Physiological and Anatomical Determinants of Locus Coeruleus Discharge:

Rita J. Valentino and Gary S. Aston-Jones

The locus coeruleus (LC) noradrenergic system has been viewed as a broadly projecting system with nonspecific functions. This view of the LC was based in part on early anatomic findings which indicated that LC efferent projections were widespread, that single LC neurons send divergent projections to target cells of very different function, and that the nucleus was not topographically organized in a manner that could confer specificity. Similarly, early studies of LC afferents suggested a diversity nearly as great as its efferent projections, implying a lack of specificity in the types of stimuli that could affect LC activity and function. This concept of the LC as a "nonspecific" system was further supported by initial electrophysiological studies demonstrating that the effect of norepinephrine (NE) application or LC stimulation on postsynaptic targets was generally inhibition, regardless of the postsynaptic site studied. The last decade has brought advances in anatomic, physiological, and behavioral technology that has allowed for a more sophisticated analysis of the input and output relations of the LC–NE system. This chapter reviews recent studies that are refining our view of the LC–NE system as one that shows specifity in its response to various stimuli and in its effects on different target neurons (see Pharmacology and Physiology of Central Noradrenergic Systems and Central Norepinephrine Neurons and Behavior for other aspects).

The anatomical and physiological characteristics of the LC have been most studied in the rat, cat, and primate. The most obvious species difference is that LC neurons are homogeneously noradrenergic in rat and primate LC, although other neurotransmitters may be colocalized with norepinephrine in these neurons (see Coexisting Neurotransmitters in Central Noradrenergic Neurons). In contrast, NE-containing neurons are interspersed with non-noradrenergic neurons in the LC of other species including cat. In order to limit discussion here to LC-noradrenergic neurons, most of the data reviewed will represent rat or primate studies, except where noted.

The LC was originally thought to receive neurochemically and regionally diverse inputs. This was based on the numerous different neurochemically identified fibers in the LC area. The localization of binding sites for many of these neurotransmitters in the LC and their reported physiological effects on LC activity supported the notion that the LC received diverse afferent input. summarizes the putative neurotransmitters that have been identified immunohistochemically in fibers in the LC region, neurotransmitter binding sites in the LC, and neurochemicals that directly affect LC neurons. However, these findings can be misleading and do not by themselves establish a particular neurotransmitter role for a substance in the LC. For example, the localization of neurotransmitter receptors does not necessarily match localization of nerve terminals containing the neurotransmitter. Additionally, the finding that a particular neurochemical affects LC discharge does not confirm that this is a physiologically relevant phenomena as opposed to a pharmacologic phenomena. Finally, the presence of neurotransmitter immunoreactivity in fibers in the LC region without ultrastructural confirmation of synapses is not convincing evidence for neurotransmission (e.g., fibers may be passing through to terminate elsewhere). Thus far, the most convincing evidence of functionally relevant input to the LC has come from a combination of (a) studies using tract tracers and immunohistochemistry to neurochemically identify and confirm afferents to the LC and (b) studies involving stimulation of putative afferents with simultaneous LC recording to physiologically confirm these inputs (see also Cytology and Circuitry and Electron Microscopy of Central Dopamine Systems).

Early retrograde tract-tracing studies of afferent input to LC which utilized the horseradish peroxidase (HRP)–diaminobenzidine (DAB) technique reported that more than 30 nuclei were labeled after injection of the tracer into the LC (6). Such findings suggested that the LC could be influenced by many structures and supported prevailing views that this was a general nonspecific system. The development in the last decade of tract tracers that could produce more focal injections has refined the number of nuclei projecting to the LC. These tracers include wheat germ agglutinin (WGA)-conjugated HRP, WGA-apoHRP coupled to colloidal gold (WGA-apoHRP-Au), the beta subunit of cholera toxin (CTb), and the fluorescent tracer, Fluoro-Gold. Using these tracers it has been demonstrated that the LC receives a far more restricted set of inputs than was originally believed (6). Two medullary nuclei, the nucleus paragigantocellularis (PGi) and nucleus prepositus hypoglossi (PrH), are prominent afferents to LC based on recent retrograde tract-tracing studies and electrophysiological evidence demonstrating that these nuclei directly impact on LC (see below). The PGi, localized in the rostral ventrolateral medulla, is of interest as an LC afferent because of its role in somatic, autonomic, and visceral functions. Less is known about the function of cells in the PrH. However, oculomotor function has been attributed to the PrH, suggesting that this input to LC may integrate visual shifts in attention with cognitive shifts. Other areas that appear to send projections into the LC include the dorsal cap of the paraventricular nucleus of the hypothalamus, the intermediate zone of the spinal cord, the Kolliker-Fuse, periaqueductal gray, the lateral hypothalamus, and the preoptic area. However, it should be noted that fewer projections to the LC are found from these areas compared to PGi or PrH (6), and that these areas require further study to confirm direct projections to the LC proper. The involvement of some of these nuclei in stress and autonomic activity suggest possible routes whereby this information may reach the LC.

Anterograde tracing studies have confirmed that the PGi, the PrH, the ventrolateral periaqueductal gray, the Kolliker-Fuse, and the medial preoptic area project to the LC (6). However, dendrites of non-LC neurons extend into the LC nucleus, so that ultrastructural and electrophysiological studies are needed to confirm inputs to noradrenergic LC neurons; such electrophysiologic studies have confirmed functional inputs from the PGi and PrH (see below). In contrast, anterograde labeling from injections into areas previously thought to innervate the LC—including the VTA, the dorsal horn of the spinal cord, the rostral solitary nucleus, and the prefrontal cortex—was not apparent in the LC but was observed in structures surrounding LC (6). These findings indicate that these structures may innervate distal dendrites of LC neurons that extend into pericoerulear areas or that they may communicate with the LC only indirectly, perhaps by innervating pericoerulear regions. Anterograde labeling from amygdala, which was initially reported as a prominent LC afferent, was meager and was confined to the extreme rostral pole of the LC where NE neurons are interdigitated with non-NE neurons (6). Interestingly, the projections from the periaqueductal gray and from the preoptic nucleus terminate, for the most part, in pericoerulear regions that contain LC dendrites. Ultrastructural analysis of this region is necessary to evaluate the functional innervation of these dendrites. It is also possible that neurons in close proximity to the LC (i.e., Barrington's nucleus, lateral dorsal tegmentum) provide afferent input to the LC. However, it is difficult to determine whether these nuclei project to the LC using the available tracers because the close proximity of these nuclei to the LC prohibits any distinction between spread of tracer and actual retrograde labeling.

Many of the neurotransmitters reported to innervate the LC based on immunohistochemical localization of fibers in the LC are found in neurons of two major LC afferents, the PGi and PrH. These include PMNT (a marker for adrenaline) (31), excitatory amino acids (18), enkephalin (6, 37), corticotropin-releasing factor (CRF) (66), substance P (38), serotonin (5-HT) (50), and gamma-aminobutyric acid (GABA) (46). Double labeling studies combining retrograde tract tracing and immunohistochemistry for glutaminase reveal putative glutaminergic neurons in PGi that project to the LC (6). This input has been confirmed physiologically and has been shown to mediate the well-characterized activation of LC neurons by footshock, as well as LC activation associated with opiate withdrawal (see below). Adrenergic innervation of the LC derives primarily from the PGi, where more than 20% of LC afferent neurons are PMNT-immunoreactive, although a small percentage of adrenergic LC innervation may also derive from PrH (6). Ultrastructural studies revealed numerous PNMT-immunoreactive varicosities making conventional symmetric and asymmetric synapses onto LC dendrites, further supporting a neurotransmitter role for adrenaline in the LC (44). Indeed, adrenaline may mediate a part of the postactivation inhibition associated with LC responses to footshock (see below). The PrH contains a large percentage (>40%) of LC projecting neurons that are GAD- or GABA-immunoreactive (6). Other nuclei, including the PGi, contain GAD- and GABA-immunoreactive neurons that project to the LC (46). The LC receives a rich enkephalinergic innervation (6), and ultrastructural studies indicate that enkephalin-immunoreactive terminals synapse on LC neurons (51), suggesting that the LC is an important target for opioid neurotransmission in the brain. This is consistent with the potent effects of opiates on LC neurons (see Signal Transduction Pathways for Catecholamine Receptors, Pharmacology and Physiology of Central Noradrenergic Systems, Neuropharmacology of Endogenous Opioid Peptides, Intracellular Messenger Pathways as Mediators of Neural Plasticity, and Opioids). Recent double labeling studies revealed that a surprisingly large percentage of the LC projecting neurons in both PGi and PrH are immunoreactive for enkephalin (6).

The CRF innervation of the LC is noteworthy because the LC–NE system is sensitive to different stressors (see below). CRF was initially characterized as the hypothalamic hormone responsible for releasing adrenocorticotropin in response to stress and, therefore, as the initiator of the endocrine cascade of the stress response (65). Shortly after its characterization, numerous anatomic and behavioral studies suggested that CRF also acted as a neurotransmitter outside of the hypothalamic–pituitary axis to initiate other aspects of stress (66). The criteria for a neurotransmitter role for CRF have been most rigorously tested in the LC, and the results of these studies indicate that CRF acts as a neurotransmitter to activate the LC during hypotensive stress (see below). CRF input to LC appears to originate predominantly from the PGi (66). Interestingly, hypotensive stress also activates PGi neurons. Additionally, some CRF-immunoreactive neurons in the dorsal cap of the paraventricular nucleus are retrogradely labeled from the LC, and there are numerous pericoerulear CRF-containing neurons along the lateral border of the LC and rostromedial to the LC in Barrington's nucleus that may be a source of CRF fibers in LC (66).

Ultrastructural studies indicate that afferent terminations in the LC synapse on dendrites ranging between 0.5 and 2.5 mm in diameter and onto spine-like appendages on dendrites and cell bodies (26). Most of these synapses fall into four categories: (i) synapses with small round, densely packed vesicles (41%); (ii) those with large rounded vesicles (20%); (iii) synapses with large flattened vesicles (23%); and (iv) those with numerous small flattened vesicles (11%). The remaining 5% had mixtures of these and/or contained dense-core vesicles. Interestingly, there was no apparent segregation of symmetric and asymmetric junctions. If these different morphologic characteristics represent different afferents, this could suggest that different inputs converge in a common spatial distribution and that the net effect of inputs to an LC neuron will be an integration of these converging inputs.

Thus, afferent input to LC appears to be more restricted than previously believed, although it is neurochemically diverse. The physiological relevance of the LC afferents proposed from anatomic studies is described below. Their identity has been useful in understanding how various stimuli communicate with the LC.

Numerous immunohistochemical studies have described NE innervation of the central nervous system (CNS); and in combination with LC lesions or tract tracing, the distribution of the massive divergent LC projection system has been described (see ref. 17 for review). An important finding revealed by these studies is that the LC is the sole source of norepinephrine in the forebrain, as well as a major contributor to other regions. The detailed distribution will not be described here except where it is applicable to the functional correlates of the LC described below.

That the LC is important in sensory processing is suggested by its innervation of the spinal cord and sensory nuclei in brainstem and pons. The target of spinally projecting LC neurons has been the subject of controversy, because earlier studies suggested that this was the ventral horn, implicating the LC in motor function (17). However, several more recent studies convincingly demonstrate a bilateral innervation from the LC to the dorsal horn, terminating particularly in superficial layers such as the substantia gelatinosa (20). There has been general agreement that the LC does not innervate intermediolateral column of the thoracic cord (20). This pattern of innervation supports a role for the LC in analgesia and sensory information processing, as opposed to motor or autonomic function. The pattern of LC innervation of brainstem and pontine nuclei are also consistent with a role in sensory information processing rather than autonomic function. Most LC innervation of the pons is to sensory and association nuclei, whereas autonomic nuclei receive norepinephrine innervation from other noradrenergic nuclei (19). It is noteworthy, for example, that the LC densely innervates the trigeminal nucleus—particularly pars caudalis, which is involved in sensory information from the face (19).

The LC is the sole source of NE in the cerebellum (17). Ultrastructural studies of this innervation reveal NE synapses onto Purkinje cell dendrites (17). These findings led to numerous electrophysiological and pharmacological studies of the effects of NE and LC stimulation on Purkinje cell activity which have served as a model for studying the postsynaptic impact of this system (see below).

In contrast to many brain areas, the hypothalamus receives only a minor noradrenergic innervation from LC that is localized in the medial part of the parvocellular paraventricular nucleus of the hypothalamus, a region containing neurons that project to the median eminence (17). This innervation suggests that the LC may modulate neuroendocrine function. Interestingly, although the LC receives a minor input from the paraventricular nucleus of the hypothalamus, cells that project to the LC are localized more dorsally, arguing against the possibility of reciprocal communications (66). The LC also projects to the preoptic nucleus and bed nucleus stria terminalis. Interestingly, the preoptic nucleus is a minor source of afferents to the LC, suggesting the possibility of reciprocal communications between the LC and this nucleus. The LC innervates thalamic sensory relay nuclei for the visual and somatosensory cortex, the lateral geniculate nucleus, and ventrobasal complex, respectively, and there is a dense NE innervation from the LC to the thalamic reticular nucleus which coordinates thalamocortical activity (17). The effects of LC stimulation or NE application on thalamic neuronal activity have been one of the more well-characterized postsynaptic effects of this system. These effects suggest a model whereby LC activation can produce a shift from drowsiness to the alert state, perhaps underlying the importance of the LC in arousal (see Pharmacology and Physiology of Central Noradrenergic Systems).

The LC is the sole source of NE in the hippocampus and neocortex, and the projection is predominantly (90%) unilateral (17). The pattern of LC innervation of cortex shows a distinct laminar distribution with little variation between different areas of cortex. The patterns of innervation of these structures and specific projection pathways from LC have been described in detail elsewhere (17).

Another LC target of interest is the olfactory bulb, which receives all NE innervation from the LC (58). The termination patterns suggest that the major target cell in this structure is the granule cell, a GABA interneuron that provides inhibitory feedback to the major output cells of the olfactory bulb, namely, the mitral and tufted cells.

An interesting characteristic of LC projections is their divergence, suggesting that changes in LC activity can influence neurons of diverse functions. This has been studied using multiple injections of retrograde tracers into the same structure but in different hemispheres or into different structures (17). The results of these studies indicate that a single LC neuron can project to different cortical hemispheres: to hippocampus and cortex, to thalamus and cortex, to thalamus and hippocampus, and to forebrain and spinal cord. This divergence supports the idea that changes in LC discharge can simultaneously impact on functionally diverse targets, and this could be one way in which the LC could coordinate the activity of multiple systems into a symptom complex as has been hypothesized to occur in opiate withdrawal or stress (see Signal Transduction Pathways for Catecholamine Receptors and Opioids).

In spite of the clear evidence for the divergence of LC projections, it is becoming recognized that there is some topographical organization of LC efferents, which implies a greater specificity in the effects of this system than was originally believed. Recent studies indicate that topographically organized LC neurons with specific projections may also be distinguished morphologically (17). This is perhaps most apparent for projections to spinal cord which appear to originate from large multipolar neurons in the ventral third of the LC (17). These neurons may also project to the cerebellum. Although this population overlaps spatially with LC cortically projecting neurons, it appears to be distinct. Anterior large multipolar neurons are thought to project primarily to the hypothalamus and perhaps the septum (17). Additionally, small round neurons in the ventral LC project to the hypothalamus. In the core, medium multipolar neurons project to multiple targets, including the neocortex, the hippocampus, the hypothalamus, the cerebellum, and the spinal cord. However, there appears to be a modest topographic organization within the core as well: Core cells in the dorsal two-thirds of the LC project to the neocortex, the hippocampus, the hypothalamus, and the cerebellum, but not to the spinal cord, and core cells in the ventral third project primarily to the spinal cord and the cerebellum, but not to the hippocampus (17). Fusiform cells on the dorsal edge of the LC are thought to project to the hippocampus (17). Waterhouse et al. (68) detailed the topography of cortically projecting LC neurons. In this study, cortically projecting LC neurons were localized in the caudal three-fifths of the dorsal ipsilateral LC nucleus. Within this region, groups of cells identified as projecting to occipital, sensorimotor, or frontal cortex were observed to form a dorsal-to-ventral gradient, whereas occipital-projecting neurons tended to be more caudal. The frontal cortex received innervation from LC neurons in both dorsal and ventral subdivisions of the nucleus. These findings differ with earlier studies concluding that cortically projecting LC neurons are randomly distributed in the nucleus with the exception of the ventral division. It is possible that LC neurons sharing similar morphology and topography within the nucleus may also share other characteristics (such as afferents or colocalization of other neurotransmitters) with NE. While further studies are needed to explore such possibilities, this would be one mechanism whereby specialization of LC function could be conferred.

Early reports using electron microscopic (EM) autoradiographic examination of NE terminals in the cerebral cortex concluded that most NE terminals did not make synaptic contacts, but rather existed as nonsynaptically arranged terminals to provide NE in a paracrine fashion to a local cortical area (11). In contrast, quantitative studies in hippocampus demonstrated that NE terminals formed specialized synaptic junctions with other neurons as frequently as did non-NE terminals (17). In addition, more recent studies using serial EM reconstruction of NE synapses in cerebral cortex identified with dopamine-b-hydroxylase (DBH) immunohistochemistry found that a great majority of NE terminals form conventional synaptic contact onto neuronal profiles (49). These studies are consistent with those of Olschowka et al. (48), who found that most DBH-immunoreactive terminals in areas of the diencephalon, cerebellum, and limbic cortex form axodendritic synapses characterized by specialized junctional appositions. Thus, considerable evidence exists to indicate that this system does use conventional synaptic transmission at many, if not all, of its terminations. It is also possible that both synaptic and paracrine modes of neurotransmission exist at the same or different NE terminals. Other ultrastructural studies reveal tyrosine hydroxylase terminals in apposition to astrocytic processes that stain for b-receptor antibody, suggesting that one mode of communication of LC with its targets may be via alterations in glial function (4). Future ultrastructural studies are necessary to confirm the mechanism by which the LC communicates with its numerous targets.

The anatomic demonstration of specific inputs to the LC is not sufficient to verify function; this requires physiological confirmation. Thus, electrical stimulation of sources of LC afferents should alter activity of LC neurons, and LC stimulation should antidromically activate neurons that project to the LC. These techniques have been used to confirm afferents from the PGi and PrH proposed by anatomic studies. Electrical stimulation of the LC was found to antidromically activate >20% of neurons recorded in the PGi and PrH, but few or no neurons in the nucleus tractus solitarius, contralateral LC, or lateral reticular nucleus (6). Interestingly, these studies suggested at least two physiologically distinct populations of LC afferents in PGi, based on differential latencies to antidromic stimulation of the LC.

Electrical stimulation of PGi produces two effects on LC neurons. An excitation of short latency which is sensitive to non-N-methyl-D-aspartate (non-NMDA) excitatory amino acid receptor antagonists is most often observed (6). This putative excitatory amino acid afferent appears to be important in LC activation by a number of stimuli. For example, the characteristic response of LC neurons to footshock or pressure is also prevented by excitatory amino acid antagonists that are selective for non-NMDA receptors and by reversible inactivation of the PGi with local lidocaine or GABA injections (6) (also see below). An inhibitory response to PGi stimulation has also been reported and is thought to be due to activation of adrenaline-containing C1 neurons because it is prevented by the a2 antagonist, idazoxan (6). These findings confirm the anatomic studies which indicate that the LC receives a major adrenergic input from the PGi.

In contrast to PGi stimulation, electrical stimulation of PrH primarily inhibited LC neurons (6). This inhibition was eliminated by GABA antagonists, but not by opiate or a2 receptor antagonists, implicating GABA in this response. This is consistent with immunohistochemical observations of GABA-containing neurons in the PrH.

Similar physiological evidence for a functionally substantive projection to LC from other proposed afferents has not been obtained to date. Electrical stimulation of the central nucleus of the amygdala, nucleus of the solitary tract, or medial prefrontal cortex does not consistently alter LC activity, although earlier anatomic studies implicated these regions as LC afferents (6). Stimulation of these regions more often resulted in synaptic activation of areas adjacent to the LC such as the parabrachial nucleus or Barrington's nucleus. These results confirm the more recent anatomic studies revealing projections to these pericoerulear regions but not to the LC nucleus proper.

Thus, electrophysiological and pharmacologic studies are beginning to confirm anatomic studies suggesting a relatively restricted input to LC. An important aspect of these studies is the neurochemical diversity of afferents to LC from a common region. These results suggest that specificity of LC responses to various stimuli may be conferred by different neurotransmitter inputs. This appears to be the case for different stressors, as discussed below.

The postsynaptic effects of NE application or LC stimulation on targets of this system are described elsewhere in this volume (see Foote and Aston-Jones) and will only be reviewed briefly here. Of interest to this chapter it is important to note that just as the LC–NE system initially appeared to be "nonselective" because of its anatomic characteristics, the initial physiological studies of the system supported this general view. Thus, early investigations into the postsynaptic effects of the LC–NE system demonstrated a generalized inhibition of target cells. This effect was initially characterized in detail on cerebellar Purkinje neurons and subsequently noted in hippocampus, cortex, thalamus, and hypothalamus, suggesting that a global effect of the LC–NE system was inhibition (see ref. 17 for review). However, later studies revealed more complex effects of NE application that were concentration-dependent, whereby low doses selectively enhanced the effects of afferent inputs (evoked activity) relative to basal or spontaneous discharge, while higher doses resulted in inhibition. These effects were observed in the cerebellum, the cortex, and the hypothalamus and led to the idea that LC activation increases the "signal-to-noise" ratio of activity in postsynaptic neurons (17). More support for this idea came from studies demonstrating that NE and LC stimulation potentiated the effects of both excitatory and inhibitory neurotransmitters on the same neuron (17). Taken together, these effects suggested that the LC–NE system functioned to increase processing of information about incoming sensory stimuli, as opposed to solely altering basal discharge rate.

Although comparable effects of LC activation are observed on different postsynaptic targets, the net effect on a particular neuron may depend on properties of the circuit in which the neuron functions. For example, one effect attributed to NE is gating of postsynaptic activity, whereby a cell which was previously unresponsive to a stimulus becomes responsive in the presence of NE. This effect was demonstrated for visually evoked responses of cerebellar Purkinje neurons in the parafloccular lobe (69). Another related effect of NE, observed in the visual cortex of the cat, is to refine receptive fields. NE application onto visual cortical neurons resulted in more sharply defined transitions between stimulus-induced inhibition and excitation; that is, it sharpened the receptive field (69). This effect would be predicted to enhance the ability to detect stimulus movement across receptive field boundaries. Another effect of NE and LC stimulation may be to alter patterns of neuronal firing. For example, in the thalamus NE can shift the pattern of neuronal activity from a "bursting" mode, which is associated with slow-wave sleep and drowsiness, to a single-spike firing mode, which is associated with (a) transmission of sensory stimuli to the cerebral cortex and (b) waking and attention. This pattern shift may, in part, underlie LC effects on arousal (see Pharmacology and Physiology of Central Noradrenergic Systems).

Although it is generally assumed that LC discharge rate is proportional to NE release in target areas, this relationship has yet to be systematically characterized. This is important because some hypothesized functions of the LC–NE system have been generated by integrating (a) studies that quantify LC discharge rate under different conditions and (b) studies characterizing the effects of NE on target cell activity. The assumption that increases in LC discharge produce sufficient NE release in targets to mimic the effects reported in postsynaptic studies underlies the integration of these two types of studies into a hypothesis of LC function. Unfortunately, this assumption can only be tested in studies that combine recordings of activity in both LC and target regions simultaneously during manipulation, and these studies are few (however, see Pharmacology and Physiology of Central Noradrenergic Systems). The link between LC discharge and target cell effect (i.e., NE release) has recently been examined in voltammetry and microdialysis studies. These studies have demonstrated that stimuli that are known to increase LC rate also increase NE levels in extracellular fluid or increase the NE signal as measured by voltammetry in postsynaptic targets (cortex, hippocampus, and thalamus) of the LC–NE system. These stimuli include footshock, restraint stress, electrical and chemical stimulation of the LC or dorsal noradrenergic bundle, and administration of a2 receptor antagonists (1, 8, 29). Interestingly, some studies demonstrate that NE release increases in a nonlinear manner with increasing frequencies of dorsal bundle stimulation, suggesting that during bursts of high-frequency activity, such as when LC discharge is evoked by phasic sensory stimuli, NE release per action potential is greater than when LC discharge is tonically elevated (8). Unfortunately, in the above neurochemical studies, LC discharge was not recorded simultaneously, so that the relationship between LC discharge and NE release still remains to be established. For example, it is not known to what extent, and for how long a duration, LC discharge must be elevated to produce increases in NE release in targets, and whether the amount of release is target-specific. An interesting finding of the neurochemical studies is that NE release in a particular region may be dependent on the mode of sensory stimulus that activates the LC. This was recently demonstrated in a study of the effects of visual stimulation on NE release (measured by voltammetry) in different cortical regions (41). This study demonstrated that NE release in monkey striate cortex exhibits an ocular dominance paralleling the ocular dominance for cortical neuron activation. Moreover, in the cat, visual stimuli that elicited NE release in visual cortex failed to do so in somatosensory cortex. A possible explanation for the local specificity of NE release is that it may be regulated presynaptically by cortical afferents activated by specific stimuli. This type of presynaptic heteroregulation of NE release could be a mechanism for conferring specificity on the LC–NE system within terminal areas.

The proposed functions attributed to the LC–NE system have been based on lesion studies, pharmacologic studies, and electrophysiological recordings from LC neurons under different conditions (see Pharmacology and Physiology of Central Noradrenergic Systems and Central Norepinephrine Neurons and Behavior). The putative roles of the LC in arousal, vigilance, attention, and learning are discussed in detail in other chapters in this volume and will not be reviewed here. This chapter focuses on putative functions derived from new knowledge of the input–output relations of the LC.

One characteristic of LC neurons that has implicated this nucleus in pain is that they are conspicuously activated by noxious stimuli. In waking animals, LC cells are activated by low-level stimuli of many modalities (described in ref. 17). However, these neurons are most reliably activated by either noxious or stressful stimuli. This fact is best illustrated by comparing LC sensory responsiveness in anesthetized versus unanesthetized animals.

The sensory sensitivity of LC neurons is markedly reduced under anesthesia. Indeed, reduced LC activity and responsiveness may be one of the hallmark actions of anesthetic treatments. For example, LC neurons are potently activated by a variety of stimuli in waking animals, including auditory, visual, and non-noxious somatosensory events. In contrast, these cells recorded under anesthesia do not respond to any of these stimuli, but are activated exclusively by noxious stimuli such as foot or tail pinch or sciatic nerve activation (17). As another example, the magnitude of LC activation by hypotensive stress is more than 10 times greater in unanesthetized rats than in anesthetized ones (66). Similarly, the effects of pharmacological agents on LC discharge are dependent on the state of anesthesia. Thus, the stress neurohormone, corticotropin-releasing factor (CRF), increases LC spontaneous discharge rate and is more potent and efficacious in unanesthetized rats than in anesthetized ones (66). In contrast, morphine, which inhibits LC spontaneous discharge rate, is much less potent and less efficacious in unanesthetized rats than in anesthetized rats (66).

As noted above, under anesthesia LC neurons lose their responsiveness to nearly all low-level stimuli and become selectively sensitive to strongly noxious or stressful events. A particularly well-studied noxious stimulus that potently activates LC neurons is sciatic nerve or subcutaneous electrical footshock stimulation; similar results are seen with tail or paw pinch. Short-lasting stimuli of this type elicit a brisk, phasic activation of LC neurons consistently over many presentations. This is seen under a variety of anesthetics when LC neurons are insensitive to non-noxious auditory, tactile, or visual stimuli. These findings indicate that LC neurons are particularly responsive to noxious sensory stimuli. Recent studies by our group reveal that activation of peripheral C fibers (thought to mediate painful stimuli) by high-intensity stimuli yields a specific, long-latency response of LC neurons not observed with stimuli that activate only rapidly conducting peripheral nerves. This C-fiber-mediated response summates with the response to lower-threshold fast-conducting fibers to produce a substantially greater response in the LC for nociceptive versus non-nociceptive stimuli. This indicates that the LC system may play a role in processing of painful stimuli, a result consistent with antinociceptive effects of LC activity (discussed below).

Another important aspect of these responses is that all LC neurons are similarly activated, so that a noxious stimulus elicits concerted activation from the entire LC nucleus. This homogeneity of LC activity is true for all other physiological properties examined as well. Recent evidence reveals that LC activation by footshock and tail or paw pinch is due to an excitatory amino acid (EAA; i.e., glutamate or aspartate) transmitter input (6); this suggests avenues for pharmacologic manipulation of LC responses to painful stimuli. Because of the problems of maintaining recordings during abrupt, severe movements, the effects of painful stimuli on LC activity have not been well-studied in unanesthetized animals.

LC activation is associated with antinociception via projections to the spinal cord. Recent neurophysiological work by a number of groups indicates that LC activation can produce potent antinociception. For example, Jones and Gebhart (35, 36) have shown that electrical or chemical activation of the LC area in lightly anesthetized rats can substantially increase the threshold for nociception, measured as increased latency for tail-flick to avoid a heat stimulus. Using a variety of receptor antagonists, these workers also found that this antinociceptive action of LC stimulation was mediated by spinal a2-type adrenoceptors. Consistent with this, pharmacological studies have shown that intrathecal administration of a-adrenergic agonists produces dose-dependent analgesia (55). The mechanism for the antinociceptive effect associated with LC stimulation is indicated by studies of activity of neurons in the spinal dorsal horn in response to LC stimulation. Results indicate that LC activation significantly decreases the response of these cells that are early in the pain pathway to noxious stimuli (30). Furthermore, the effect of LC activation appeared to be somewhat selective because responses of dorsal horn neurons to noxious stimuli and input mediated by C and Ad fibers were inhibited to a greater extent by LC activation than were responses of the same neurons to non-noxious stimuli or inputs mediated by faster-conducting fibers (45). These results have given rise to the idea of a descending noradrenergic antinociceptive system from the LC to the spinal dorsal horn and trigeminal system. These findings fit well with the above results for potent activation of LC neurons by noxious stimuli, and they suggest that antinociception may be part of a global influence of the activated LC system to facilitate rapid and adaptive responses to urgent stimuli (discussed below).

Virtually all classes of abused drugs affect LC discharge characteristics at doses that are in the range of those abused by humans. These include hallucinogens, stimulants, opiates, alcohol, nicotine, and benzodiazepines. Much of the work on the LC in substance abuse has focused on physical dependence and withdrawal symptoms. This interest has, in part, been generated by the usefulness of a2-adrenergic agonists such as clonidine, which suppresses LC activity, in the alleviation of withdrawal symptoms for several dependence-producing substances. However, pharmacological investigations into the acute effects of some of these agents on LC discharge suggest that the LC may also play a role in other central effects of these drugs, and they are revealing the importance of particular LC afferents in these effects.

Nicotine produces a potent activation of LC neurons when administered systemically (16). Surprisingly, recent evidence indicates that this effect of nicotine is not mediated in the LC, or even initiated in the brain, but results from nicotinic activation of primary sensory C-fiber afferents (28). For example, in contrast to intravenous (i.v.) administration, local iontophoretic application of nicotine is ineffective on LC activity (16). Systemic administration of quaternary nicotinic agonists, which would not be expected to cross the blood–brain barrier, mimic the effects of nicotine, and systemic administration of quaternary nicotinic antagonists prevent the effect of i.v. nicotine, consistent with the idea that nicotine is acting at peripheral receptors. Several findings indicate that LC activation by systemic nicotine is mediated by primary sensory C-fiber afferents and subsequent activation of excitatory amino acid LC afferents from PGi. Thus, nicotine effects on LC are prevented by capsaicin lesion of primary sensory C-fiber afferents (28), by excitatory amino acid antagonists, and by chemical inactivation of the PGi (16). In addition to increasing LC discharge, nicotine also alters discharge pattern such that burst activity is more frequent. Assuming that NE release from LC neurons is greater during burst modes, the net effect of nicotine may be to elicit more effective NE release in targets and, via effects on thalamocortical activity, produce short-lasting periods of enhanced arousal (see also Pathophysiology of Tobacco Dependence).

Like nicotine, cocaine and amphetamine-like drugs produce arousal and heightened vigilance at certain doses. However, in contrast to nicotine, the action of these CNS stimulants on LC activity is primarily inhibitory, presumably through activation of presynaptic a2 receptors by increased levels of synaptic NE. Because synaptic levels of NE are also elevated in postsynaptic targets by these stimulants, the net effect of a dose of cocaine or amphetamine will be an integration of these two opposing effects. In this regard, doses of cocaine that inhibit LC discharge rate enhanced cerebellar responsiveness to synaptic activation and iontophoretically applied neurotransmitters in a manner similar to the effects of iontophoretically applied NE (33). These studies suggest that cocaine effects on NE release in terminal regions predominate over inhibitory effects on LC cell bodies. It is likely that there is an optimal dose of cocaine at which the function of the LC–NE system is enhanced. Consistent with this, in unanesthetized rats, doses of cocaine that are in the intermediate range of the dose response for LC inhibition (0.3–3.0 mg/kg, i.v.) (10) are similar to those that have been shown to improve performance by rats in tasks requiring sustained attention (2.5 mg/kg) (25). Doses higher than this did not improve performance (see also Pathophysiology of Tobacco Dependence).

The prominent enkephalinergic input to LC, taken together with the high density of opiate receptors in this nucleus and the potent effects of opioid peptides and synthetic opiates on LC discharge, has implicated the LC as integral to endogenous opioid function as well as to the pharmacologic effects of opiates. Unfortunately, the role of LC in endogenous opioid function has been difficult to determine because opiate antagonists have little effect on LC discharge, indicating that this input is not tonically active. The LC is not unique in this regard because opiate antagonists have also been reported to have little effect in other brain regions that contain opioid peptidergic terminals (e.g., hippocampus). Future studies aimed at determining conditions under which the opioid input to LC is active will be necessary to determine the function of the LC in endogenous opioid effects.

In contrast to their lack of effect in naive animals, opiate antagonists produce a dramatic, potent, long-lasting excitation of LC neurons of rats that have chronically received opiates (see Electrophysiology and Pharmacology and Physiology of Central Noradrenergic Systems). This effect has been the basis for rationalizing the use of clonidine, which inhibits LC discharge, in the treatment of opiate withdrawal (23). Only a fraction of this excitation occurs in vitro, suggesting that a critical site of action of opiate antagonists is on cell bodies of LC afferents rather than on LC neurons or presynaptic terminals of LC afferents (9).

An excitatory amino acid input from PGi has been implicated in the bulk of LC activation associated with opiate withdrawal. Excitatory amino acid antagonists administered into the LC prior to opiate antagonists attenuate LC withdrawal excitation, and the same drugs significantly reverse this response when given after withdrawal is precipitated (2). Excitatory amino acid antagonists selective for non-NMDA receptors are more effective than selective NMDA antagonists, although both produce a significant attenuation of the activation. PGi lesions also attenuate increases in LC discharge associated with opiate withdrawal (53). The finding that excitatory amino acid antagonists do not completely reverse withdrawal-induced LC activation suggests that although a major component of this activation is mediated by opiate antagonist actions on cell bodies of excitatory amino acid LC afferents, a minor component is mediated by nonexcitatory amino acid mechanisms.

The elucidation of the circuitry underlying LC activation during opiate withdrawal has led to novel approaches of manipulating this activation. Initial studies focused on directly inhibiting LC discharge using a2-adrenergic agonists, such as clonidine. More recent studies suggest that serotonergic agents may attenuate LC activation without affecting basal LC discharge. Iontophoresis of serotonin onto LC neurons markedly attenuated activation of LC neurons by the EAA glutamate, but had no consistent effect on either basal activity or acetylcholine-evoked activity of LC neurons (5; see also Electrophysiology and Pharmacology and Physiology of Central Noradrenergic Systems). Because EAAs mediate hyperactivity of LC neurons during opiate withdrawal, these results led to the prediction that augmentation of serotonergic neurotransmission within the LC may attenuate withdrawal-induced hyperactivity. Indeed, our group has recently found that intravenous administration of indirect serotonin agonists, including the uptake blockers fluoxetine or sertraline or the serotonin releaser/uptake blocker fenfluramine, substantially attenuates the hyperactivity of LC neurons associated with morphine withdrawal (3). A similar effect was also seen in preliminary studies using local administration of fenfluramine, indicating that a possible site of action was within the LC. Furthermore, this attenuation of withdrawal hyperactivity was prevented by depletion of serotonin with PCPA, indicating that the effect was mediated through endogenous serotonin systems. Because hyperactivity of LC neurons has been implicated in a number of symptoms of opiate withdrawal, these results suggest a new pharmacotherapy for treating withdrawal which may be without the side effects of presently used agents.

The translation of LC activation to behavioral or physiological withdrawal symptoms remains controversial. The LC area was found to be the most sensitive site for the elicitation of motor aspects of opiate withdrawal by intracerebrally injected quaternary opiate antagonists (40). Moreover, electrolytic lesion of the LC attenuated wet dog shakes, mastication, rearing, piloerection, hyperactivity, ptosis, and eye twitch associated with precipitated opiate withdrawal, although these elicited signs were not completely abolished (39). Interestingly, attenuation of LC activation by administration of excitatory amino acid antagonists does not consistently alter the withdrawal syndrome as determined by scoring elicited behaviors in rats (54). However, as mentioned above, these agents do not completely abolish LC activation associated with precipitated withdrawal (see Signal Transduction Pathways for Catecholamine Receptors, Pharmacology and Physiology of Central Noradrenergic Systems, Intracellular Messenger Pathways as Mediators of Neural Plasticity, and Opioids).

Neurochemical and electrophysiological studies over the past 15 years support a strong link between stress and activation of the LC–NE system. The earliest studies implicating the LC in stress demonstrated that stress was associated with increased NE turnover in brain regions known to receive their sole NE input from the LC (i.e., hippocampus, cortex), suggesting that stress increases NE release (see ref. 66 for review). More recently, this has been substantiated by microdialysis studies, which provide a better indication of NE release (1). A likely cause of enhanced NE release during stress is increased discharge of LC neurons, and this is supported by findings that stress increases tyrosine hydroxylase expression in LC cell bodies (see Signal Transduction Pathways for Catecholamine Receptors and Intracellular Messenger Pathways as Mediators of Neural Plasticity). Electrophysiological studies in which LC activity is recorded during stress have provided more direct evidence that the increases in NE function in postsynaptic targets is due to increased LC activity. In unanesthetized cats, both environmental and physiological challenges including hypoglycemia, hyperthermia, hemorrhage, hypotension, restraint, and aversive auditory stimuli increase LC discharge, and this is usually accompanied by autonomic activation (32, 66). Similarly, in anesthetized rats, physiological challenges such as hypotension and hypercarbia increase LC discharge rate and activate sympathetic nerve discharge (63, 66). The frequent coactivation of sympathetic activity and LC discharge may be functionally significant and may reflect the fact that the PGi prominently innervates both the LC and the preganglionic sympathetic nucleus in spinal lateral horn. However, it should be noted that coactivation of sympathetic and LC activity does not always occur and the two effects are not always temporally correlated, suggesting that sympathetic activation is neither the initial stimulus for, or a consequence of, LC activation (15, 63). In spite of this distinction, the finding that both LC and the autonomic nervous system are often activated in parallel is consistent with the possibility that a common brain site (e.g., nucleus PGi) relays stress-related information to both the LC and autonomic nervous system.

Although stressors and physiological challenges have long been associated with activation of the LC–NE system, only recently has the circuitry underlying this activation been investigated. Particularly, it was not clear whether stressors of different modalities activated the LC via a common "stress" pathway or by different afferents, specific for the type of stressors. Recent studies have addressed this by pharmacologically characterizing LC activation by two physiological challenges, namely, hypotension and bladder distention. These studies suggest that LC activation by different stressors is mediated by different neurotransmitter inputs. Several findings indicate that LC activation by hypotension is mediated by CRF afferents to LC. For example, hypotensive stress elicited by i.v. nitroprusside infusion mimics the effects of intracerebroventricularly administered CRF on LC discharge; that is, it increases spontaneous discharge rate and disrupts LC responses to phasic sensory stimuli. Intracerebroventricular (i.c.v.) administration or intracoerulear microinfusion of CRF antagonists prevents LC activation by hypotensive stress, but not LC activation by footshock (66). This is in contrast to excitatory amino acid antagonists which prevent LC activation by footshock, but not by hypotensive stress (66). Taken together, these findings suggest that hypotensive stress is associated with neuronal CRF release within the LC which then activates LC neurons.

Similar pharmacologic experiments have elucidated the neurotransmitter involved in LC activation by bladder distention (66). In contrast to LC activation by hypotension, LC activation by bladder distention is largely prevented by i.c.v. or local administration of excitatory amino acid antagonists, but not by i.c.v. administration of the CRF antagonist, a-helical CRF9–41. Antagonists that are more selective for non-NMDA receptors are much more effective than NMDA-type receptor antagonists. Thus, the pharmacology for LC activation by bladder distention is similar to that for LC activation by footshock and opiate withdrawal and may involve similar LC afferents.

The studies cited above indicate that LC activation by stressors of different modalities is mediated by different neurotransmitter inputs to LC. illustrates the potential pathways and neurotransmitters thought to be involved in LC activation by different challenges. Thus, LC activation by footshock, bladder distention, nicotine, and opiate withdrawal require excitatory amino acid neurotransmission in LC which may originate (at least in part) from the nucleus PGi. In contrast, LC activation by hypotensive stimuli require CRF afferents to the LC. This argues against the idea of a common neural substrate for LC activation by stressors and suggests a certain degree of specificity with regard to responses of LC neurons to stress. Because the LC is activated by a variety of stressors, the elucidation of the neurotransmitters and pathways involved in the effects of specific stressors will likely be a focus of future investigations.

Recent studies combining LC unit recording with cortical electroencephalographic (EEG) activity have begun to address the functional consequences of elevated LC discharge during stress. These studies demonstrated that LC activation produced by hypotension is accompanied by, and necessary for, forebrain EEG activation (66). Bilateral LC inactivation produced by intracoerulear clonidine injection prevented EEG activation associated with hypotension (66). Likewise, bilateral injection of CRF antagonists into the LC prevented both LC and EEG activation by hypotensive challenge (66). These studies suggest that CRF serves as a neurotransmitter to activate the LC during hypotensive stress and that one function of this is to maintain or increase EEG arousal. Interestingly, LC activation during hypotensive stress does not appear to impact on autonomic function because when this is prevented by CRF antagonists, the magnitude and duration of the hypotensive response is unaltered (66). This is consistent with anatomic findings that the LC does not substantially project to autonomic nuclei.

Other findings are consistent with the view that cortical arousal may be a common consequence of LC activation by multiple stimuli. Bladder distention, like hypotensive stress, increases LC discharge and activates the electroencephalogram, and both of these effects are prevented by pretreatment with excitatory amino acid antagonists (66). Recently, selective pharmacologic manipulation of LC discharge has been shown to have profound effects on forebrain electrophysiology recorded as electroencephalogram (see Electrophysiology and Pharmacology and Physiology of Central Noradrenergic Systems). Taken together, these results argue for at least one common consequence of LC activation regardless of the input (i.e., EEG arousal).

Because of the limitations of animal models of psychiatric disease, it has been more difficult to ascertain a role of the LC or its afferents in specific psychiatric disorders. Of the clinical phenomena that have implicated the LC–NE system, depression has been the most thoroughly studied, perhaps because of the neurochemical and pharmacological evidence that strongly supported the original biogenic amine hypothesis of depression. This hypothesis has been much revised since its original formulation such that it is now thought that dysfunctions, rather than decreased functions, of serotonin or norepinephrine systems are important (59). In addition to biogenic amine function, hypothalamic–pituitary–adrenal function is also abnormal in depression (61). However, the relationship between dysfunctions in these two systems was relatively understudied. The integration of findings from basic and clinical studies in the last few years has suggested that CRF may be an important link between neuroendocrine and biogenic amine dysfunctions in depression. This has been partly based on the dual role of CRF as a hypothalamic neurohormone that initiates endocrine components of the stress response and as a neurotransmitter serving to activate the LC–NE system in stress. Substantial clinical neuroendocrine findings implicate enhanced hypothalamic CRF activity in the endocrine dysfunctions that characterize depression (24). A parallel increase in CRF neurotransmitter activity in the LC would be predicted to cause persistent activation of this nucleus, with disrupted responses to brief sensory stimuli. The consequences of this may be hyperarousal and inability to concentrate, two symptoms that are characteristic of depression. Nonetheless, these predictions are dependent on the assumption that hypothalamic and extrahypothalamic CRF are hypersecreted in parallel in depression, and this has yet to be demonstrated (see Corticotropin-Releasing Factor: Physiology, Pharmacology, and Role in Central Nervous System and Immune Disorders and Neuropeptide Alterations in Mood Disorders).

There are numerous other conditions in which the LC–NE system has clearly been implicated and which deserve mention. Dysfunction of the LC–NE system may be involved in attention deficit disorder based on the role of the LC in attention which is reviewed in more detail elsewhere in this volume (see Pharmacology and Physiology of Central Noradrenergic Systems and Central Norepinephrine Neurons and Behavior). Indeed, drugs that are effective in this disorder profoundly affect LC activity and NE in postsynaptic targets. Future treatments of attention deficit disorder may focus on more controlled pharmacological manipulation of this system.

Marked parallels between the symptoms of post-traumatic stress disorder (PTSD) and LC hyperactivity suggest that LC dysfunction may be involved in this phenomenon. For example, PTSD is precipitated by stressful events, and stressors activate the LC. Certain symptoms of PTSD, including hypervigilance and sleep abnormalities, are also predictable signs of LC hyperactivity. Other symptoms of PTSD are characterized by autonomic hyperactivity (tachycardia, hypertension, pallor, flushing, and sweating), and there appear to be mechanisms by which peripheral sympathetic and central noradrenergic function can be activated in parallel (see above). These similarities between PTSD and LC function suggest that this may be a key nucleus whose dysregulation leads to perhaps some of the symptoms of PTSD (see also Noradrenergic Neural Substrates for Anxiety and Fear: Clinical Associations Based on Preclinical Research).

In the last four years, detailed analyses of the anatomy and physiology of the LC–NE system have provided substantial evidence that this system is more specifically organized than previously thought. Future questions will be directed at the level of specificity of this system and the mechanisms by which specificity is conferred. The studies necessary to answer these questions must focus on a better understanding of the topography of LC afferents and efferents, identification of the circuitry and neuromediators underlying LC responses to stimuli, more detailed analysis of the cellular effects of norepinephrine and/or LC stimulation on different postsynaptic targets, and quantification of the translation of LC discharge to postsynaptic effect. These future studies will help reveal the function of the LC–NE system in normal and pathologic conditions and how this system can be pharmacologically or behaviorally manipulated to treat psychiatric disorders.

published 2000