Plasticity of sensory responses of locus coeruleus neurons in the behaving rat: implications for cognition

The gating and tuning actions of noradrenaline (NA) at post-synaptic sites have been highly suggestive of an important role for the locus coeruleus (LC) in attention, learning and memory. By recording the activity of single units in the LC in behaving rats in a strictly controlled conditioning paradigm, direct evidence was provided that this nucleus is engaged during specific aspects of learning. The neuronal response to a discrete sensory stimulus was monitored as a function of the changing significance of the stimulus i.e., when it was novel, during habituation, associative learning, reversal and extinction. Both appetitive and aversive paradigms were used. We consistently observed differential conditioned responding with food reinforcement, while when footshock reinforcement was used, there was an increase in response to both CS+ and CS–. In both paradigms, the LC response disappeared when the conditioning was expressed at a behavioral level, to reappear vigorously as soon as the stimulus reinforcement contingencies were changed, i.e., during reversal or extinction. These results suggest that the LC does not mediate specific sensory or associative information necessary for ongoing performance but shows remarkable plasticity of sensory responding as a function of changing cognitive significance of the stimulus.

Introduction

In spite of the abundance of electrophysiological evidence that noradrenaline (NA) is a potent neuromodulator, involved in enhancing and gating signals (Madar and Segal, 1980; Rogawski and Agahjanian, 1980; Waterhouse and Woodward, 1980; Waterhouse et al., 1988; Kayama et al., 1982) and promoting synaptic plasticity throughout the nervous system (Neuman and Harley, 1983; Stanton and Sarvey, 1986; Hopkins and Johnston, 1988; Dahl and Sarvey, 1989; see Harley, 1987 for review), the functional significance of noradrenergic action must ultimately be studied at a behavioral level. There the evidence for an essential role for NA in learning and memory is scarce and far from compelling. In awake, behaving animals the first demonstration of the neuromodulatory properties of NA was in a study showing that iontophoretic application of NA to the auditory cortex of a monkey enhanced signal/noise in response to the vocalization of a conspecific (Foote et al., 1975). In the rat, electrical stimulation of the locus coeruleus (LC) selectively enhanced hippocampal neuronal responses to tones associated with food, while increasing inhibition to the same tones before they had acquired a significance (Segal and Bloom, 1976). While these two studies demonstrated in awake animals that NA could modify neuronal responses to discrete stimuli, especially if they had a biological significance for the organism, they do not provide evidence that this modulation actually improves cognitive performance. The obvious strategy in addressing this question lies in enhancing or inhibiting the release of NA and evaluating its effect on evoked potentials at postsynaptic sites and at the same time on learning or memory. The localization of the entire population of cortically projecting NA neurons in one small pontine nucleus LC simplifies the first part of the problem to some extent.

Neurotoxic or electrolytic lesion to the LC or its forebrain projection fiber system, the dorsal bundle, depletes the forebrain of NA. Most behavioral investigators have addressed the question of the role of NA in learning and memory by using specific neurotoxic lesions to the coeruleocortical system. This has generated substantial literature which has been thoughtfully reviewed on several occasions (Robbins, 1984; Robbins and Everitt, 1987). For the most part these studies have failed to provide unequivocal support for the role of this system in attention, learning and memory (Pisa and Fibiger, 1983; Robbins, 1984; but see also Pineda et al., 1989; Selden et al., 1990a,b). Although deficits do exist, animals with very little NA in the forebrain do well on many challenging cognitive tasks (Sara et al., 1984; Sara, 1985a,b; Robbins and Everitt, 1987).

Electrical or pharmacological stimulation of neurons within the LC increases release of NA in the forebrain (Tanaka, et al., 1976). Since the activity of the cells of the LC is regulated by inhibitory autoreceptors, agonists and antagonists of these receptors are useful pharmacological tools to excite or inhibit the system and can be used during behavioral tasks. Systemic injections of idazoxan, the α₂-adrenoceptor antagonist, increases firing of LC cells by about 80% at 2 mg/kg (Sara and Devauges, 1989). We used idazoxan to assess the effects of stimulation of the NA system in complex behavioral situations. Idazoxan is particularly effective in facilitating performance in tasks which require shifts in attention between sensory modalities, accompanied by changes in behavioral strategies. The same stimulation enhances behavioral response to novelty and change in response-reinforcement contingencies (Sara et al., 1988; Devauges and Sara, 1990).

The relatively few behavioral studies using electrical stimulation of the LC have shown facilitation of successive discrimination reversals (Segal and Edelson, 1978) and of memory retrieval after a long retention interval (Sara and Devauges, 1988). Stimulation in the region of the dorsal bundle produced similar memory facilitation (Sara et al., 1980; Dekeyne et al., 1987). Recent evidence suggests that the effect is mediated through a β-receptor, since the animals pretreated with the β-receptor antagonist propranolol are not facilitated by the LC stimulation (Devauges and Sara, 1991).

While these studies show that LC stimulation is most effective in facilitating cognition, if it is applied when the animal has to adapt to novelty or change in significance of a stimulus, to further determine the role of LC in learning it is important to know when in the learning process the LC is active and engaged. We suggest that the stimulation is effective because it enhances a physiological event, i.e., activation of LC cells in a particular cognitive context. Therefore the LC should be preferentially engaged during learning when stimuli are novel or when there is a change in the significance of the stimulus. We recorded activity of LC cells in response to novel to-be-conditioned stimuli, habituation, and associative conditioning, reversal and extinction, to ascertain when in the learning process the LC is actually active and engaged. A Pavlovian differential conditioning paradigm was used in order to precisely control the conditioned stimuli (CSs) and their relation to the reinforcement and to establish a strict temporal relation between the neuronal and the behavioral response. Both appetitive and aversive versions of the paradigm were used, because Rasmussen and Jacobs (1986) reported that, in cats, there was no conditioning to a stimulus associated with food reinforcement, but good differential responding in a conditioned emotional response (CER) protocol, concluding that LC is involved in learning exclusively when the level of stress and anxiety is high.

In the aversive conditioning studies, the CSs were preceded by, and contained within, a flashing light which served as an experimentally controlled context. Interest in the LC response to context lies in the fact that previous experiments have shown that contextual cues can serve as very effective memory facilitators (Deweer et al., 1980). We have suggested that the mechanism of action might lie in a conditioned response of LC neurons to the context associated with the prior learning (Sara, 1985b). Lesion studies from two different laboratories have led to diametrically opposite conclusions concerning the role of the coeruleocortical system in mediating context effects (Archer et al., 1982; Mohammed et al., 1986; Selden et al., 1990).