Noradrenergic Neural Substrates for Anxiety and Fear

Dennis S. Charney, J. Douglas Bremner, and D. Eugene Redmond, Jr.

Preclinical studies investigating the physiological mechanisms of fear and anxiety states have strongly suggested that multiple brain neurochemical and neuropeptide systems, including noradrenergic, benzodiazepine, serotonergic, and corticotropin-releasing hormone, are involved in the pathophysiology of human anxiety. In addition, the evidence implicates specific brain structures such as the amygdala, thalamus, hypothalamus, central gray, hippocampus, locus coeruleus, and prefrontal cortex as mediators of the broad range of behaviors and physiological responses associated with anxiety and fear.

The focus of this review is the role of brain noradrenergic neural systems in the development of human anxiety disorders—clinical associations based on preclinical research. Neuroanatomical, neurochemical, neurophysiological, and behavioral studies of the noradrenergic system provide a basis for relating increased activity of this system to the expression of anxiety and fear and the somatic symptoms and cardiovascular changes that accompany severe anxiety states (). We will first review some of the main preclinical findings and then describe the status of clinical studies largely based on these preclinical data (see also Pharmacology and Physiology of Central Noradrenergic Systems, Central Norepinephrine Neurons and Behavior, and Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications).

The neuroanatomy of the largest central noradrenergic nucleus, the locus coeruleus (LC), is characterized by an extensive efferent projection system and a more restricted afferent input (5, 20, 40). A retrograde labeling study of afferent input to the LC using iontophoretic horseradish peroxidase (HRP) injections found that forebrain structures, such as the neocortex, amygdala, and hypothalamus, project to the LC. The hypothalamic innervation is noteworthy because it provides a potential regulatory role for the LC with regard to autonomic function. Brainstem monoaminergic neurons, such as the raphe nuclei and a variety of sensory relay areas, also project to the LC. The existence of these sensory afferents from the spinal cord and the nucleus of the solitary tract provide insight into the mechanisms by which the LC is responsive to noxious stimuli and alterations in cardiovascular function (10). Consistent with these observations, a recent neuroanatomical investigation involving anterograde labeling with PHA-L in the cat and monkey found terminal fibers in the LC following injections in lamina I of the spinal or medullary dorsal horn (20). These observations are supported by an electrophysiological mapping study which revealed that spinal cord lamina I cells terminated in the LC (40). There are reports, in contradiction to the above, that direct afferents to the LC almost exclusively come from the paragigantocellularis (PGi) and the prepositus hypoglossi (PrH) (4). These studies used very small injections of wheat germ agglutinin conjugated to HRP (WGA-HRP) that were restricted to the LC in the rat. The discrepant observations may be due to the fact that WGA-HRP is not an efficient retrograde marker for lamina I neurons (21). The finding of additional afferents to the LC using choleratoxin B subunit as a retrograde tracer instead of WGA-HRP is consistent with this assertion (38). The hypothesis that PGi and PrH are the predominant afferents to the LC is also not supported by the demonstration that destruction of these nuclei fail to block LC responses to somatosensory stimuli (51).

If not the primary LC afferent, the PGi remains an important afferent to the LC and is a key sympathoexcitatory region in brain (5). There are widespread afferents to the PGi from diverse brain areas (81). The PGi may be an integrative pathway for activating the LC by a variety of mechanisms. The PGi is involved in control of arterial blood pressure, cardiopulmonary reflexes, and parasympathetic function. It has been suggested that the peripheral sympathetic nervous system is activated in parallel with the LC by projections to both areas from the PGi (5). The LC also appears to be activated by projections responsive to corticotropin-releasing factor (CRF), and considerable evidence suggests that CRF has anxiogenic properties. Intracerebroventricular infusion of CRF increases norepinephrine (NE) turnover in several forebrain areas (24). In a dose-dependent fashion, CRF increases the firing rate of LC–NE neurons (80). Stress, which activates NE neurons, markedly increases CRF concentrations in the LC (11). Moreover, it has recently been demonstrated that infusion of CRF into the LC produces anxiogenic activity and significant increases in the NE metabolite 3,4-dihydroxyphenylglycol in forebrain areas such as the amygdala and hypothalamus (6).

The LC has diffuse projections to the entire cortical mantle, as well as thalamus, hypothalamus, amygdala, hippocampus, cerebellum, and spinal cord. Through this broad efferent network the LC can mediate a range of cognitive, neuroendocrine, cardiovascular, and skeletal motor responses that accompany anxiety and fear.

The LC is the primary source of NE fibers in a number of neocortical regions in the primate. Consistent with postsynaptic inhibitory effects of NE, activation of the LC neurons produce inhibition of neuronal activity in brain regions receiving projections from the LC, including the cochlear nucleus, cerebral and cerebellar cortices, spinal trigeminal nucleus, hippocampus, caudate, and superior colliculus. Similarly, LC stimulation reduces cerebral metabolism and blood flow in these areas (31, 58). Because NE has greater inhibitory effects on spontaneous than on evoked neuronal activity, it has been suggested that a function of the LC–NE system is to enhance neuronal responses to inputs of behavioral significance or, put another way, to increase the signal-to-noise ratio (3); (also see Pharmacology and Physiology of Central Noradrenergic Systems and Central Norepinephrine Neurons and Behavior).

Electrical stimulation of the LC produces a series of behavioral responses similar to those observed in naturally occurring or experimentally induced fear (55). These behaviors are also elicited by administration of drugs, such as yohimbine and piperoxone, which activate the LC by blocking alpha-2-adrenergic autoreceptors. Drugs which decrease the function of the LC by interacting with inhibitory opiate (morphine), benzodiazepine (diazepam), and alpha-2 (clonidine) receptors on the LC decrease fearful behavior and partially antagonize the effects of electrical stimulation of the LC in the monkey (53). These studies suggest that abnormally high levels of LC activity producing increased release of NE at postsynaptic projection sites throughout the brain may act to augment some forms of fear or pathological anxiety, depending on the environmental conditions (19).

A single report of electrical stimulation of the region of the LC in humans reported that the subjects experienced feelings of fear and imminent death. In a subject with chronically implanted stimulation electrodes in the vicinity of the LC, electrical stimulation during sleep was associated with marked insomnia (32) ().

Bilateral lesions of the LC in the monkey decrease the natural occurrence of these behavioral responses in a social group situation and in response to threatening confrontations with humans (53). Agents which decrease the function of the LC also decrease fearful behavior and partially antagonize the effects of electrical stimulation of the LC in the monkey. Lesions of the dorsal bundle which originates in the LC and projects mostly to the neocortex, hippocampus, and cerebellar cortex produce anxiolytic-type responses in anxiogenic behavioral test situations (82).

Many studies in rodents using the neurotoxin 6-hydroxydopamine (6-OHDA) have shown effects on a variety of behavioral tests which have been interpreted as inconsistent with the hypothesis that noradrenergic systems are critically involved in anxiety and fear (41). The validity of these behavioral paradigms to reflect anxiety and fear has been the subject of much debate (34, 53, 54).

In laboratory rats, chronic stress results in an increased firing of the LC (47, 64). Animals exposed to chronic inescapable shock, which is associated with learned helplessness, have an increase in responsiveness of the LC to an excitatory stimulus in comparison to animals exposed to escapable shock (64).

The effect of stressful and fear-inducing stimuli on LC activity has been assessed in freely moving cats (37, 52). Conditions which are behaviorally activating, but not stressful, such as exposure to inaccessible rats or food, do not increase LC firing in cats as they do in rats (see Pharmacology and Physiology of Central Noradrenergic Systems). In contrast, stressful and fear-inducing stimuli, such as loud white noise, air puff, restraint, and confrontation with a dog, produce a rapid, robust, and sustained increase in LC activity (37). Of interest, these increases in LC function are accompanied by sympathetic activation. Generally, the greater the sympathetic activation in response to the stressor, as indicated by heart rate, the greater the correlation observed. Thus, a stimulus intensity threshold for coactivation of central and peripheral NE systems may exist.

A parallel activation of LC neurons and splanchnic sympathetic nerves is produced by noxious stimuli. The LC, like sympathetic, splanchnic activity, is highly responsive to various peripheral cardiovascular events, such as alterations in blood volume or blood pressure. Internal events that must be responded to for survival, such as thermoregulatory disturbance, hypoglycemia, blood loss, an increase in pCO2, or a marked reduction in blood pressure, cause robust and long-lasting increases in LC activity (67).

There are also peripheral visceral influences on LC activity. In rats, distention of the urinary bladder, distal colon, or rectum activates LC neurons. These findings suggest that changes in autonomic or visceral function may result in specific behavioral responses via the brain LC–NE system. The LC–NE network may help determine whether, under threat, an individual turns attention toward external, sensory stimuli or to internal vegetative events. The system, when functioning normally, may be important in facilitating the planning and execution of behaviors important for survival (67) ().

Stressful stimuli of many types produce marked increases in brain noradrenergic function. Stress produces regionally selective increases in NE turnover in the LC, limbic regions, and cerebral cortex. It has recently been demonstrated that immobilization stress, footshock stress, and tail-pinch stress increase noradrenergic metabolism in the hypothalamus and amygdala (29, 70, 71). Stress also increases tyrosine hydroxylase levels in the LC (42). Anxiolytic agents reverse the effects of stress on noradrenergic metabolism (29, 62, 69). Consistent with these findings, acute cold restraint stress results in decreased density of alpha-2-adrenergic receptors in the hippocampus and amygdala (73). The stress-induced increase in NE turnover is also associated with a decrease in postsynaptic beta receptor density (79) ().

Several behavioral paradigms indicate an important role for noradrenergic neuronal systems in the processes involved in fear conditioning. Neutral stimuli paired with shock produce increases in brain NE metabolism and behavioral deficits similar to those elicited by the shock (88). In the freely moving cat, the firing rate of cells in the LC can be increased by presenting a neutral acoustic stimulus previously paired with an air puff to the whiskers, which also increases firing and is aversive to the cat (52). There is also a body of evidence indicating that an intact noradrenergic system may be necessary for the acquisition of fear conditioned responses (74).

Sensitization generally refers to the increase in behavioral or physiological responsiveness that occurs following repeated exposure to a stimulus. Behavioral sensitization can be generally context-dependent or conditioned, such that animals will not demonstrate sensitization if the stimulis is presented in a different environment (50). However, if the intensity of the stimulis or drug dose is high enough, behavioral sensitization will occur even if the environment is changed. It has been suggested that different mechanisms are called into play with environment-independent sensitization (84).

Behavioral sensitization to stress involves alterations in noradrenergic function. Limited shock exposure that does not increase NE utilization in control rats does increase NE release in animals previously exposed to the stressor. Moreover, changes in noradrenergic function in animals subjected to long-term shock require lower shock currents (decreased stressor intensity) than required under acute conditions (30. An in vivo study observed augmented extracellular NE concentrations in the hippocampus, whereas ex vivo measurements of noradrenergic metabolites in response to chronic stress indicated a sensitized response in the hypothalamus but not in the hippocampus (45). It is not clear to what degree this reflects differences in metabolic disposition of NE in the two regions, as opposed to actual differences in sensitization processes. Nonetheless, regional specificity in biochemical indices of the expression of sensitization may be important. A recent in vivo dialysis investigation demonstrated stress-induced sensitization of NE release in the medial prefrontal cortex (26).

Stress and learning alter gene expression by modifying the binding of transcriptional activator proteins to each other and to the regulatory regions of genes (8). It has been suggested that immediate early genes, such as C-fos, whose proteins, acting in the nucleus as "third messengers," produce stable modifications in the transcriptions of late genes to long-term memory (1). Immediate early genes may serve a variety of functions in mediating genomic responses to extracellular stimuli (66).

Recently, it has been demonstrated that C-fos and other immediate early genes in the brain are activated by increases in brain noradrenergic neurotransmission. For example, yohimbine, an alpha-2-receptor antagonist, increases c-fos expression in rat cerebral cortex (66).

In support of studies that demonstrated that acute and chronic stress activate brain NE systems, stress increases expression of c-fos immunoreactivity in the LC (9). Furthermore, activation of the LC with glutamate produces marked and widespread increases in c-fos protein in brain regions receiving projections from the LC. If c-fos induction is a component of the noradrenergic signal transduction pathways, then basal levels of c-fos mRNA in brain could represent a summation of activity of noradrenergic neurons (see also Pharmacology and Physiology of Central Noradrenergic Systems, Central Norepinephrine Neurons and Behavior, and Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications).

Clinical Investigation of Noradrenergic Function in Healthy Subjects and Anxiety Disorder Patients ()

Accumulated evidence suggests that brain noradrenergic systems play a role in mediating normal state anxiety and the response to stress in healthy human subjects. States of anxiety or fear appear to be associated with an increase in NE release in healthy subjects (83).

Levels of the NE metabolite, MHPG, have also been found to increase in healthy subjects during emotional stress (36). Plasma MHPG was correlated with state anxiety in healthy subjects exposed to the anticipatory stress of receiving an electric shock, while there was no such correlation in the absence of the electric shock threat (75, 78). Significant within-individual correlations between changes in urinary MHPG and changes in state anxiety have been found in healthy human subjects (68).

In comparison to other anxiety disorders, the evidence for an abnormality in noradrenergic function is most compelling for panic disorder and Post-traumatic stress disorder (PTSD). Panic disorder and PTSD patients frequently report cardiovascular, gastrointestinal, and respiratory symptoms. Because the LC is responsive to peripheral alterations in the function of these systems, minor physiological changes in these patients may result in abnormal activation of LC neurons and, consequently, panic attacks and flashbacks. These functional interactions may explain the association of anxiety symptoms with tachycardia, tachypnea, hypoglycemia, and visceral and organ distention, as well as the marked sensitivity of panic disorder and PTSD patients to interoceptive simuli. The important role of the noradrenergic system in fear conditioning may account for the development of phobic symptoms in these patients. Finally, the involvement of noradrenergic neurons in learning and memory may relate to the persistence of traumatic memories in PTSD.

Generally, measurement of peripheral NE and its metabolites have revealed concentrations in panic disorder patients similar to those in controls. Several studies have not found elevated plasma catecholamines following spontaneous or situationally provoked panic attacks (7, 85). Moreover, plasma and urinary catecholamine levels (43, 44), as well as cerebrospinal fluid (CSF) MHPG levels, in panic disorder patients are generally not different from those in healthy controls (25, 76; B Lydiard, personal communication, April 1992).

Two studies have found significantly elevated 24-h urine NE excretion in combat veterans with PTSD compared to healthy subjects or patients with schizophrenia or major depression (35, 87). A wide variety of investigations have been conducted to evaluate sympathetic nervous system (SNS) function in panic disorder. These studies have produced markedly divergent findings yielding no firm concensus on whether an SNS dysfunction exists in panic disorder (57).

The regulation of noradrenergic neuronal function has been examined by determining the behavioral, biochemical, and cardiovascular effects of oral and intravenous yohimbine, an alpha-2-adrenergic receptor antagonist, in a spectrum of psychiatric disorders, including schizophrenia, major depression, obsessive–compulsive disorder, generalized anxiety disorder, panic disorder, and PTSD (19). Specific abnormalities have been identified in panic disorder and PTSD. Approximately 60–70% of panic disorder patients experience yohimbine-induced panic attacks, and these patients have larger yohimbine-induced increases in plasma MHPG, blood pressure, and heart rate than do healthy subjects and those with other panic psychiatric disorders (15, 16, 18).

Similar to the panic disorder patients, approximately two-thirds of PTSD patients experience yohimbine-induced panic attacks, and, in addition, 40% report flashbacks after yohimbine. As a group, PTSD patients also have greater yohimbine-induced increases in plasma MHPG, sitting systolic blood pressure, and heart rate than do healthy subjects. A striking effect of yohimbine is its ability to increase the severity of the core symptoms associated with PTSD, such as intrusive traumatic thoughts, emotional numbing, and grief (65). This may be due to the involvement of noradrenergic systems in the mechanisms by which memories of traumatic experiences remain indelible for decades and are easily reawakened by a variety of stimuli and stressors.

Recently, the effects of yohimbine on regional cerebral blood flow (rCBF) and metabolism have been evaluated in panic disorder and PTSD patients. In panic disorder patients, yohimbine significantly reduced frontal rCBF rates in patients compared to healthy subjects (86). Similar to these observations, administration of yohimbine decreased regional glucose metabolic rates in several cortical regions in PTSD patients, but not in healthy subjects. Because cortical rCBF, metabolism, and spontaneous neuronal activity all generally decrease with noradrenergic stimulation, these data are consistent with a role for excessive stimulation of noradrenergic projections to the cortex in the pathophysiology of panic disorder and PTSD.

A consistent finding in the literature is that the growth hormone rise induced by clonidine is blunted in panic disorder patients (77). In a recent investigation, clonidine-growth hormone response was found primarily in the patients who experienced yohimbine-induced panic attacks (18). This suggests that the diminished postsynaptic alpha-2-adrenergic receptor function reflected by the blunted clonidine–growth-hormone response may relate to presynaptic noradrenergic neuronal hyperactivity.

Several previous investigations observed that clonidine produced greater decreases in plasma MHPG and blood pressure in panic disorder patients compared to healthy subjects (18, 46; J. D. Coplan and colleagues, personal communication, December 1993). The clonidine-induced decreases in plasma MHPG may be greatest in the panic disorder patients who experienced yohimbine-induced panic attacks (18), suggesting that there is a distinct subgroup of panic disorder patients who manifest noradrenergic neuronal dysfunction. The effects of clonidine on these parameters have not been investigated in PTSD patients.

Infusion of isoproterenol, a peripherally acting compound that is selective for the beta adrenoceptor, has been reported to trigger anxiety responses in panic patients compared with controls (49). Successful treatment of panic patients with tricyclic antidepressants blunted isoproterenol-induced anxiety and systolic blood pressure responses (48). These studies are consistent with the hypothesis of increased beta-1-adrenoceptor sensitivity in panic disorder, which is normalized by effective pharmacotherapy (48).

Evidence is emerging that the efficacy of some tricyclic and monoamine oxidase inhibitor drugs against panic may be related to their regulatory effects on noradrenergic activity (14). The effects of chronic treatment with these agents on the regulation of noradrenergic activity are complex. Some of these effects, such as reduced tyrosine hydroxylase activity, LC firing rate, NE turnover, and postsynaptic beta-adrenergic receptor sensitivity, diminish noradrenergic function. It is interesting that the only antidepressant drugs that do not exhibit antipanic efficacy—bupropion and trazodone—have effects on noradrenergic function different from those of the tricyclic and monoamine inhibitors (60). In the rat brain, bupropion does not down-regulate beta-adrenergic receptors or decrease NE turnover. While trazodone does down-regulate the beta-adrenergic receptors, it does not decrease the LC firing rate or the spontaneous activity of cortical neurons receiving noradrenergic innervation.

Benzodiazepines are highly effective treatments for panic disorder, generally at higher doses than those needed for generalized anxiety disorder. Clearly, the anxiolytic effects of benzodiazepines are related to their agonist actions at benzodiazepine receptors at a variety of brain sites. However, it has been hypothesized that the antipanic properties of benzodiazepines may also relate to inhibitory effects on noradrenergic function, because these drugs reduce LC neuronal activity (13).

The antipanic efficacy of the potent serotonin reuptake inhibitors (SRIs) such as clomipramine, fluvoxamine, zimelidine and fluoxetine is well-documented. The mechanism of action of SRIs in panic disorder has not been established. However, 5-HT2, 5-HT1C, and 5-HT1A receptors are unlikely to be directly involved because ritanserin, a 5-HT2 and 5-HT1C antagonist, and buspirone, a 5-HT1A agonist, lack antipanic efficacy (23, 61).

It is possible that interactions between the serotonin (5-HT) and noradrenergic systems may be related to the antipanic properties of SRIs. Preclinical studies suggest that 5-HT–NE interactions occur between the LC and the dorsal raphe. Direct application of 5-HT to LC neurons results in a tonic inhibition of electrical activity. Phasic 5-HT inhibition of LC function may be mediated by an excitatory amino acid (EAA) pathway from the nucleus paragigantocellularis, possibly via a 5-HT1A receptor (2). In this context it is notable that fluvoxamine has been found to alter EAA receptor mRNA expression throughout the brain (28). The interactions between the noradrenergic and serotonin systems are supported by the finding that fluvoxamine, but not placebo treatment, reduced yohimbine-induced anxiety in panic disorder patients (27).

If the noradrenergic system is dysregulated in panic disorder, the mechanism of action of antipanic therapy may be its ability to decrease the wide and unpredictable fluctuations in noradrenergic activity and improve efficiency by reducing basal activity (decreasing noise) while effecting a more specific responsiveness to specific stimuli (increasing signal-to-noise ratio).

The treatment implications of the reported abnormalities in noradrenergic function in PTSD patients remain to be established. Specific PTSD symptoms (anxiety, flashbacks, and autonomic arousal) may be particularly responsive to drugs that reduce noradrenergic function (33). It should be noted, however, that patients with panic disorder and those with PTSD have different therapeutic responses to tricyclic, SRI, and benzodiazepine compounds. Patients with panic disorder derive great benefit from these drugs, whereas those with PTSD have more modest responses. This may be related to the large number of neural systems that are affected by stress and likely to be involved in the pathophysiology of PTSD (12).

There is little clinical evidence supporting a primary role for the noradrenergic system in the pathophysiology of GAD. Plasma MHPG levels have been shown to be both increased (59) and not different (17) compared to normal controls. Similarly, increases in resting plasma NE in GAD patients have been reported in some studies (59) but not others (39). Growth hormone response to clonidine has been found to be blunted in GAD patients (22). Patients with GAD have been found to have normal responses to the alpha-2 antagonist, yohimbine (17).

Very limited evidence supports any important role for noradrenergic systems in the pathophysiology of obsessive–compulsive disorder or phobic disorder, but full review of the literature is beyond the scope of this chapter (63, 72).

Some studies have shown evidence for a relationship between CSF MHPG and anxiety (but not depression) in patients with major depressive disorders. Successful antidepressant treatment also seemed to begin with improvements in anxiety during the first week of treatment. These and other interactions with noradrenergic function in depression are consistent with a noradrenergic role in the emotion of anxiety or fear (56). Some data also exist showing increased noradrenergic function during mania, but like many of the correlative measurements they are confounded by physical activity and other extraneous factors.

While preclinical and clinical studies suggest that dysfunction of brain noradrenergic neurons may be involved in the development of specific human anxiety disorders, clinical anxiety is unlikely to be due entirely to abnormalities in a single neurotransmitter system. First, it is generally agreed that models which postulate too little or too much of a single neurotransmitter are not consistent with the complex regulation of neurotransmitter systems which involve multiple neurotransmitter receptor subtypes. Second, there are major functional interactions among different neurotransmitter and neuropeptide systems which make single neurotransmitter theories simplistic.

Similar arguments can be made against hypotheses that overemphasize a single brain structure to account for the spectrum of symptoms associated with human anxiety disorders. Brain structures, such as the amygdala, hypothalamus, hippocampus, and LC, are neuroanatomically and functionally interrelated and are probably responsible for different components of anxiety disorder syndromes.

Clinical research findings in panic disorder support these assertions. Drugs that produce descriptively similar panic anxiety states, such as yohimbine, MCPP, FG-7142, caffeine, lactate, and CCK-4, have many actions on neurotransmitter function, neuropeptide secretion, and cardiovascular and respiratory function. A task for future investigations of the pathophysiology of anxiety disorders is to develop clinically applicable biological tests that can assess the functional interactions among different neurotransmitter and neuropeptide systems and specific brain structures. Successful development of such paradigms could result in improved diagnostic classification and prediction of treatment response of human anxiety disorders.

This work was supported by the National Center for Post Traumatic Stress Disorder, West Haven (Connecticut) Veterans Affairs Medical Center, by grants RO1-MH40140 and MH30929 from the National Institute of Mental Health, Bethesda, Maryland (Dr. Charney), and by Research Scientist Award MH00643 (Dr. Redmond). The authors thank Evelyn Testa and Karen Tyler for excellent manuscript preparation.

published 2000