Psychiatric symptoms of noradrenergic dysfunction: A pathophysiological view

Authors

  • Ken-ichi Yamamoto MD, PhD,

    Corresponding author
    1. Stress Disorder Research Project Team, Tokyo Metropolitan Institute of Medical Science (the former Tokyo Institute of Psychiatry), Tokyo, Japan
    • Correspondence: Ken-ichi Yamamoto, MD, PhD, Shikiba Mental Hospital, 6-1-14, Kounodai, Ichikawa City, Chiba 272-0827, Japan. Email: yamaken@river.ocn.ne.jp

    Search for more papers by this author
  • Toshikazu Shinba MD, PhD,

    1. Stress Disorder Research Project Team, Tokyo Metropolitan Institute of Medical Science (the former Tokyo Institute of Psychiatry), Tokyo, Japan
    Search for more papers by this author
  • Mitsunobu Yoshii MD, PhD

    1. Stress Disorder Research Project Team, Tokyo Metropolitan Institute of Medical Science (the former Tokyo Institute of Psychiatry), Tokyo, Japan
    Search for more papers by this author

Abstract

What psychiatric symptoms are caused by central noradrenergic dysfunction? The hypothesis considered in this review is that noradrenergic dysfunction causes the abnormalities in arousal level observed in functional psychoses. In this review, the psychiatric symptoms of noradrenergic dysfunction were inferred pathophysiologically from the neuroscience literature. This inference was examined based on the literature on the biology of psychiatric disorders and psychotropics. Additionally, hypotheses were generated as to the cause of the noradrenergic dysfunction. The central noradrenaline system, like the peripheral system, mediates the alarm reaction during stress. Overactivity of the system increases the arousal level and amplifies the emotional reaction to stress, which could manifest as a cluster of symptoms, such as insomnia, anxiety, irritability, emotional instability and exaggerated fear or aggressiveness (hyperarousal symptoms). Underactivity of the system lowers the arousal level and attenuates the alarm reaction, which could result in hypersomnia and insensitivity to stress (hypoarousal symptoms). Clinical data support the hypothesis that, in functional psychoses, the noradrenergic dysfunction is in fact associated with the arousal symptoms described above. The anti-noradrenergic action of anxiolytics and antipsychotics can explain their sedative effects on the hyperarousal symptoms of these disorders. The results of animal experiments suggest that excessive stress can be a cause of long-term noradrenergic dysfunction.

NORADRENALINE (NA) WAS identified by von Euler in 1946 as a neurotransmitter released from the sympathetic nerve terminal.[1] NA is a demethylated form of adrenaline, which is a hormone released from the adrenal medulla. The functional significance of the sympatho-adrenal system had previously been explained by Cannon using the idea of ‘the emergency reaction (fight or flight reaction)’[2] and by Selye as ‘the alarm reaction in stress’.[3] In 1954, Vogt predicted the existence of NA neurons in the central nervous system (CNS) based on unequal distribution of NA in the brain.[4] The existence of the central NA system was demonstrated directly with the invention of fluorescent histochemistry in the 1960s.[5] During the 3 decades following this invention, remarkable development was achieved in the basic neuroscience of this system.

However, application of this development in neuroscience to psychiatry is still largely delayed.[6] Two reasons may explain this delay.

One potential explanation relates to overexpectation for the nosology of psychiatric disorders. Although NA research has not been the major issue of biological psychiatry, evidence is accumulating independently for the correlations of noradrenergic (NAergic) dysfunction to various psychiatric disorders within distinct nosological classification (e.g. post-traumatic stress disorder [PTSD], anxiety disorder, mood disorder and schizophrenia). If the biological bases of these nosologically different disorders are completely unrelated, these results appear to be contradictory. However, from a pathophysiological point of view, it is possible that NAergic dysfunction is related to multiple disorders.

Second, it is difficult to translate findings from animal experiments to psychiatric concepts in humans. In this review, psychiatric symptoms of NAergic dysfunction were inferred pathophysiologically by integrating developments from neuroscience, particularly those from basic study on behavioral pharmacology and psychophysiology. This inference was examined based on the literature on the biology of psychiatric disorders and psychotropics. Additionally, the potential role of excessive stress was considered as a cause of NAergic dysfunction.

Anatomical Basis

Using fluorescent histochemistry, Darsröm and Fuxe discovered seven NA-containing cell groups in the brainstem and named them A1–A7.[7] A6, which is a particularly large cell group in humans, coincides with the so-called ‘locus coeruleus (LC)’ in classic neuroanatomy. Nonmyelinated nerve fibers from these seven cell groups diffusely cover nearly the entire CNS.[8] This morphological feature of the central NAergic system suggests the system's importance in modulating brain functions.

Given the difference in the distribution of fiber projections, the central NA neurons can be divided into two major systems with some overlap:[8] the LC system, which originates in the LC (A6) and A4, and the non-LC system (the lateral-tegmental NA system), which originates in other NA cell groups (A1, 2, 3, 5, and 7). Although LC neurons send out axons that broadly cover the entire CNS, phylogenetically new regions, such as the cerebral cortex, hippocampus and neocerebellum, receive fiber projections only from the LC system. In contrast, older regions, including the hypothalamus and the archicerebellum, appear to receive rich fiber projections primarily from the non-LC system. These differences in the fiber projections have also been observed in the spinal cord. The pre-ganglionic neuron pool of the sympathetic nervous system receives dense projections from non-LC neurons, whereas that of the parasympathetic system receives projections from LC neurons.[9]

It is known that the number of LC neurons in various mammals increases logarithmically in proportion to the volume of the cerebral cortex.[10, 11] Thus, the energy consumption of a single LC neuron should increase exponentially with the phylogenetic development of the cerebral cortex, suggesting that the LC is a‘locus minoris’ (weak point) in humans.[12]

What Mental Symptoms Does Noradrenergic Dysfunction Cause?

Correlation to sympathetic activity

Of the non-LC neuron groups, A5 is known to have a strong correlation to the sympathetic pre-ganglionic neurons in the spinal cord.[9, 13, 14] In addition, it has been demonstrated that A1 neurons have strong projections to the supraoptic nucleus, which secretes vasopressin, and that the A2 neurons project to the paraventricular nucleus, which secretes corticotrophin-releasing factor (CRF) and activates the ACTH-adrenocortical system.[15, 16] Thus, the non-LC system appears to be connected directly to the center of the sympathetic and hypophysis-adrenocortical system.

It has been reported that the activity of A5 neurons, which are identified by an antidromic response to electrical stimulation in the cervical cord, correlates well with sympathetic activity.[17] Although electrical stimulation of A5 elevates the blood pressure, it also decreases the heart rate.[18] The decrease in the heart rate becomes an increase when the afferent pathway of the vagal nerve is destroyed prior to stimulation. Therefore, the heart rate decrease that follows A5 stimulation is likely a result of feedback inhibition. According to Huangfu et al.,[14] this elevation of blood pressure by A5 stimulation does not occur when A5 is destroyed by the microinjection of 6-hydroxydopamine (6-OHDA), a selective neurotoxin of catecholamine (CA) neurons.[19] (CA is the collective name of NA and dopamine [DA]). The electrical stimulation of A1 and A2 is reported to increase secretion of vasopressin and CRF, respectively. In contrast, the destruction of this area is known to reduce CRF secretion after immobilization stress.[20, 21]

However, the LC is also related to sympathetic activity. A substantial number of studies[22] note that CRF excites LC neuronal activity. A correlation between LC neuronal activity and sympathetic activity, as evidenced by parameters such as heart rate, blood pressure and cervical sympathetic tone, has been reported.[23, 24] Abercrombie and Jacobs recorded LC unit activity, heart rate and plasma NA level in unanesthetized cats and showed that plasma NA level was increased in relation to these sympatho-adrenal activities when the cat was in stressful circumstances, such as when restrained or when exposed to white noise.[25]

Modulation of arousal level

The single-cell firing of LC neurons is completely silent during REM sleep.[26] However, the firing becomes slightly active during slow wave sleep and increases even more during waking.[27] Foote et al. observed that LC neuronal firing in awake rats and monkeys was correlated with the desynchronization of electroencephalogram (EEG) activity and was activated by ‘arousing stimuli’, such as the entrance of an unfamiliar person into the laboratory or the sight of a favorite food.[28]

It has also been reported that the microinfusion of yohimbine, an α2-autoreceptor antagonist that excites NA neurons, into the LC causes behavioral arousal, including increased locomotor and exploratory activity, tachypnea and EEG desynchronization with a significant reduction in total voltage power. In contrast, microinfusion of clonidine, an α2-autoreceptor agonist that inhibits NA neurons selectively, into the LC is reported to cause marked behavioral sedation (hypomotility) and sleep with EEG synchronization.[29]

Using operant conditioning in rats trained to respond to a reaction-time task when given a warning stimulus (the so-called CNV task), Yamamoto and Ozawa found that LC neuronal firing is increased in association with alertness for prompt response after the presentation of the warning stimulus (preparatory set).[30] Furthermore, in monkeys trained to track a visual target with their eyes, the activity of LC neurons is observed to increase with fixation on a stationary target or with accurate tracking of a moving target, which suggests a close correlation of LC activity to attention.[31]

These neurophysiological findings strongly support the view that the LC system is of essential importance in controlling the arousal level or vigilance.[11, 32]

The correlation between the LC and arousal level explains why LC activity modulates all of the brain's sensorimotor activity.[12] The well-known inverted U-shaped correlation between arousal level and efficiency of higher intellectual functions, such as selective attention and memory,[33] has also been confirmed between LC activity and executive functions in animal experiments.[34]

Amplification of emotional reaction in stress

Stress and NA

A substantial number of studies report that NA content (tissue concentration) in most brain regions decreased immediately after acute stress under various conditions that would provoke the fight/flight reaction.[35, 36] Because the content of 3-methoxy-4-ethylenglycol (MHPG), the major metabolite of NA after its release, is always increased, suggesting increased NA turnover during stress, this finding has been interpreted to result from NA depletion in nerve terminals after the release of NA. This interpretation has been confirmed directly via microdialysis.[37] These biochemical studies strongly support the view that the NA neurons in the CNS mediate the alarm reaction during stress in a manner similar to those in the peripheral nervous system.[35, 36]

Furthermore, experimental data suggest that the central NAergic system amplifies the emotional response to stress.[36] In particular, anxiety, fear and rage are emotions that are closely related to stress.[38] The emotion of anxiety is experienced to prepare for unknown stressor. Fear and rage are emotions that facilitate flight (or freeze) and fight responses, respectively, to concrete threatening stimuli. Typically, anxiety can be regarded as a slight, undifferentiated form of fear or rage.

NA and anxiety

As early as 1961, Holmberg and Gershorn reported that after yohimbine injection in human subjects, participants exhibited ‘tense and anxious facial expressions’, and feelings of ‘irritability’, ‘tremulousness’, ‘impatience’ and ‘unrest’ in association with an increased heart rate and increased blood pressure.[39] This observation suggests that anxiety, irritability and emotional instability are consequences of overactivity of the central NAergic system.

Additional quantitative studies were conducted later to examine further the anxiogenic effect of yohimbine in human subjects.[40, 41] In these studies, increased anxiety on a subjective anxiety scale was observed to co-occur with increased sympathetic activity, as evidenced by elevated systolic blood pressure, piloerection and the plasma MHPG level. Additionally, these studies reported that these changes were antagonized by clonidine or benzodiazepines.

Yohimbine's anxiogenic effect has been demonstrated in rats using an elevated plus-maze test, which is a paradigm used to test anxiety levels in animals.[42] Yohimbine reduced the time required to enter the open arm and the percentage of time spent in the open arm. These changes were antagonized by clonidine. Cocaine, which potentiates DA and NA activity, is a substance known not only to have reward effects and addictive potential but also anxiogenic effects. The anxiogenic effect of cocaine has also been demonstrated using the elevated plus-maze test.[43] This effect of cocaine is blocked by propranolol or disulfiram, an inhibitor of dopamine-β-hydroxylase (DBH), the enzyme that converts DA to NA.[44] In addition, this effect was not found to occur in DBH-knockout mice.[44]

Influence on fear

The startle response to air puffs is augmented by d-amphetamine administration, and the habituation of the response is prolonged, whereas these changes are antagonized by pretreatment with FLA-63, which is an NA synthesis inhibitor.[45] Davis et al. have demonstrated that the amplitude of the acoustic startle response is increased by ‘conditioned fear’ (i.e. by presenting a cue previously paired with a footshock).[46] This fear-potentiated startle is augmented by the systemic administration of NA agonists, such as yohimbine and piperoxan, whereas this conditioned fear response is reduced by the systemic administration of NA antagonists, such as clonidine and propranolol.[46, 47] More recently, Neophytou et al. demonstrated the diminution of fear-related behaviors by bilateral LC destruction.[48] These results suggest that LC and non-LC systems are related to the emotional response. Notably, it has been reported that the microinfusion of clonidine into the amygdala or into the bed nucleus of the stria terminalis, both of which are crucial parts of the anxiety/fear circuit,[38, 49] also reduces the augmentation of the startle response to conditioned fear.[49-51]

Influence on aggressiveness

When 6-OHDA is administered intraventricularly into the brain, it initially destroys the CA neurons selectively, particularly their terminals (the denervation phase), but later, it causes a compensation for the reduction of CA terminals, such as postsynaptic supersensitivity (the compensation phase).[19] In the denervation phase, animals become hypersomnic and insensitive to external stimuli. However, in the compensation phase, animals become irritable and display emotional instability, such as an exaggerated reaction to slight stressors.[52] This phenomenon, which is known as ‘passive aggressiveness’,[52] exhibits the following pattern: Without external stimuli, the animals are quiet and lack spontaneity. However, when the animals are stimulated by certain slightly stressful stimuli, such as a rod touch, an air puff,[52] crowded housing[53] or the removal of their pups,[54] they suddenly become extremely agitated and aggressive. Pharmacological studies have revealed that the enhancement of shock-induced aggression after intraventricular 6-OHDA treatment derives from the denervation supersensitivity of the central NA system.[55, 56]

Psychophysiological changes

Traditionally, psychophysiologists have assumed that there is an arousal level associated with vigilance or alertness during waking.[57, 58] In classic psychophysiology, measurement of sympathetic activities was often employed to estimate the arousal level.[59] Of these activities, electrodermal activity, which measures palm perspiration as a proxy for mental alertness, was a representative measure because palm perspiration sensitively reflected arousal levels in a wide range of wakeful states.[57, 59] However, the development of computer technology is opening a new field in psychophysiology: computerized EEG analysis. In this section, the influence of central NAergic activity on electrodermal activity and EEG analysis will be briefly reviewed.

Influence on electrodermal activity

The skin conductance (SC) level increases in proportion to sweat gland activity. The skin conductance response (SCR) represents the phasic increase in the SC in response to a novel stimulus and typically tends to habituate with the repetition of the stimulus. SF is the spontaneous fluctuation of SC in the absence of external stimuli. Based on a series of clinical studies on electrodermal activity, Lader discovered a strong correlation between the habituation rate of the SCR and the frequency of SF and observed that both are good indices of the arousal level.[57, 60]

In animal studies, it has been reported that the systemic administration of clonidine accelerates the habituation of the SCR and reduces the SF frequency, whereas yohimbine prolongs the SCR habituation and increases the SF frequency.[61] These effects of clonidine and yohimbine are interpreted as being mediated by central NA activity, because in the peripheral sympathetic nervous system, the post-ganglionic innervation of the sweat gland is exceptionally cholinergic. In addition, the destruction of the ascending NA bundle by 6-OHDA microinjection completely obliterates the SCR and SF, whereas the destruction of the DA system does not change SC activity.[62] The habituation failure of the SCR, which occurs during the compensation phase after intraventricular 6-OHDA,[63] has been demonstrated to be caused by NAergic supersensitivity.[64] The reason why basic studies on electrodermal activity in unanesthetized animals are rare may be that training animals to remain quiet during recording is difficult.

Influence on EEG

EEG frequency analysis: Using simultaneous recording of cortical EEG and LC unit activity in rats, Berridge et al. demonstrated that the microinfusion of clonidine to the vicinity of the LC could inactivate LC neurons and shift the EEG pattern from low-amplitude, high-frequency to large-amplitude, slow-wave activity.[65] In contrast, bethanechol, which is a cholinergic agonist, infused into the area that surrounds LC neurons could activate the neurons and shift the EEG pattern from low-frequency, high-amplitude to high-frequency, low-amplitude activity.[66]

Evoked potential (EP): The EP represents the response of the brain to certain sensory stimulus and is extracted from the EEG background using averaging. It is expected that central NAergic activation (or alertness) should accelerate information processing in the CNS. In reality, the phase-advance effect of the systemic administration of yohimbine on the neuronal response in the auditory cortex and on the auditory EP to a tone stimulus has been demonstrated in rats.[67] In humans, it is reported that systemic yohimbine administration shortens peak latencies of auditory EP, whereas systemic clonidine administration prolongs the EP latencies.[68]

P3 potential: Regarding event-related potentials (ERP), there is substantial evidence to suggest that the P3 potential reflects the phasic activation of the LC (see review[69]). Using operant conditioning of the oddball visual discrimination task in monkeys, Aston-Jones et al. discovered that LC neurons are phasically and selectively activated in response to an infrequent (10–20%) target cue that was randomly intermixed with non-target cues. This condition that activates LC neurons is identical to that which elicits the P3 potential in the human ERP.[32] In addition, Pineda et al. have observed that electrolytic lesion of the LC decreases the P3-like component of scalp ERP in monkeys conditioned to an oddball task.[70] These findings suggest that the P3 component of the ERP could reflect the phasic activation of the LC.

The pharmacological suppression of NA neurons by the systemic administration of clonidine also reduces the P3 amplitude in monkeys[71] and humans.[72] However, the effects of tonic activation of the NA system by pharmacological means on the P3 amplitude have been inconsistent,[72, 73] which may reflect the inverted U-shaped correlation between tonic arousal levels and the phasic elevation of arousal.[70, 74] Therefore, decrease in the P3 potential can be interpreted to result from tonic underactivity or overactivity of the NA system, whereas increase in the P3 potential suggests phasic overactivity of the system.

Based on the experimental findings described above regarding NAergic dysfunction, the following inference seems plausible (Table 1). Overactivity of the NA system increases sympathetic tone and arousal level and amplifies emotional reaction to stress. This overactivity could manifest itself as a cluster of symptoms, such as insomnia, anxiety, irritability, emotional instability and augmentation of fear or aggressive reaction. In psychiatry, these symptoms can be included in the concept of so-called ‘hyperarousal symptoms’. In contrast, the underactivity of the system decreases sympathetic tone and arousal level, and attenuates stress reactivity, which could result in a cluster of symptoms, such as hypersomnia, behavioral sedation and insensitivity to stress (‘hypoarousal symptoms’). The psychophysiological abnormalities caused by NAergic dysfunction are also summarized in Table 1.

Table 1. Two symptom clusters caused by overactivity and underactivity of the central noradrenergic system
Central noradrenergic activityOveractiveUnderactive
  1. Overactivity not only increases sympathetic tone and the arousal level but also amplifies sensitivity to stress, resulting in insomnia, anxiety, emotional instability, agitation and an increased tendency to fear or aggression (hyperarousal symptoms). Underactivity not only decreases sympathetic tone and arousal level but also attenuates stress sensitivity, resulting in hypersomnia and blunted reactivity (hypoarousal symptoms). Psychophysiological changes caused by the noradrenergic dysfunction are also shown in regard to electrodermal activity and electroencephalography.
Sympathetic activityIncreaseDecrease
Arousal levelHyperarousalUnderarousal
SleepInsomniaHypersomnia
EmotionSensitive, anxious, instableInsensitive, sedated
 → Agitation, fear or aggression→ Blunted affect, apathy
Electordermal activity  
 Skin conductance response (SCR)Slow habituationFast habituation
 → Habituation failure→ No response
 Spontaneous fluctuation (SF)High frequencyLow frequency
  → Absent
Electroencephalography  
 Frequency analysisHigh frequency ↑Low frequency ↑
 Evoked potentialLatency reductionLatency prolongation
 Event-related potentialP3 amplitude ↑ or ↓P3 amplitude ↓
 Hyperarousal symptomsHypoarousal symptoms

Does Noradrenergic Dysfunction Exist in Psychiatric Disorders?

Does NAergic dysfunction actually occur in association with the psychiatric symptoms described above? In the following discussion of this question, the name of the disorder and the diagnostic criteria used depend on the studies cited, most of which were published after the introduction of the DSM-III.[75]

Anxiety disorders

PTSD

With the publication of the DSM-III, the concept of PTSD was introduced into psychiatry. In the diagnosis of this disorder, ‘tonic symptoms of heightened arousal’ are the decisive criterion, and ‘difficulty in falling asleep’, ‘irritability’, ‘difficulty in concentration’, ‘excessive alertness’ and ‘excessive startle response’ have been regarded as concrete examples of hyperarousal symptoms.[75, 76] In early psychophysiological studies, PTSD patients were considered to be always in a hyperaroused state. However, later psychophysiological studies were unable to confirm this observation consistently, most likely due to differences in experimental conditions. A consistent finding in PTSD patients is that a hyperaroused state is easily provoked even by mildly stressful stimuli.[77, 78] Shalev et al. demonstrated that physiological responses to loud tones as evidenced by the eye-blink, electromyogram, SCR and heart rate were larger and habituation was prolonged in PTSD patients.[79] A meta-analysis of ERP studies[80] reveals that PTSD patients consistently exhibited an enhanced P3 amplitude in response to distracting stimuli, particularly distracting stimuli related to trauma.

Geracioti et al. demonstrated that CSF concentrations of NA were tonically elevated in veterans with PTSD and that the elevation correlated with the clinical PTSD symptom scale.[81] The favorable clinical effects of clonidine, propranolol and prazosin, an α1-antagonist, have also been reported.[82-86] Although Bracha et al. reported a substantial decrease in the LC neuronal count in three postmortem brains of war-related PTSD patients, they interpreted that the hyperarousal symptoms in PTSD resulted from the upregulation of the NA biosynthetic capacity in the surviving LC neurons.[87] The pivotal role of central NAergic dysfunction in the hyperarousal symptoms of PTSD has been proposed by multiple authors.[88, 89]

Panic disorder

For numerous reasons, panic attacks can also be considered to be a phasic hyperarousal and hyperNAergic state provoked by some stimulus (often unknown). Patients who suffer from these attacks tend to exhibit high electrodermal activity, high plasma NA levels and high subjective anxiety even when not experiencing attacks. In addition, these indices increase significantly with slightly stressful stimuli, such as mental arithmetic or improvised public speech, which suggests emotional instability.[90-92] In addition, an enlargement of the P3 potential is reported among patients with ‘typical’ panic disorder.[93, 94] It is also well established that after the administration of yohimbine, these patients easily fall into a panicked state and exhibit an increased heart rate and increased plasma MHPG levels compared with normal subjects.[95] In addition, idazoxan, a more selective α2-antagonist than yohimbine, is reported to provoke panic attacks in normal subjects.[96]

Generalized anxiety disorder

Patients characterized by tonic anxiety are classified into this subgroup. Restlessness, tension, hypersensitivity, concentration difficulties, irritability, increased muscle tone and sleep disturbance, as described in the DSM-III and IV, all indicate a hyperaroused state. The sleep architecture of this disorder has been analyzed and is characterized by an increase in the time required to fall asleep, an increase in stage 1 time, a decrease in the time spent in stage 4 during non-REM sleep, and a decrease in REM density, which is an indicator of the depth of REM sleep.[97] These sleep characteristics suggest tonic NA overactivity during sleep.

Patients with generalized anxiety disorder have high plasma NA and MHPG levels.[98] However, their response to yohimbine tends to be no higher than that of normal subjects.[40] In generalized anxiety disorder, the number of platelet α2-receptors is decreased, and this phenomenon is explained by the downregulation of tonic NA release.[98] The assumption that the downregulation of α2-receptors also occurs in the CNS may explain the blunted response to yohimbine in generalized anxiety disorder.

Thus, so-called ‘anxiety disorders’ can be considered to be biological changes associated with a hyper-NAergic state. The systemic administration of benzodiazepines is known to decrease brain NA turnover.[36] The LC activity is known to receive GABAergic inhibition, and a good correlation is reported between the affinity of benzodiazepines for the GABAA receptor and the SCR habituation rate,[99] which suggests that the anxiolytic effect of benzodiazepines is mediated by their NAergic inhibition. Clonidine, prazosin, and β-blockers are known to be useful as anxiolitics.[86, 100-102] However, direct biochemical reports on brain NA levels in anxiety disorders are very limited in number. One reason for the limitation may be that the change in NA levels is small in anxiety disorders compared with other functional psychoses. As Kalk et al.[103] have noted, the NAergic pathophysiology of anxiety disorders may vary depending on the disorder's clinical sub-classification.

Mood disorders

Depression

Although the existence of depressive mood or anhedonia (with psychomotor retardation) is regarded to be important for the diagnosis of depression, depression is often accompanied by hyperarousal (or hypoarousal) symptoms.[75, 76, 104] Recently, depression with anxiety disorders (or ‘anxious depression’) is attracting attention.[78, 105, 106]

In a classic electrodermal study on depression, Lader discovered that there are two types of depression.[57, 107] In one type (‘agitated depression’), the habituation of the SCR is slow, and the SF frequency is high, whereas in the other type (‘retarded depression’), the habituation of the SCR is fast, and the SF frequency is low. This observation suggests that the former type involves a hyperaroused state and that the latter type involves a hypoaroused state.

In a biochemical study of the CSF,[108, 109] it was found that the MHPG levels in depressive patients were not elevated overall; however, the MHPG levels were elevated in cases with anxiety, agitation and insomnia. In a clinicopharmacological study, clonidine was effective for certain depressive patients.[110] In other studies, a favorable effect of yohimbine was observed on depression in the presence of desipramine, which is a selective NA reuptake inhibitor.[111]

Mania

Manic patients typically exhibit signs not only of psychomotor excitation and euphoric mood but also of hyperarousal symptoms.[76, 104] In manic patients, the CSF MHPG levels are elevated overall.[109] Swann et al.[112] reported that this elevation was particularly evident in ‘mixed mania’, and the extent of the elevation was larger than that observed in ‘agitated depression’. Post et al. demonstrated that the increased levels of NA in the CSF of manic patients correlated with measures of ‘anxiety’, ‘anger’ and ‘dysphoria’.[113] Kraepelin noted that mixed mania had an unfavorable prognosis and distinguishing between dysphoric mania and schizophrenia was often difficult.[114]

Clonidine has been shown to be effective in 50–60% of manic patients.[115, 116] The dose of clonidine effective against mania (450–900 μg/day) is intermediate between those used to treat anxiety disorder and schizophrenia.

Schizophrenia

Biological studies on schizophrenia appear to support the distinction between positive and negative symptoms.[117-121]

Psychophysiological data

The acute state of schizophrenia is typically accompanied by severe insomnia.[122, 123] In the acute phase of schizophrenia, the SC level and the SF frequency are high, and the SCR habituation is extremely slow or absent, suggesting that the patient is in a state of extreme hyperarousal. In contrast, in the defect state, the skin conductance level is extremely low, and the SF and the SCR disappear, which suggests a state of hypoarousal.[124, 125] The correlation of these changes in electrodermal activity to NAergic dysfunction is described above.

Using frequency analysis, the EEG of patients with schizophrenia who exhibit predominantly positive symptoms are often accompanied by augmented β-activity and attenuated α-activity, which suggests a hyperaroused state, whereas schizophrenics with predominantly negative symptoms exhibited augmented δ/θ-activity, which suggests a state of underarousal.[126] A recent finding on the correlation between auditory hallucination and β/γ-activity in the left auditory cortex[127, 128] is noteworthy.

Concerning the EP, the peak latencies tend to be reduced in patients with predominantly positive symptoms, whereas they are prolonged in patients with predominantly negative symptoms.[129, 130]

An inverse correlation between the negative symptom scale and the P3 amplitude is well established.[131] However, the correlation between positive symptoms and the P3 amplitude is not as consistent. Interestingly, an enlarged P3 potential is reported in ‘cycloid psychoses’ (according to Leonhard's subclassification of schizophrenia).[132]

These psychophysiological findings suggest that positive symptoms are associated with an extreme hyper-NAergic state, whereas negative symptoms are associated with a hypo-NAergic state.[133]

Psychopharmacological data

In 1963, using a pharmacological study of chlorpromazine and haloperidol, Carlsson and Lindqvist first obtained evidence that suggested the importance of the CA system for a biological understanding of schizophrenia.[134] Table 2 shows the effects of drugs that clearly affect the CA system and modulate schizophrenic symptoms (see review[133] for more details). The drugs in group A suppress the activity of NA or DA (or both), whereas the drugs in group B facilitate NA or DA activity (or both). The drugs in group A are known to ameliorate positive symptoms but do not improve negative symptoms.[118, 135-141] The drugs in group B often exacerbate positive symptoms but have favorable effects on negative symptoms.[41, 118, 142-151] In this table, ‘chlorpromazine’ represents the antipsychotics that block both NA and DA receptors, whereas ‘haloperidol’ represents the antipsychotics that are thought to block DA receptors selectively. However, this DA blockade is only the acute effect, whereas the anti-schizophrenic effect of haloperidol emerges only after chronic use. The mechanism of antipsychotic action, which is the basis of the DA hypothesis, will be discussed as follows.

Table 2. Drugs known to modulate schizophrenic symptoms
GroupSite of actionNANA,DADA
  1. In the columns ‘NA’, ‘DA’ and ‘NA, DA’, the drugs that predominantly affect the neurotransmission of NA, DA and both (NA, DA) are shown. Additionally, the site of action is also indicated. The drugs in Group A suppress NA or DA activity, whereas the drugs in Group B facilitate NA or DA activity. References for each drug are indicated by numerals.
  2. NA, noradrenaline; DA, dopamine.
AReceptorClonidine[135, 136]Chlorpromazine[139]Haloperidol[139]
Propranolol[137]
Synthesis α-methyl-p-tyrosine[140] 
StorageOxypertine[138]Reserpine[141] 
BReceptorYohimbine[41] Apomorphine[151]
Ephedrine[142]
Release Amphetamine[145, 146] 
Methylphenidate[147]
ReuptakeImipramine[143]Cocaine[148] 
Desipramine[144]
Synthesis L-DOPA[149] 
Degradation Nialamide[150] 

Positron emission tomography (PET) studies have shown that DA receptor occupancy in the brain occurs soon after haloperidol administration.[152] After haloperidol administration, the drug's effect on psychomotor excitation and its extrapyramidal side-effects appear within several hours, whereas it may take several weeks (known as the ‘therapeutic latency’[153]) to observe the drug's therapeutic effects on the positive symptoms of schizophrenia, including auditory hallucinations and delusions of persecution.

In contrast, many typical antipsychotics, for example, haloperidol, chlorpromazine, pimozide, fluphenazine and perphenazine, gradually decrease NA and MHPG levels in CSF within several weeks of the start of administration, and there is a good correlation between the decrease in NA and MHPG levels and decreases in positive symptoms, as assessed by the Comprehensive Psychopathological Rating Scale (CPRS).[154-156] In monkeys, the firing rate of the LC neurons decreases with the chronic administration of haloperidol.[157] However, the mechanism for these chronic effects remains unclear.

The anti-schizophrenic effect of apomorphine is questionable because its reported effect is too short to evaluate.[151] Therefore, this table suggests that the overactivity of the NA system rather than the DA system is related to the positive symptoms, and that the underactivity of the NA system is related to the negative symptoms. To judge from the dose of drugs required to improve positive symptoms, the NA overactivity that underlies the positive symptoms is likely large.

Biochemical evidence

These psychophysiological and psychopharmacological data suggest the hypothesis that positive and negative symptoms are associated with overactivity and underactivity of the central NA system, respectively.[133] There is biochemical evidence that appears to support this hypothesis.

With respect to CSF studies, NA elevation has been confirmed in at least six out of eight studies since 1980. In the studies in which subtyping has been performed, patients with paranoid schizophrenia exhibited the most significant NA elevation.[158, 159] In schizophrenic patients who were receiving haloperidol, van Kammen et al. demonstrated that high CSF NA levels during the medication's administration could predict a relapse after the medication was discontinued.[160] In more recent studies, Kemali et al.[161] demonstrated that high NA levels correlated with positive symptoms on the CPRS and a shift in the EEG toward hyperarousal. Post et al. reported that the CSF MHPG levels in patients with acute schizophrenia were positively correlated with Schneiderian first-rank symptoms and negatively correlated with subjective distress.[162]

However, Nyback et al. reported a negative correlation between the MHPG level and ventricular enlargement.[121, 163] Furthermore, the DBH activity in CSF tended to be low in most schizophrenic patients, although the result was not significant.[164] Based on correlation analysis, it was found that low DBH activity was associated with brain atrophy.[165] These findings suggest a correlation between negative symptoms and NAergic underactivity.

Regarding postmortem studies on NA content, seven of eight studies reviewed confirmed the original finding by Hornykiewicz's group[166, 167] that NA levels are elevated in the schizophrenic brain. However, it should be noted that the regions with significant NA elevation varied among the studies. In addition, Gay et al. observed that DBH activity was significantly elevated in the rostral subdivision of the LC in those with paranoid schizophrenia.[168] However, in elderly schizophrenic patients, a marked decrease in DBH activity compared with an age-matched control was reported.[169] A recent finding, using stereological morphometry, that perikaryon volume of LC neurons is increased on average in those with chronic schizophrenica[170] may have some relationship to these biochemical abnormalities.

In summary, these psychopharmacological, psychophysiological and biochemical data support the hypothesis that extreme overactivity of the NA system underlies the positive symptoms of schizophrenia, whereas underactivity of the system is related to its negative symptoms (see reviews[171, 172] for more details). According to the Positive and Negative Symptom Scale (PANSS) by Kay et al.,[173] positive symptoms include tension, suspiciousness, hostility and excitation as well as delusions and hallucinations, whereas negative symptoms include blunted affect, lack of spontaneity and emotional withdrawal. Suspiciousness is a result of supersensitivity to stressful scenario. The delusions and hallucinations in the acute phase of schizophrenia are often accompanied by the subjective feeling that he or she is in an extremely stressful situation.

As is well known, the two syndromes often co-exist.[174] However, the actual mechanism that underlies their coexistence remains unclear. Because the brain possesses strong compensatory abilities, the coexistence of overactivity and underactivity should be common, as is the case in most neurological diseases.

Patients with schizophrenia often have symptoms of other psychiatric disorders, particularly mood disorders. This phenomenon was described in detail by Bleuler[175] and is currently reflected in the ‘general psychopathology scale’ of the PANSS. Additionally, the involvement of other aminergic dysfunctions should be considered in the pathophysiology of schizophrenia.

Others

It is possible that ‘essential’ hypertension and stress-induced insomnia, both of which are frequently observed among adults in industrialized societies, may also be a result of hyper-NAergic dysfunction. These disorders often co-exist, and NA antagonists are effective for both.

In patients with essential hypertension, overactivity of the sympathetic nervous system is usually present.[176] In addition, NA spillover from subcortical brain regions into the minor jugular vein has been found to be significantly higher in patients with essential hypertension than in normal subjects,[177] which suggests a correlation between essential hypertension and overactivity of the non-LC system.

In animals, footshock stress, which causes delayed NA elevation (described later in more detail[178]), causes resistance to drug-induced sleep 1 day after the stress. This abnormality can be prevented by pretreatment with clonidine (0.1 mg/kg, i.m.) prior to the stress.[179] A comparison of the time course between stress-induced NA elevations[178] and stress-induced insomnia[179] may suggest that the latter is a result of NA elevation in the LC system rather than in the non-LC system.

The importance of NAergic dysfunction in the pathophysiology of behavioral and psychological symptoms of dementia (BPDSD) has also been reviewed.[180]

Excessive Stress Causes Long-term Noradrenergic Dysfunction

Although acute stress results in decreased NA levels,[36, 181] chronic stress is reported to increase brain NA content.[182-184] Two possibilities could explain this discrepancy. The mechanism that causes NA elevation after chronic stress may differ from the mechanism that is activated after acute stress. Alternatively, NA elevation may in fact be caused by acute stress but hidden by the initial depletion immediately after the stress.

Shinba et al. observed a long-term effect of acute stress on brain NA levels.[178] In their study, rats were exposed to intermittent intense footshock stress for 1 h, and the NA content in three brain regions was measured for up to 7 days after the stressor. Confirming previous reports, the hypothalamic NA content was decreased immediately after the stress but recovered within 24 h. A significant NA increase was then noted 7 days after the footshock. In the cerebral cortex and the hippocampus, an increase in NA content was observed even 1 day after the stress and lasted for at least 7 days. These results indicate that an increase in brain NA content can be caused by acute stress when the stress is intense enough, although the emergence of the NA increase is delayed.

Several interesting studies on tyrosine hydroxylase (TH), which is the rate-limiting enzyme in NA biosynthesis, are helpful in understanding the mechanism of this stress-induced NA increase. Salzman and Roth demonstrated that electrical stimulation of the LC activates TH, which results in an enhancement of NA synthesis in the cerebral cortex and the hippocampus of rats.[185] Under more natural conditions, footshock stress could enhance NA biosynthesis in the rat brain as a result of TH activation.[186, 187] Furthermore, accumulating evidence indicates that an overactivity of NA neurons also triggers TH induction.[188] Using pheochromocytoma cell culture, Killbourne and Sabban[189] discovered that membrane depolarization elevates TH mRNA in the cell. Furthermore, this elevation is inhibited by chelation of Ca2+ with ethylenglycotetraacetic acid (EGTA), which suggests that intracellular Ca2+ mediates the induction of TH.

Although TH is the rate-limiting enzyme in NA biosynthesis, the alteration of TH activity is not the only factor that contributes to the delayed elevation of NA after stress. Reports have demonstrated the stress-triggered gene transcription of DBH and guanosine 5'-triphosphate (GTP) cyclohydroxylase I, a rate-limiting cofactor for NA synthesis.[190] Changes in other mechanisms, such as turnover and reuptake, are also known to contribute to NA elevation.[191] All of these adaptive mechanisms could contribute to the overuse-induced accumulation of NA after acute intense stress and possibly after chronic or repeated stress.

The relationship among stress, NA neuronal activity, TH activity (or level) and NA levels is illustrated in Figure 1. This correlation suggests that NA accumulation frequently occurs after excessive stress in the human brain. Although this change may be an adaptive change, it may also result in maladaptation. The delayed elevation of NA appears to explain the summation effect of stress when the stressors are repeated or when they are of various modalities.

Figure 1.

Schematic drawing of the hypothesis that excessive stress can cause overuse-induced oversynthesis of brain noradrenaline (NA). Stress elevates NAergic neuronal activity in the brain (LC activity), which results in the activation and induction of NA synthesizing enzymes, such as tyrosine-hydroxylase (TH). Sequential changes in the levels of brain NA and its metabolite, 3-methoxy-4-ethylenglycol (MHPG), are also illustrated. This change is expected to be reversible. N, normal level.

In a study using forced-running stress with a rotating drum for 2 weeks, Hatotani et al. found that half of the female rats exhibited extreme reduction in spontaneous movement (hypomotility) and lacked a menstrual cycle several weeks after the stress.[192] The rats were considered to represent an animal model of depression. Additionally, it was reported that the model rats showed evidence of insomnia, irritability, aggressiveness and social isolation.[193] In the model rats, NA florescence was enhanced in the LC.[192, 193] However, in spite of the increase of NA content, NA turnover was observed to be decreased.[194] In addition, the degeneration of NA nerve terminals was observed in the model rats.[195]

Recently, Olson et al. reported a PTSD mode in mice, in which sensitization to social aggressiveness developed by repetition of stress reminders can be normalized by clonidine or prazosine.[196]

In the pathophysiology of somatic disease, the overactivity and underactivity of a system typically display strong mutual correlations and can often co-exist in the setting of compensation. A mechanism must exist to inhibit unnecessary or exhaustive overactivity. The decreased turnover of NA[194] and the degeneration of NA terminals[195] may be examples of such an adaptive change. The LC neuronal loss,[87] the downregulation of NA receptors[197] and the moderating effect of endogenous opioids[198] may also be examples of a change in the opposite direction. Thus, excessive stress may cause not only hyper-NAergic dysfunction but also hypo-NAergic dysfunction.

These findings suggest that central NAergic dysfunction can be caused by excessive stress, such as psychosomatic diseases (e.g. a peptic ulcer). However, stress is not only the result of the environment but may also be caused by an interaction between the environment and the individual. There may be substantial individual variation in the vulnerability to NAergic dysfunction after stress, as confirmed in animal experiments, for example in a depression model by Hatotani et al.[192] and in a PTSD model by Olson et al.[196] In addition to predisposition, heredity, endocrinological conditions and brain lesion,[199] indirect internal factors that may make adaptation difficult, such as personality problems and personal values, could contribute to the vulnerability.

Discussion

Experimental studies suggest that the central NA system, like the peripheral NA system, mediates the alarm reaction in stress. Overactivity of the system increases the arousal level and amplifies emotional reaction to stress, which could manifest as insomnia, anxiety, irritability, emotional instability and exaggeration of fear or aggressiveness (hyperarousal symptoms). In contrast, underactivity of the system appears to decrease the arousal level, and to attenuate stress reactivity, and could result in hypersomnia and insensitivity to stress (hypoarousal symptoms). Clinical data support the idea that NAergic dysfunction is in fact associated with the arousal symptoms described above, which are often observed in functional psychoses. In addition, the findings from animal experiments suggest that excessive stress causes the long-term NAergic dysfunction.

As reviewed above, a substantial number of studies already exist on the correlation between NAergic dysfunction and psychiatric disorders from different nosological classifications. The original point of this paper is to present a pathophysiological view that NAergic dysfunction is related not to one but to many disorders within the functional psychosis. The stress-induced hyperarousal symptom is a frequently observed and early sign of functional psychoses, which suggests that the role of NAergic dysfunction may be of primary importance in the development of the psychoses. The causative potential of stress for NAergic dysfunction suggests that stress is not related solely to PTSD but may be related more broadly to other functional psychoses.

A similar idea has been proposed previously in a Japanese paper by Yamamoto and Shinba.[200] The present review is an updated version of that paper in English.

After the introduction of operational diagnostic criteria, such as the DSM-III in 1980, the nosological psychiatry proposed by Kraepelin[201] became well recognized. Certainly, the use of operational diagnostic criteria has provided clinicians with a common language in psychiatry and has encouraged much biological research. However, after this 3-decade trial of nosological psychiatry or ‘Neo-Kraepelinism’, many questions concerning this diagnostic system are arising. Notably, recent developments in molecular genetics question the dichotomy between schizophrenia and mood disorder.[202] Furthermore, in actual practice this nosological diagnosis is not useful for pharmacological therapy.[203]

In somatic disease, pathophysiological diagnosis is occasionally more useful for treatment than nosological diagnosis. In pathophysiology, the dysfunction of a system is understood based on the system's normal function. The pathophysiological view also suggests that a psychiatric disorder can develop from the normal state and can become another psychiatric disorder, thus raising doubts concerning the absolute belief in discrete disease entities within psychiatry, particularly with regard to functional psychosis. Prior to Kraepelin's nosological psychiatry, most psychiatric disorders were understood dynamically via functional states, as represented by Griesinger in the 19th century.[204] This paper is a review only of the role of NAergic dysfunction in psychiatric disorders. Further research on psychopathology related to other monoaminergic dysfunction will enrich the pathophysiological understanding of functional psychosis in the future. The question of which biological factors produce the variation observed in functional psychosis is also important.

Of course, the central NAergic system may not be the only system which modulates arousal level. For example, studies suggest that cholinergic neurons and histaminergic neurons in the CNS play a crucial role in controlling consciousness state.[205-207] Therefore, the hyper- (or hypo-) arousal symptoms in functional psychoses might be explained by dysfunction in these other transmitter systems. However, selective blockers of these transmitters, such as acetylcholine blockers and histamine blockers, are not used as psychotropic medications to treat arousal symptoms in clinical psychiatry. These transmitters might contribute to other aspects of wakefulness. In addition, the reasons described in this review seem to suggest the importance of NAergic dysfunction in explaining abnormal arousal symptoms. However, other possibilities should not be completely excluded at present.

There may be some functional differentiation between the LC and non-LC systems. NA metabolic turnover appears to differ between the LC and non-LC systems. A particularly interesting finding is the difference between the two systems with regard to distributional changes in NA turnover under controllable and uncontrollable stress conditions, even when physically identical stress stimuli are used. Tsuda and Tanaka[208] demonstrated that in the process of learning to avoid footshock (controllable stress), NA turnover in the LC system, such as in the cerebral cortex, was more evident than in yoked rats unable to control the stress stimulus. Alternatively, the turnover in the non-LC system, such as in the hypothalamus, in the yoked uncontrollable rats was more evident compared with controllable rats. However, once the shock-avoidance response was well established, the turnover of NA in controllable animals became less than that of uncontrollable animals in both NA systems. In addition, using microdialysis, McQuade and Stanford[209] observed that when a tone stimulus was used as an aversive stimulus in a Pavlovian conditioning paradigm, the tone stimulus alone resulted in NA release from the cerebral cortex but not the hypothalamus.

These findings, combined with the differences in fiber projection, suggest that the functional difference between the two NA systems is that the non-LC system may be related to the physical aspect of the alarm reaction in stress, whereas the LC system may be related to the mental aspect of the reaction.

This functional differentiation between the two NA systems suggests the important role of the LC system in psychiatric disorders. The importance of the LC in functional psychoses has been proposed previously by Svensson.[210] In the future, this idea should be investigated further from the perspective of biological psychiatry.

If NAergic dysfunction exists in the brains of psychiatric patients, its existence and extent will eventually be detected objectively. As reviewed above, psychophysiological testing, including electrodermal measurements, nocturnal polygraph recordings and computer analysis of EEG, will be increasingly important in determining this dysfunction's existence. Recently, the computerized analysis of heart rate variability, which was developed in internal medicine to measure the net activity of the sympathetic and parasympathetic nervous systems independently, has been introduced into psychiatry.[211] In the future, monoamine metabolism in CSF may be analyzed directly in treatment-resistant cases. However, to date, the most practical and reliable method to assess monoaminergic dysfunction is to carefully observe each patient's responsiveness and resistance to each medication using up-to-date knowledge of psychopharmacology. Additionally, taking a pathophysiological perspective to elucidate the role of NAergic dysfunction in psychiatry will contribute to the development of psychotropic medications.

Acknowledgments

We would like to thank Professor Shoji Nakamura (Department of Neuroscience, Yamaguchi University, Graduate School of Medicine) and Professor J. Allan Hobson (Department of Psychiatry, Harvard Medical School) for their valuable comments. We also thank Dr Yoko Hoshi (Tokyo Metropolitan Institute of Medical Science) for her kindness in preparing this manuscript. This study was funded by Tokyo Institute of Psychiatry (Stress Disorder Research Project) and the Japanese Ministry of Science and Education (No.18591322). No author has any conflict of interest.

Ancillary