conditions, spindles appear during drowsiness and slow wave sleep and disappear on arousal and during EEGdesynchronized sleep. 1g,29Experimentally, spindles can be induced in acute animals by transections that separate the brainstem from the upper mesencephalot? and they are readily blocked by MRF stimulation. It may then be expected that the blockage of spindle oscillations produced by MRF stimulation is exerted at the very site where these oscillations are generated (i.e. the RE nucleus). A series of studies using extracellular recordings in anaesthetized animals has shown that MRF stimulation or iontophoretic ACh applications depressed the spontaneous and evoked discharges of RE neurons.2~4~‘2-‘4,24 This inhibitory action of ACh has also been confirmed by intracellular recordings in guineapig thalamic slices. 2o This set of results seems to support the hypothesis of Singer,25,26who proposed that the MRF-induced arousing reaction was mainly mediated at the thalamic level by a cholinergic inhibition of RE cells resulting in disinhibition of thalamocortical systems. On the other hand, extracellular recordings in chronic cats3’ and in urethaneanaesthetized rats’* have reported that MRF stimulation excited RE neurons and that ACh could also induce tonic firing. In the present series of experiments, the effects of PB stimulation were studied in thalamic neurons of the LG and PG nuclei. Our investigation concentrated on the LG-PG nuclei for the following reasons: (1) the connectivity and intrinsic organization of the LG-PG complex are the best known among all thalamic regions; (2) the brainstem inputs to the LG and PG nuclei have been well determined both anatomically and histochemically; (3) it was easy to acutely deprive the LG-PG complex of its peripheral and cortical inputs, thus preventing the occurrence of indirect polysynaptic effects. In addition, the cholinergic effects induced by PB stimulation could be studied in isolation after massive depletion of transmitters in the noradrenergic and serotonergic afferent pathways. EXPERIMENTAL PROCEDURES

Two types of preparations were used in of experiments: cats anaesthetized with (Nembutal) (35 mg/kg; n = 7) or urethane and non-anaesthetized deafferented cats these animals were also pretreated

the present series Na-pentobarbital (1.4 g/kg; n = 22) (n = 4). Some of with reserpine

(1.4 mg/kg, i.p.) 24 h before recording sessions. Such doses of reserpine were shown to deplete brain monoamines to less than 10% of the control values after 5 h. The maxima1 depletion (less than 5%) was observed after 24 h.’ The deafferented preparations were performed in the following way. Prior to the recording sessions, large bilateral electrolyte lesions were made under ketamine anaesthesia (4Omg/kg) in the lateral pons close to the point of emergence of the trigeminal nerves. This procedure was performed in four cats under aseptic conditions. Immediately after the lesions, cats suffered from a Cheyne-Stokes respiratory syndrome whose severity diminished when animals recovered from anaesthesia. Effectiveness of trigeminal deafferentation was demonstrated by jaw muscle paralysis and by the absence of motor reactions to noxious stimuli applied to the skin area innervated by trigeminal afferents. During the post-operative period, animals were hand-fed with a high-protein milk substitute. After three days of survival, cats were anaesthetized with Fluothane (3.5%) for surgery. A bulbospinal cut was then made and anaesthesia was stopped. As an additional precaution, all head wounds were generously infiltrated with Duracaine every 3 h. Animals were paralysed with Flexedil and respired artificially. The CO, level of expired air was kept at 3.7 f. 0.3% and rectal temperature was maintained at between 37 and 39°C. The EEG of these deafferented cats was monitored and appeared similar to that observed during natural slow-wave sleep, with numerous sequences of spindle waves. In order to reduce synaptic noise and prevent indirect polysynaptic effects that might result from PB stimulation, both retinae were acutely lesioned and the ipsilateral cortical areas 17 and 18 were removed. These surgical procedures were carried out under Fluothane (3.5%) or urethane anaesthesia. For retina1 lesions, both eyeballs were first injected with Duracaine and then emptied of their content. Small crystals of silver nitrate were used for burning ganglion cells’ axons. The cortex over the LG was also removed to reduce the incidence of micropipettes breaking during penetrations. In some cats, recordings were also performed in the rostra1 thicker part of the RE complex. In those cases, stimulating electrodes were implanted into the motor cortical area for the identification of the recorded cells. Micropipettes used for intracellular recordings were filled with a K-acetate (3.5 M) or KC1 (3 M) solution. For iontophoretic studies, an NaCl-filled micropipette was glued to a double-barrel pipette (tip separation 30 pm). One barrel contained NaCl (3 M) for injecting neutralizing currents while the other contained either hexamethonium (2 M, pH 4) or scopolamine (0.2 M, pH 4). Bipolar stimulating electrodes were implanted into the optic chiasma (OX) and into the ipsilateral PB region at the midbrain-pontine junction (AP, - 1 to + 1; L, 3 to 4; D, -2)5 where cholinergic PB neurons projecting to the LG-PG nuclei have been previously localized.‘0~27Stimuli were delivered as single shocks (duration: 200 ps; intensity: 5&400 PA) or as short trains (3-10 shocks at 10&200 Hz). Data were stored on magnetic tapes (bandwidth: DC -2.5 kHz) and processed using a digitizing waveform analyser. Long episodes of recording were also displayed directly on a rectilinear pen recorder by reducing the tape recorder speed during playback. However, this procedure reduced by lO-20% the amplitude of action potentials. At the end of experiments, cats were killed with a large dose of barbiturate and perfused with formalin. Electrode location and brainstem lesions were examined using Nissl stained sections. Cell identificationand data basis In the course of the present experiments, two neuronal populations were identified: LG relay neurons and PG neurons. The typical intracellular responses of LG and PG neurons to OX stimulation are shown in Fig. 1. LG cells responded to OX stimulation with a short latency excitatory

The blockage of thalamic oscillations

-65

dx

mV

10ms

Fig. 1. Electrophysiological identification of LG relay neurons and PG cells by OX stimulation. Inserts show the isolated EPSPs. Spike firing and inward current activation were prevented by hyperpolarizing the cells. Recordings performed under barbiturate anaesthesia. Time calibrations under the LG and PG traces also apply to insert traces. Membrane potentials for the insert traces: -65 mV for the LG cell and -74 mV for the PG neuron. postsynaptic potential (EPSP) (usually below 2 ms) and single spike discharge followed by a long period of hyperpolarization. PG cells responded to the same stimulus with a long depolarization (latency: 2-5 ms) crowned with a spike burst that often repeated at intervals of 150 ms. Though LG and PG cells both have the intrinsic conductance to generate burst responses, the PG cell burst is much longer and is never followed by a hyperpolarization. PG cells demonstrated the same burst structure as that previously reported for rostra1 RE cells identified by intracellular staining.u The effects of PB stimulation were studied in 19 PG neurons and additionally in six rostra1 RE cells having resting potentials negative to - 50 mV. Those rostra1 RE units were identified by their repetitive long burst discharges triggered by motor cortex stimulation?2 Over 100 additional PG or rostra1 RE neurons were also studied extracellularly. Statistics concerning relay neurons will be presented in the companion papers.‘5,‘6 RESULTS

Extracellular

recordings

All RE cells recorded extracellularly were inhibited by PB stimulation. This inhibition was best demonstrated by using a conditioning testing paradigm where the burst discharges evoked by OX stimulation in PG cells were suppressed by conditioning PB stimuli (Fig. 2). This inhibition started within the first 1Oms after the shocks and could last for up to 1 s (usually about 500 ms).

Intracellular

recordings: the depolarizing

response

When RE cells were recorded intracellularly, the most obvious effect of PB stimulation was indeed a large and long-lasting hyperpolarization (Figs 3, 6, 7, 8 and 9). However, this hyperpolarization was always preceded by depolarizing responses that rarely reached firing threshold. These depolarizing responses started at a latency of 8-10 ms after the first shock and lasted up to 200ms in non-anaesthetized or in urethaneanaesthetized animals. The leading part of the depolarizations was made of discrete EPSPs appearing at fixed latency after the stimulus artefact (see inserts in Fig. 3). Under barbiturate anaesthesia, these early EPSPs were immediately followed by the hyperpolarizing response. In non-anaesthetized or in urethane-anaesthetized cats, the depolarizing response persisted for about 200 ms. This prolonged depolarization was characterized by an indented envelope, suggesting that it may result from the repetitive firing of thalamic relay cells. A comparison between the firing of LG neurons and the depolarization of PG cells is shown in Fig. 4. The PC cell in trace A and the LG neuron in traces B were recorded in the same animal under urethane anaesthesia. In spite of the close temporal relationship between relay cell firing and the rapid EPSPs in the PG neuron, it

4

B. Hu et al.

I

B

I----------~-----. -‘_I

AREA OFCONTROL REVISEf mean+BSD)

._,-,--_I IOR

600

INTERVALS BiTEEN PB AND OX STIWLdTIONS Imsi

* +

SOS38

Fig. 2. inhibition of OX-evoked burst discharges in PCi neurons by PB stimulation. Control responses to OX and PB stimulations are shown in (A) and (8). The depressive effect of PB stimulation is displayed in (C). The gap in trace C5 represents 400 ms. The time course of the PB-evoked inhibition for the same unit is shown in the right-hand panel chart. The area of control response results from the average of IO sweeps. A polynomial regression pracedure was used for curve fitting.

is quite obvious that the discharges of the LG cell never occurred at late&es short enough to produce the early phase of the depolarizing response observed in PG neurons. No relay cell fired at a latency shorter than 20 ms following PB stimulation. The time relationship between the latency of the first PB-evoked EPSP in 13 PG cells and the latency of the first spike discharge evoked by PB stimulation in a similar number of randomly chosen relay cells is shown in Fig. 5. This histogram makes it clear that the earliest EPSPs observed in PG cells after PB stimulation did not result from the firing of LG relay neurons. However, it is also evident that the delayed firing of relay cells could be responsible for the prolongation of the depola~~ng response observed in PG cells. This view is also supported by the persistance of an early depolarizing response in PG neurons recorded under barbiturate anaesthesia, a condition making thalamic relay neurons completely unresponsive to PB stimulation (see companion paper”). The unresponsiveness of relay ceil under barbiturate anaesthesia was indeed correlated with the disappearance of late tonic EPSPs in PG cells resulting in an apparent earlier onset of the hyperpolarization (Fig. 3A).

The hyperpolarizing response of RE neurons to PB stimulation always lasted for at least 500 ms. It could be obtained with single stimuli (Fig. 3C) but short trains were usually used to produce larger responses.

Its latency could not be precisely dete~ined due to the presence of the earlier depolarization. Nevertheless, this latency appears to be short enough (l&20 ms) to prevent spike firing by the early depolarization. The occasional firing observed intracellularly (Fig. 3A) might have resulted from a positive shift ‘of the resting potential due to cell impalement. indeed, spike discharges were never observed extracellularly. The amplitude of the hy~rpolarization decreased with inward current injections or when the cells hy~r~lar~zed spontaneously. In the traces shown in Fig. 6A, the membrane potential hy~rpolar~~ by about 20mV when the PG cell resting potential was ciose to - 50 mV but PB stimulation did not hyperpolarize the eelI during a spontaneous interspindle lull when its membrane potential was already at - 70 mV (Fig. 6A2). In the other example shown in Fig. 6B, the hy~rpola~zation was reduced in size by injection of inward current. The amplitude and polarity of the response were never affected by intracellular Cl injections. The conductance change associated with the hyperpolarizing response was measured in four neurons according to the procedure illustrated in Fig. 7. Control current pulses were injected at rest and produced rebound burst discharges. A pulse was then preceded by PB stimulation and, thereafter, the membrane was hype~~a~zed with DC current in order to take into account the anomalous rectification of the RE cell’s membrane.22 When measured in this

The blockage

of thalamic

oscillations

E

0 el

1

-60mV

50 ms

B,AA

C.

-63

mV

-63

mV

Fig. 3. Effects of PB stimulation on PG neurons. The PG cell in (A) was recorded under barbiturate anaesthesia and those displayed in (B) and (C) were recorded from a deafferented preparation. Three PB shocks (A and B) evoked first step depolarizations at fixed latencies (see inserts) that were followed by a long period of hyperpolarization. Note that in the deafferented preparation (B and C) the early depolarization was more prolonged and delayed the apparent onset of the hyperpolarization. In some cases, the whole sequence could also be obtained by a single PB stimulus (C). Inserts in (B) and (C) are expansions of the main traces. Insert in (A) was taken from a different trace in the same cell. The calibrations for the main trace and insert in (A) also apply for (B) and (C).

way, the conductance increase associated with the response was of the order of 40-50%. The increase in membrane conductance was also clearly demonstrated by conditioning the OX-evoked response by PB stimulation. This is illustrated in Fig. 8, where the top traces show the control burst discharges triggered in a PG cell by a single OX stimulus. The burst response was greatly reduced down to a single spike when it was preceded by a short PB train (Fig. 8B and C). It must be noted in the traces of Fig. 8B that the response was depressed even during the early phase of depolarization preceding the apparent onset of the liyperpolarizing response. These results strongly suggest that the mechanism of PB inhibition in RE cells involves a

large drop in membrane input resistance that renders these cells virtually transparent to synaptic currents and also short-circuits intrinsic inward currents. The blockage of spindle oscillations In anaesthetized as well as in non-anaesthetized cats, the spontaneous activity of PG cells was characterized by the recurrence of spindle waves at intervals of 6-10 s (Fig. 9A). Short trains of PB stimuli prevented the occurrence of spindles or readily blocked ongoing spindle sequences (Fig. 9A2, A3 and A4). It can be noted that PB stimulation produced very little hyperpolarization when the resting potential spontaneously hyperpolarized during interspindle lulls but that a large hyperpolarization followed PB stimu-

B. Hu et al.

A

I$r 0

-65mV

Fig. 4. Time course of the PB-evoked depolarizations in PG and LG relay neurons in urethane anaesthetlzed cats. Both cells, the PG cell in (A) and the LG neuron in (B), were recorded in the same experiment. Note the time relationship between the occurrence of fast rising EPSPs in the PG unit and LG cell firing. Traces Bl and B2 are representative of the shortest and longest latencies of discharges seen in LG relay neurons.

lation during the spindles (Fig. 9A4). In LG relay neurons, the blockage of spindle o~illations was characterized by the disappearance of the rhythmic inhibitory postsynaptic potentials (IPSPs) (Fig. 9B). It must be stressed that, under barbiturate anaesthesia, spindling was blocked without the occurrence of any tonic discharges in relay neurons. Usually, a single rebound spike terminated the aborted spindle sequence. The cholinergic nature of the peribrachial-induced ~yper~olarizat~on It is now well established th& most PB neurons projecting to the LG-PC complex are cholin. 10~27~28*32 and that ACh application on RE neurons ergic produces a slow muscarinic hyperpolarization by increasing the membrane conductance to K.20 Because of technical problems associated with longterm intracellular recording of RE cells in vivo, it has not been possible to record from the same cells before and after i.v. injection of ACh antagonists. However, three RE neurons were recorded within 1 h after i.v. injection of scopolamine (1 mg/kg). In those three cases, the PB-induced hy~~ola~~tion was not observed. Instead, the early depolarization was more prominent and could generate burst discharges. Although the number of neurons recorded intracellularly after scopolamine is small, this result contrasted polarizing

with the constant presence of the hyperresponse in every cell in the absence of this

choiinergic antagonist. The example shown in Fig. 10 is from a rostra1 RE neuron recorded under barbiturate anaesthesia. After scopolamine injection, a prolonged burst discharge followed PB stimulation and this long-lasting depolarization was reduced to a sub-threshold EPSP by h~rpola~~tion of the cell membrane. The muscarinic nature of the PB-evoked inhibi~on of PG cells was also demonstrated by iontophoretic applications of ACh antagonists. The inhibition of OX-evoked burst discharges in PG neurons by PB stimulation was abolished by sco~lamine ionto-

33 2

1 0

LATENCY OF PE-EVOKED EXCITATIUNS(ms1

Fig. 5. Distribution of latencies of the PB-evoked early EPSP in PG neurons (blank bars) and of PB-evoked earliest spike discharges in LG relay cells (filled bars). Data were taken from 13 PG neurons and from 13 relay cells.

The blockage of thalamic oscillations

A

B I -62mV P

2 -2nA

100

ms

Fig. 6. Effect of membrane potential on the PB-evoked hyperpolarization in PG neurons. Cell in (A) was recorded in a pentobarbital (Nembutal) anaesthetized cat and cell in (B) was recorded in a urethane anaesthetized preparation. In (Al), the cell displayed a depolarizing shift of its membrane potential during a spindle sequence. The membrane potential was hyperpolarized to - 70 mV by a long PB train (Al). During the interspindle lull (A2), when the membrane potential was already at -7OmV, no net hyperpolarization resulted after the PB shocks. Note the plateau of the PB-evoked depolarization during the train. In the second example (traces B) the PB-evoked hyperpolarization was greatly reduced by the injection of an inward current. The amplitude of the depolarizing events was however enhanced. phoresis (Fig. ll), while the OX-evoked discharges remained untouched. This test was carried out in five PG cells with similar results. On the other hand, application of the nicotinic blocker hexamethonium did not produce any significant changes (N = 6).

DISCUSSION Peribrachial stimulation produces two different and direct effects on RE neurons: a rapid depolarization followed by a prominent long-lasting

Fig. 7. Change in membrane resistance induced by PB stimulation in a PG neuron. This cell was recorded in a non-anaesthetized deafferented cat. Current pulses intensity: 2 nA. In the right-hand part of the trace, a sustained hyperpolarizing current was injected to estimate the amount of anomalous rectification in the conductance change observed during the response.. Note that, in spite of the anomalous rectification, the drop in membrane resistance during the response was of the order of 50%.

B. HU et al.

8

2

2

-0SmV

Y 2

d o’x Fig. 8. Decreased responsiveness of RE neurons after a conditioning PB stimulation. Recording performed in a PG cell under urethane anaesthesia. The control response to OX stimulation is shown in (Al). The first part of the response is expanded in (A2). In (B) and (C), the same OX stimulus was delivered at different intervals after a short PB train. The respective responses are depicted in (B2) and (C2). Note that the cell responsiveness was already decreased in (B) during the early depolarization. hyperpolarization. Our results show that the hyperpolarization is generated by a large conductance increase, presumably to K ions. They further suggest that a muscarinic mechanism is involved. This hyperpolarizing response renders RE cells virtually transparent to synaptic currents and prevents the activation of intrinsic inward currents that underlie the genesis of spindle oscillations.

Origin and nature of the early depolarization Previous studies bearing on the effect of MRF stimulation on RE neurons have been carried out by means of extracellular recordings in anaesthetized*~‘*~” or chronic animals.30 In barbiturized no early excitation has been reported preparations?” but such an excitation has been observed in chronic cats” and in urethane-anaesthetized rats.18 However,

in the latter cases, it has not been clearly demonstrated that this excitation was direct and did not result from a parallel activation of thalamic relay neurons either by PB afferents or by co-stimulation of specific pre-thalamic pathways (medial lemniscus, spinothalamic pathway or brachium conjunctivum afferents). In the present study, these possible indirect effects have been precluded by barbiturate anaesthesia, which renders thalamic relay neurons completely unresponsive to PB stimulatioqn and by reserpine treatment, which prevents monoaminergic effects. Besides PB and monoaminergic afferents, no other brainstem inputs are known to project to PG cells. It is then concluded that PG cells and possibly all RE neurons receive a direct excitatory input from the PB area. The latency of this EPSP (7-10 ms) agrees well with the latencies of antidromic invasions of PB neurons to PG microstimulation in cats.’ 3

The blockage of thalamic oscillations

PB

PB

500 ms

-63 mV

-60mV

500 ms P.B

Fig. 9. Blockage of spindle oscillations in thalamic neurons by PB stimulation. The PG cell in (A) and the LG relay neuron in (B) were both recorded in a non-anaesthetized deafferented cat. The rhythmic occurrence of spontaneous spindle sequences in (Al) and (AZ) was prevented by stimulation of the PB area. Traces (Al) and (A2) are continuous. An expanded spindling sequence is shown in (A3) and in (A4) another sequence was aborted by PB stimulation. Traces (Bl) and (BZ) show respectively a complete spindle sequence in an LG relay neuron and a spindle sequence abbreviated by PB stimulation.

Two possibilities are to be considered for the genesis of this depolarization. (1) Some afferents of PB origin may release a still unidentified excitatory neurotransmitter. This transmitter may be present in a subset of non-cholinergic PB neurons or may be co-localized with ACh in PB cells. (2) The early excitation may also be cholinergic but depend upon the activation of nicotinic or muscarinic, scopolaminBinsensitive receptors on RE neurons. In this latter case, receptors mediating the early depolar-

ization and the following hyperpolarization may be postsynaptic to a unique set of cholinergic PB afferents or postsynaptic to different subsets of cholinergic PB elements. On the basis of microstimulations performed in cats, it was suggested that PB afferents to the PG nucleus arose from different subsets of PB cells having different terminal arborizations in the LG-PG complex.’ However, it is not known if these different terminal arbors were all issued from cholinergic cells.

B. Hu et al.

demonstrated a muscarinic hyperpolarization of PG cells following ACh applications. However, nicotinic receptors have been mapped within the RE nuclear compleff of rats’ and monkeys” and the failure to observe a nicotinic response in slices may be ascribed to the mode of ACh application leading to a rapid desensitization of the response. The synaptic mechanism responsible for the early PB-evoked depolarization in RE neurons as well as the possible depolarizing action of ACh on these cells remains to be elucidated. Origin and nature of the hyperpolarization

of PB-induced hyperpolarization in a after i.v. injection of scopolamine (1 mg/kg). After scopolamine, the PB stimulus triggered a prolonged burst (A) which was reduced to a subthreshold EPSP by hyperpolarizing the cell membrane (B, C). Recordings made in a barbiturized cat. Fig.

10. Absence

rostra1 RE neuron

Until now, only one study has reported that RE neurons might be depolarized by ACh iontophoresis and this study was performed in rats.‘* In vitro recordings in guinea-pig thalamic slices2’ have only

The long-lasting hyperpolarization triggered in RE cells by PB stimulation is consistent with the depression of RE cells’ excitability observed extracellularly after MRF stimulation.2*‘2 Its cholinergic nature now appears to be well established. The inhibitory effects of MRF stimulation or ACh application are similarly suppressed by muscarinic antagonists.12 Furthermore, ACh application in vitro was reported to hyperpolarize RE cells by increasing K conductance.20 An increase in K conductance is also the most likely mechanism underlying the hyperpolarization observed in the present study. Although the PB area was recognized as the most effective site for inhibiting PG neurons, it was found that electrical stimulation of extensive brainstem areas that have no direct connections with the RE Recovery

Scopolamine

Control

.

* 100 ms

B

81

seronds

A PB * ox

seconds

seconds

Fig. 11. Blockage of PB-evoked inhibition in PG neurons by iontophoretic application of scopolamine. (A, upper row): effect of scopolamine application (70 nA for 60 s) on OX-evoked burst discharges in a PG cell; (second row): a similar application of scopolamine on the same cell prevented the PB-evoked inhibition of burst discharges triggered by OX stimulation. Peristimulus histograms in (B) summarize the results of scopolamine application on the conditioning testing procedure described above (bin size = 50 ms; eight trials).

The blockage of thalamic oscillations

nucleus could also inhibit PG cells2 This apparent ubiquity of inhibitory sites deserves some comments. Firstly, cells of the PB region are likely the targets of numerous cellular groups disseminated in the lower brainstem. Electrical stimulation of various brainstem areas may then be expected to excite PB neurons and inhibit disynaptically PG cells. Secondly, PG cell inhibition could also have resulted from activation of serotonergic cells of the dorsal raphe. In the present study, however, RE cell hyperpolarization does not appear to be serotonergic since it was observed after amine depletion and it was prevented by scopolamine. In our restricted experimental conditions, the inhibition of PG neurons could only result from the activation of the PB input. This does not necessarily imply that every IPSP evoked in PG cells by PB stimulation was monosynaptic since PB neurons could have been activated indirectly by the firing of local axonal branches presynaptic to PB cells. Anyhow, the final effect on PG neurons had to be mediated by PB afferents. The blockage

of thalamic

oscillations

The RE nuclear complex has been identified as the generator of thalamic spindle oscillations.3’ The above set of data clearly shows that these oscillations are blocked at the very site where they originated. It must be stressed that the PB-induced blockage of spindle waves could also be obtained in deeply barbiturized animals when thalamic relay neurons are completely unresponsive to PB stimulationi Under this anaesthetic condition, the suppression of rhythmic IPSPs in relay cells by PB stimulation abolishes burst discharges and leaves these neurons in a state of low excitability where neither bursts nor tonic discharges occur. The passage from an osillatory to a tonic mode of firing would then require additional mechanisms.

11

In the cat, it has been proposed that the synchrony of spindle oscillations involved the cooperativity of RE cells’ intrinsic membrane properties with a dendrodendritic synaptic mechanism.“~U This model implies a decoupling mechanism in order to stop oscillations in the RE network. The decoupling mechanism appears to be a large drop in membrane resistance that shunts very effectively inward currents which set off these oscillations. The large drop in membrane resistance produced by ACh renders RE cells very leaky and almost transparent to synaptic currents. This was indeed observed in urethane anaesthetized or deafferented animals where the tonic barrage of EPSPs resulting from the activation of relay cells by PB stimuli failed to induce any firing in PG neurons. However, it is very likely that in a normal preparation where the retinal and cortical inputs are intact, these additional excitatory drives would contribute to induce tonic firing in PG cells. Given the high degree of convergence of corticothalamic and thalamocortical fibres on RE cells, they would suffer from an enormous depolarizing pressure on arousal when cortical and thalamic neurons start firing tonically. The PB cholinergic input appears then as one of the systems that could dampen

the excitability of RE neurons. The somatic location of choline&c synapses on PG cells’ fits irrwell with such a dampening function. Other systems (especially the serotonergic input) may also act similarly and the present demonstration that spindle oscillations are blocked by PB afferents does not necessarily imply that this is the only way by which spindling can be blocked. wish to thank P. G&u&e and D. Drolet for their technical assistance. This work was supAcknowledgements-We

ported by the Canadian Medical MT-5877 and MT-3689.

Research

Council,

Grants

REFERENCES 1. Ahlsen G. (1984) Brain stem neurones with differential projection to functional subregions of the dorsal lateral geniculate complex in the cat. Neuroscience 12, 817-838. 2. Ahlsen G., Lindstrom S. and Lo F. S. (1984) Inhibition from the brainstem of inhibitory intemeurones in the cat’s dorsal lateral geniculate nucleus. J. Physiol., Lend. 347, 59349. 3. Ahlsen G. and Lo F. S. (1982) Projection of brainstem neurones to the perigeniculate nucleus and the lateral geniculate nucleus in the cat. Bruin Res. 238, 433-438. 4. Ben-Ari Y., Dingledine R., Kanazawa I. and Kelly J. S. (1976) Inhibitory effects of acetylcholine on neurones in the feline nucleus reticularis thalami. J. Physiol., Land. 261, 647-671. 5. Berman A. L. (1968) The Brainstem of the Cut. University of Wisconsin Press, Madison, WI. 6. Bremer F. (1935) Cerveau isole et physiologie du sommeil. C.r. SPanc. Sot. Biol. 118, 1235-1241. 7. Carlsson A. (1966) Drugs which block the storage of 5-hydroxytryptamine and related amines. In Handbook of Experimental Pharmacology (eds Eichler 0. and Farah A.), pp. 529592. Springer, New York. 8. Clarke P. B. S., Schwartz R. D., Paul S. M., Pert C. B. and Pert A. (1985) Nicotinic binding in rat brain: autoradiographic comparison of [3H]acetylcholine, [3H]nicotine and [‘251]-a-bungarotoxin. J. Neurosci. 5, 1307-1315. 9. De Lima A. D., Montero V. M. and Singer W. (1985) The cholinergic innervation of the visual thalamus: an EM immunocytochemical study. Expi Bruin Res. 59, 206212. 10. De Lima A. D. and Singer W. (1987) The brainstem projection to the lateral geniculate nucleus in the cat. 1. camp. Neural. 259. 92-l 2 1. 11. Desch&nes M., Madariaga A. and Steriade M. (1985) Dendrodendritic synapses in the cat reticularis thalami nucleus: a structural basis for thalamic spindles synchronization. Bruin Res. 334, 165-168. 12. Dingledine D. and Kelly J. S. (1977) Brain stem stimulation and the acetylcholine-evoked inhibition of neurons in the feline nucleus reticularis thalami. J. Physiol., Land. 271, 135-154.

12

B. HU et al.

13. Eysel U. T., Pape H. C. and Van Schayck (1986) Excitatory and differential disinhibitory actions of acetylcholine in the lateral geniculate nucleus of cat. J. Physiol., Lond. 370, 233-254. ‘4. Godfraind J. M. (1978) Acetylcholine and somatically evoked inhibition on perigeniculate neurons in the cat. J. Pharmacol. 63, 295-302. IS. Hu B., Steriade M. and Deschines M. (1989) The effects of brainstem peribrachial stimulation on neurons of the lateral geniculate nucleus. Neuroscience 31, 13-24. Hu B., Steriade M. and Deschenes M. (1989) The cellular mechanism of thalamic ponto-geniculo-occipital waves. Neuroscience 31, 25-35. Jones E. G. (1983) The thalamus. In Chemical Neuroanatomy (ed. Emson P. C.), pp. 257-293. Raven Press, New York. Kayama Y., Sumitomo I. and Ogawa T. (1986) Does the ascending cholinergic projection inhibit or excite neurons in the rat thalamic reticular nucleus? J. Neurophysiol. 56, 1310-1320. Livingstone M. S. and Hubel D. H. (1981) Effects of sleep and arousal on the processing of visual information in the cat. Nature 291, 554561. McCormick D. A. and Prince D. A. (1986) ACh induces burst firing in thalamic reticular neurons by activating a K conductance. Nature 319, 4022405. Morrison J. H. and Foote S. L. (1986) Noradrenergic and serotonergic innervation of cortical, thalamic and tectal visual structures in Old and New World monkey. J. camp. Neural. 243, 117-138. Mulle C., Madariaga A. and Deschenes M. (1986) Morphology and electrophysiological properties of reticularis thalami neurons in the cat: in uiuo study of a thalamic pacemaker. J. Neurosci. 6, 21342145. Pare D., Smith Y., Parent A. and Steriade M. (1988) Projections of brainstem core cholinergic and non-cholinergic neurons of cat to intralaminar and reticular thalamic nuclei. Neuroscience 25, 69-86. Sillito A. M., Kemp J. A. and Berardi N. (1983) The cholinergic influence on the functions of the cat dorsal lateral geniculate nucleus. Brain Res. 280, 299-307. Singer W. (1973) The effect of mesencephalic reticular stimulation on intracellular potentials of cat lateral geniculate neurons. Brain Res. 61, 3554. Singer W. (1977) Control of thalamic transmission by corticofugal and ascending reticular pathways in the visual system. Physiol. Rev. 57, 386420. Smith Y., Pare D., Deschenes M., Parent A. and Steriade M. (1988) Cholinergic and non-cholinergic projections from the upper brainstem core to the visual thalamus in the cat. Expl Brain Res. 70, 166180. Sofroniew M. W., Priestly J. V., Consolazione A., Eckenstein F. and Cue110 A. C. (1985) Cholinergic projections from the brainstem and pons to the thalamus in the rat identified by combined retrograde tracing and choline acetyltransferase immunohistochemistry. Brain Res. 329, 213-223. Steriade M. and Deschines M. (1984) The thalamus as a neuronal oscillator. Brain Res. Rev. 8, l-63. Steriade M., Domich L. and Oakson G. (1986) Reticularis thalami neurons revisited: activity changes during shifts in states of vigilance. J. Neurosci. 6, 68-81. Steriade M., Domich L., Oakson G. and Deschenes M. (1987) The deafferented reticular thalamic nucleus generates spindle rhythmicity. J. Neurophysiol. 57, 26(t273. Woolf N. J. and Butcher L. L. (1986) Cholinergic systems in the rat brain. III. Projections from the pontomesencephalic tegmentum to the thalamus, tectum, basal ganglia and basal forebrain. Brain Res. Bull. 16, 603-637. Yoshida M., Sasa M. and Takaori S. (1984) Serotonin-mediated inhibition from dorsal raphe nucleus of neurons in dorsal lateral geniculate and thalamic reticular nuclei. Brain Res. 290, 955105 (Accepted 6 December

1988)