Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

The Role of Acetylcholine Mechanisms in Affective Disorders

David S. Janowsky and David H. Overstreet


Considerable information suggests a role for monoaminergic-cholinergic balance in the pathogenesis of mood disorders. As proposed by Janowsky et al. (43, 44), depression may be a manifestation of a central cholinergic predominance, whereas mania, conversely, may be due to a relative monoaminergic (i.e., adrenergic or possibly serotonergic) predominance. This chapter on cholinergic mechanisms in the affective disorders summarizes research findings from both animal and human studies that suggest that central muscarinic and possibly nicotinic mechanisms are likely to contribute to the psychopathology of the affective disorders (70)


It has long been recognized that cholinergic effects in animals are mediated by both muscarinic and nicotinic mechanisms. However, muscarinic mechanisms have been the focus of most investigations into the potential role of the central cholinergic system in affective psychopathology, and indeed muscarinic mechanisms are the major focus of this review. Nevertheless, as a part of this review, we also consider a potential role of nicotinic mechanisms in the etiology of affective disorders.

Many preclinical animal behavioral models of depression have been developed, including the self-stimulation model, the hypoactivity model, the learned helplessness model, the chronic stress model, the behavioral despair or forced swim model, and the HPA axis activation model (69, 88). Significantly, increasing central cholinergic tone with such centrally active cholinomimetic agents as physostigmine, arecoline, and oxotremorine usually induces or enhances the behavioral analogs of depression in such models of depression. Thus, centrally acting cholinomimetic drugs consistently produce behavioral inhibitory effects including lethargy and hypoactivity, activation of the HPA axis, decreases in self-stimulation (43, 54, 55), increases in behavioral despair in the forced swim test, and decreases in saccharin preference (88). These cholinergically induced phenomena are generally reversible with centrally active sympathetic agents and antimuscarinic drugs, thus supporting evidence of a balance between adrenergic and muscarinic cholinergic factors in their regulation (43).

Several animal species, developed by selective breeding, have demonstrated differentially enhanced responses to cholinergic agonists; these animals appear to represent genetic animal models of depression. The psychogenetically selected Roman low avoidance (RLA) rats, which do not effectively learn avoidance responses, are relatively more sensitive to cholinergic agonists (66) and have been considered to be an animal model for anxiety/ depression (122). The hypercholinergic Flinders sensitive line (FSL) of rats, which were selectively bred to have differentially increased responses to the anticholinesterase agent diisopropylfluorophosphate (DFP), as compared to control Flinders resistant line (FRL) rats, are differentially sensitive to other muscarinic agonists. Significantly, this cholinergic hypersensitivity occurs when the FSL rats are only several weeks old, and appears to be dissociated from muscarinic receptor upregulation (86). Overstreet and colleagues have proposed that, like the RLA rats, the FSL rats are an animal model of depression (86).

Similarities between FSL rats and depressed humans include reduced body weight, learning difficulties, reduced general activity and locomotion, increased REM sleep and reduced REM sleep latency, and exaggerated HPA axis activation (i.e., corticosterone release) after cholinomimetic administration (86). The FSL rats also demonstrate a greater reduction in saccharin preference than do control FRL rats during exposure to chronic mild stress (86), the latter suggesting a greater degree of stress induced anhedonia.

In addition to possessing a number of characteristics that parallel those of depressed humans (86), FSL rats also have baseline exaggerated immobility in the forced swim test, a test commonly used to screen for antidepressant drugs, and considered by some also to be an animal model of depression. This exaggerated immobility is not unexpected; there is literature indicating that cholinergic agonists accentuate and cholinergic antagonists reduce swim test immobility. Hasey and Hanin (40) confirmed the acute depressive effects of cholinergic agonists (i.e., physostigmine) on swim test immobility. These investigators also reported that these cholinergic effects could be partially counteracted by noradrenergic manipulations, and they proposed a balance model reminiscent of the original adrenergic/cholinergic balance model of affective disorders (43).


If an aberrant mood is caused by an imbalance between adrenergic and cholinergic factors, it is logical to expect that increases in central cholinergic activity might decrease manic symptoms. Several studies have shown that centrally active cholinergic agonists and cholinesterase inhibitors possess antimanic properties. In a seminal study by Rowntree et al. (100), the centrally active cholinesterase inhibitor DFP was given to manic–depressive patients and normals. The normal subjects and remitted manic–depressives developed irritability, lassitude, depression, apathy, and slowness and/or poverty of thoughts. Two patients who were hypomanic at the time of the study improved with DFP and continued euthymic after its administration. One hypomanic patient became less manic and was minimally depressed after each of two courses of DFP, but relapsed upon DFP withdrawal.

Janowsky et al. (43, 44, 45) found that the centrally active cholinesterase inhibitor physostigmine caused a dramatic but brief reduction in hypomanic and manic symptoms in bipolar patients. Neither placebo nor the non-centrally acting cholinesterase inhibitor neostigmine produced such changes, thus suggesting a central mechanism. After physostigmine administration, the manics became significantly less talkative, active, euphoric, happy, grandiose, and friendly, and showed a decrease in flight of ideas on the Beagle Murphy Mania Rating Scale. The effects of physostigmine lasted from 20 to 90 min. Modestin et al. (73, 74) subsequently also reported a lessening of manic symptoms following infusion of physostigmine, but not neostigmine, and Davis et al. (17) reported that physostigmine caused dramatic antimanic effects, particularly in patients with low levels of hostility and/or irritability. Carroll et al. (8) and Shopsin et al. (104) also reported a decrease in euphoria and mobility in manics after physostigmine infusion. Berger et al. (3) reported pilot data suggesting that RS86, a relatively specific muscarinic (M1) agonist, has significant antimanic effects.

Although most data to date are supportive of cholinomimetic agents exerting an antimanic effect, some authors have reported effects only on the affective and motoric components of mania, and no effects on the cognitive aspects of mania such as grandiose thinking and expansiveness (8). Skeptics of the possibility that cholinomimetics exert antimanic effects have pointed out that manic patients, treated with centrally acting cholinomimetic drugs still continue to show grandiosity and to have manic delusions (8).

In addition to the antagonism of motoric behaviors, there is evidence that when the central cholinergic system is activated, there may be a later compensatory and antagonistic activation of the adrenergic system. Fibiger et al. (29) demonstrated that the increase in cholinergic activity caused by the administration of physostigmine led to an eventual increase in locomotion in rats, as presumed adrenergic mechanisms began to exert effects. This activating effect became apparent as the cholinergic predominance induced by physostigmine decreased and was exaggerated if a centrally acting anticholinergic agent such as scopolamine was given at the beginning of the hyperactivity phase.

As with the work of Fibiger et al. (29), there is some evidence that a behavioral rebound following physostigmine infusion can also occur in humans. A case of rebounding into mania was noted by Rowntree et al. (100), and by Shopsin et al. (104), who demonstrated rebounding into hypermania in two of three manic patients given physostigmine. However, "rebounding" in humans has rarely been observed following physostigmine infusion.


Although there is considerable evidence that centrally acting muscarinic cholinergic drugs can effectively decrease manic symptoms, at least acutely, few reliable direct markers of a cholinergic deficit have been noted in mania. One measure of such function may be reflected in the measurement of erythrocyte choline activity. Erythrocyte choline levels have been investigated as a biochemical marker of acetylcholine activity since choline is the major precursor, as well as a major metabolite of acetylcholine. Changes in central choline have been shown to affect cholinergic neurotransmission. Slight elevations in erythrocyte choline have been noted in patients with bipolar disorders (115) and have also been observed in schizophrenics and depressives (6, 115). Stoll et al. (115) also noted that there exists an increased level of choline in a subgroup of manic bipolar patients at pretreatment. These patients had more severe illnesses at admission and a less desirous outcome at discharge. In addition, and of potential theoretical significance, bipolar patients with low concentrations of red blood cell choline had four times as many previous episodes of mania as they had episodes of depression. This contrasted with the finding in patients with high erythrocyte choline levels, who had similar numbers of manic and depressive episodes. Thus, those patients with low choline and presumably low central acetylcholine activity had a relatively greater predominance of manic episodes. Interestingly, patients with low choline showed no depressive symptoms in their clinical presentation. If erythrocyte choline in any way reflects central brain choline and acetylcholine activity, the above findings may be consistent with high central acetylcholine levels causing depression.


Probably the most convincing evidence that acetylcholine is involved in the regulation of the affective disorders is the observation that centrally active cholinomimetic drugs rapidly induce depressed moods. In addition to observations of depression-induction caused by DFP (100) and cholinomimetic insecticides (34), Janowsky et al. found induction and/or intensification of depressive symptoms in actively ill bipolar manic patients given physostigmine, as well as a worsening of depression in groups of unipolar depressed and schizoaffective depressed patients (45). Similarly, Davis et al. (17) and Modestin et al. (73, 74) demonstrated an increase in depression in manic patients given physostigmine. In addition, Risch et al. (99) and Nurnberger et al. (78, 80) found that depressed patients given the direct cholinergic agonist arecoline also developed depression and other forms of negative affect, including hostility and anxiety. Physostigmine also caused a depressed mood in a majority of euthymic bipolar patients maintained on lithium (84). Risch et al. (95, 98, 99) found a statistically significant increase in self- and observer-rated negative affect on the Brief Psychiatric Rating Scale (BPRS), Profile of Mood States (POMS), and the Activation-Inhibition Rating Scales in normals receiving intravenous physostigmine or arecoline. Likewise, Mohs et al. (75) reported severe depression occurring in Alzheimer's patients receiving the cholinergic agonist oxotremorine. El-Yousef et al (28) reported that normals, having smoked marijuana, became profoundly depressed after receiving physostigmine, an effect that was atropine reversible.

Evidence supportive of a role for acetylcholine in the phenomenology of affective disorders also comes from descriptions of the anergic-inhibitory behavioral effects, as opposed to the mood effects of centrally acting cholinesterase inhibitors and cholinergic agonists. These drugs induce a psychomotor retardation that is very similar to that occurring in endogenous depression. Thus, Rowntree et al. (100) and Modestin et al. (73, 74), studying normals, depressives, and manics, and Gershon and Shaw (34), observing normals, all reported that cholinesterase inhibitors exerted anergic and behavioral-inhibitory effects, as did Janowsky et al (45) in their physostigmine-treated subjects.

Depressed moods have also been observed in subjects receiving acetylcholine precursors, including deanol, choline, and lecithin. Davis et al. (18) and Tamminga et al. (117) found that depressive symptoms occurred in some schizophrenic patients who were treated with choline, a phenomenon that was atropine-reversible. In a subgroup of cases, it was noted that depressed mood was a side effect of choline and lecithin treatments employed to try to reverse the memory deficits of Alzheimer's Disease (117). Also, Casey (9) observed that a depressed mood and, in some cases, a paradoxical hypomania occurred in some deanol-treated tardive dyskinesia and other movement-disorder patients. Thus, precursors of acetylcholine may induce a depressed mood, a finding that is consistent with the adrenergic-cholinergic imbalance hypothesis.


The vast majority of evidence suggesting that increasing central cholinergic activity can induce depression and supporting the validity of an adrenergic/cholinergic balance hypothesis of affective disorders has come from the utilization of cholinergic agonist/antagonist strategies. These strategies have proven promising in suggesting a cholinergic defect in the affective disorders. However, they are all indirect indicators.

Two more direct techniques have provided evidence for a role for increased acetylcholine in depression. The first is the measurement of red blood cell choline in manic and bipolar patients as described above (115), with choline being a presumptive acetylcholine precursor. The second, clinical in vivo proton magnetic resonance spectroscopy and other imaging techniques such as PET scanning provides another means for more directly assessing presumed human cholinergic function in vivo. To this end, magnetic resonance spectroscopy has been non-invasively used to measure choline-containing substances in the brain, and these compounds may reflect acetylcholine activity. Charles et al (10) have observed that there is a state-dependent increase in choline in the brains of patients with major depression, as compared to controls. This increase in choline reverted to normal after successful drug treatment of the depression. Subsequently, Renshaw et al (91) have studied the basal ganglia of depressed and control subjects, and noted an alteration in the metabolism of cytosolic choline compounds in the basal ganglia of depressives, particularly those responsive to fluoxetine. In addition, Hankura et al (39) found that depressed Bipolar Disorder patients had higher absolute subcortical choline containing compounds than did normal subjects. They found that Bipolar Disorder patients also had higher choline/creatinine plus phosphocreatine and choline/NNA peak ratios in the depressive and euthymic states, as did patients with Major Depressive Disorder. CNS imaging of direct indicators of cholinergic function by techniques such as measurement of muscarinic and nicotinic receptor binding, ligand displacement strategies, and blood flow studies following cholinomimetic administration in affective disorder patients is in its infancy. However, the potential for applying those imaging techniques used to study dopamine and serotonin in schizophrenics and affective disorder patients to the study of acetylcholine is great and could provide more direct evidence for a role for cholinergic mechanisms in the pathophysiology of the affective disorders.


Patients with affective disorders appear to be more sensitive than normal subjects to the effects of cholinomimetics, and an inherent muscarinic receptor hypersensitivity has been proposed to underlie this hyper-reactivity. With respect to the affect-inducing and behavioral inhibitory effects of cholinomimetics, Janowsky et al. (45, 52, 54) noted that those patients with depression, mania, or schizoaffective disorders, as compared to schizophrenics without a significant mood component to their illness, became significantly sadder and more depressed after receiving physostigmine.

Oppenheimer et al. (84) likewise observed that a significant percentage of euthymic bipolar patients receiving lithium developed a depressed mood after receiving physostigmine, whereas normal controls who received physostigmine alone did not become depressed. Furthermore, Janowsky et al. (45, 49, 50, 52) found that rater-evaluated behavioral inhibition and self-rated anxiety, depression, hostility, confusion, and elation subscales of the POMS showed significantly greater increases in affect disorder patients than in other psychiatric patient groups or normals after physostigmine infusion. That physostigmine may differentiate behaviorally patients with affective disorder diagnoses from others has received further support from the work of Edelstein et al. (27). These investigators used physostigmine to differentiate schizophrenic patients who were responsive to lithium carbonate therapy from those who were not. They found that patients who responded to physostigmine with a clearing of psychotic symptoms were significantly more likely to respond to a trial of lithium, presumably because they represented an affective disorder variant. Similarly, Casey (9) noted that those tardive dyskinesia patients with a strong history of affective disorder selectively showed increased affective symptoms while receiving the presumed acetylcholine precursor deanol. Conversely, Silva et al (unpublished data) noted anergia, but no depressed mood after physostigmine infusion in a group of carefully screened normals.

Steinberg et al (113) noted that increases in negative affect after physostigmine administration occurred selectively in those personality disordered patients with pre-existing affectively unstable personalities, as compared with those who were affectively stable or predominantly had impulsive traits. In contrast, those affectively unstable patients who reacted to physostigmine with negative affect did not show mood changes following noradrenergic, serotonergic, and placebo challenges. In a somewhat similar way, Fritze et al. (30) and Fritze (31) noted that in 11 healthy male volunteers, behavioral and cardiovascular sensitivity to physostigmine correlated with irritability and emotional lability, and with habitual passive stress coping strategies. A rise in plasma cortisol correlated with "motor retardation," and an epinephrine rise correlated with active coping strategies. These authors have proposed that cholinergic supersensivity may be predominately related to stress sensitivity and coping profiles, rather than to specific affective disorder diagnoses as such.

Nevertheless, non-affective disorder subjects and normals receiving physostigmine sometimes do develop depression and other negative affects after receiving cholinomimetic drugs, and the differences between affective disorder patients and controls are not profound. Furthermore, the observation by Nurnberger (80) that a differential behavioral sensitivity was not seen in a group of euthymic affective disorder patients is suggestive of the possibility that increased behavioral sensitivity to cholinomimetics is a state, rather than a trait phenomenon.


Although not as specific as previously believed, major depression is generally associated with a series of characteristic sleep changes, including decreased rapid eye movement sleep (REM) latency and increased REM density. In parallel with these characteristics of depression, such cholinergic agonists as arecoline, physostigmine, and pilocarpine cause a shortening of REM latency and an increase in REM density (4, 5, 15). In addition, Sitaram et al (106, 108) have shown that unmasking of muscarinic up-regulation by the withdrawal of chronic scopolamine leads to a shortening of REM latency and an increase in REM density. Furthermore, Sitaram et al. (106, 108) have found that REM latency shortened significantly more following arecoline infusion in patients with an affective disorder episode, a history of affective disorder, or a family history of affective disorder than in those with no such history (107). Similarly, Gillin et al (37) replicated Sitaram's earlier work, showing enhanced cholinergic-induced REM latency shortening in depressives following arecoline infusion.

Berger et al. also found that a super-shortening of REM latency in endogenous depressives occurred following administration of the long-acting oral muscarinic agonist, RS86 (3), when compared to normals and to eating disorder patients. Possibly related to all of the above findings, Berger et al. (4) also found that physostigmine-induced arousal and awakening from sleep more frequently occurred in affective disorder patients than in normals.

Gann et al. (33) investigated sleep electroencephalogram (EEG) profiles during placebo administration and after cholinergic stimulation with RS86 in patients with major depression, healthy subjects, and patients with anxiety disorders. Like arecoline's effects in previous studies, RS86 had a more profound impact in patients with major depression with respect to sleep onset REM episodes, shortening of REM latency, increasing REM density and REM duration. Similar results with respect to REM sleep responses were noted by Riemann et al (92); and Riemann found that to a lesser extent schizophrenics show parallel findings to those noted in affective disorder patients (93, 94). Significantly, patients with anxiety disorders and associated secondary depression did not show enhanced REM abnormalities following RS86 administration. Anxiety disorder patients, in fact, showed decreased REM density compared to controls. Dube and co-authors (25) were able to show that the REM sleep response to cholinergic stimulation was significantly more pronounced in cases of primary depression, compared to patients with manic disorders and to those who constituted a mixed anxious/depressive group. Rapaport et al. (89) reported that panic disorder patients responded similarly to physostigmine as did controls, supporting the concept of cholinergic supersensitivity being unrelated to anxiety disorders.

Conversely, Poland et al (87) demonstrated that scopolamine caused a differential effect on REM density, reducing REM activity in a way consistent with a cholinergic abnormality in depressives. This finding was replicated by McCraken et al (68) in adolescents.

A controversy exists as to whether or not the supersensitivity to cholinomimetic agents demonstrated by enhanced REM shortening and increases in other REM parameters is a state or a trait phenomena. The work of Sitaram et al. (107) and Nurnberger et al (80) would suggest the latter, because remitted bipolar's showed exaggerated REM latency shortening after receiving arecoline. In contrast, Berger et al (3) noted exaggerated REM latency shortening following administration of RS86 only in actively depressed patients, and Lauriello (61) did not find a supersensitive REM latency shortening response to pilocarpine in mildly depressed patients, but did find that a greater shortening of REM latency did occur in their most symptomatic depressed patients.

The presumed hypersensitivity of REM sleep parameters to cholinomimetics appears to be genetically linked, as noted in monozygotic twin studies in which arecoline was administered to paired twins (80, 81) and caused significantly correlated changes on REM sleep parameters, and as observed in the recent work of Sitaram et al. in which affectively ill members of the families of affectively ill patients showed exaggerated shortening of REM latency after arecoline infusion (107). Schreiber et al. (103) has observed exaggerated shortening of REM latency and increased spontaneous sleep onset REM periods following RS86 administration in non-depressed first-degree relatives of patients with a DSM III diagnosis of major depression. There is evolving evidence that those then non-depressed relatives of depressives, who initially showed prominent decreases in REM latency following RS86 were those who, years later, eventually became clinically depressed (Holsboer, F. Personal communication, 1998). However, in contrast to evidence supportive of cholinergic REM related supersensitivity in depression, Gillin et al. (37) noted that depressives, withdrawn from chronically administered scopolamine did not show differential cholinergic rebound effects, as measured by sleep EEGs.


In humans, a major characteristic of depression is the apparent activation of the hypothalamic-pituitary-adrenal (HPA) axis. A variety of studies have shown that increased cortisol and adrenocorticotropic hormone (ACTH) release occurs in depressed patients, and that some depressed patients fail to suppress cortisol secretion after dexamethasone. There is evidence from a variety of studies that cholinomimetic drugs can release CRF and elevate serum ACTH and cortisol in animals, in normals, and in psychiatric patients (54, 96, 97). Also, physostigmine has been shown to reverse dexamethasone induced suppression of cortisol in normals (16, 23) and in depressives. Thus, it appears that physostigmine and other cholinomimetic induced increases in HPA axis activity occur, and that these parallel other phenomena noted in endogenous depression, such as increased cortisol secretion, cortisol resistance to suppression by dexamethasone, and elevated ACTH levels.

The secretion of b-endorphin appears linked to the secretion of ACTH, probably by a common precursor, b-lipotropin. Like ACTH and cortisol, b-endorphin and b-lipotropin are often elevated in depressives, and like ACTH and cortisol, they are released by physostigmine (54, 96, 97). Furthermore, affective disorder patients have been found to show significantly greater increases in both ACTH and b-endorphin levels after physostigmine infusion (96), when compared to normal controls and to non-affective psychiatric patients.


Acetylcholine can increase plasma growth hormone levels (67). As reviewed elsewhere (54), pilocarpine, acetylcholine, and physostigmine all have been noted to increase growth hormone release in vivo in rats, and in vitro in rat pituitaries. However, this increase is prevented and reversed by the concurrent administration of atropine or methscopolamine. Likewise, piperadine, a nicotinic cholinergic receptor agonist, has been found to enhance growth hormone secretion in humans during sleep (71), and methscopolamine, a non-centrally-active peripheral anticholinergic agent inhibits nocturnal growth hormone secretion in man (98).

Whereas work in humans by Janowsky et al. (50), Davis and Davis (16), and Risch et al. (98) did not demonstrate statistically significant growth hormone increases after physostigmine infusion in subjects pretreated with peripheral anticholinergic agents such as probanthetine and methscopolamine. O'Keane et al (83) have reported growth hormone release following administration of the cholinominotic agent pyridostigmine in depressed patients and normals, who were not pretreated with a peripheral anticholinergic drug. Their depressed patients, especially those with high baseline cortisol levels and those who were males, showed an exaggerated release of growth hormone, a finding also noted in manic patients by Dinan et al. (21, 22). Thus, enhanced growth hormone release in depressives may have been due to a supersensitive cholinergic system or possibly an increase in basal cortisol levels. With respect to specificity, Lucey et al. (64) recently reported exaggerated pyridostigmine induced growth hormone release in obsessive compulsives, and O’Keane et al (82) noted an enhanced growth hormone response to pyridostigmine in schizophrenics. However, Cooney et al (13) noted that patients with schizophrenia and those with panic disorder, associated with low depression scores, did not differ from control groups.


Sokolski and DeMet (109) have reported that the pupillary miotic response to the cholinergic agonist pilocarpine was exaggerated in patients with major depression. They suggest that a trait-dependent supersensitivity exists. These authors asserted that the supersensitive pupillary response to pilocarpine in depressives is probably mediated by M3 muscarinic receptors, working through a G protein–phosphoinositol system. Thus, their data support the notion of both peripheral and central muscarinic supersensitivity in depressives. In addition, Sokolski and DeMet (personal communication) have found that lithium and valproate acid induced improvements in manic patients were correlated with increases in pupillary sensivity to pilocarpine, and that manic patients, like depressives, showed increased pupillary sensivity to pilocarpine, a phenomena which correlated with the degree of mania.


There is evidence that patients with major depressive disorder often have increased urinary epinephrine excretion and, to a lesser extent, increased norepinephrine excretion (51). As with many parameters, physostigmine infusion causes effects in normals with respect to catecholamine release similar to those occurring in depressives. Administration of physostigmine to normals and affective disorder patients causes profound increases in serum epinephrine levels, and slight increases in serum norepinephrine levels (48). Furthermore, Janowsky et al (48) have demonstrated a blunting of the epinephrine response to physostigmine in affective disorder patients, possibly due to receptor down-regulation caused by chronically elevated epinephrine levels. Physostigmine and arecoline have both been shown to increase pulse rate and blood pressure levels in subjects treated with peripherally acting anticholinergic drugs. These changes can be profound and appear, as with cholinomimetic-induced increases in epinephrine release, to be due to muscarinic activation of centrally mediated sympathetic outflow (53), and these changes parallel preclinical results in animals (60).


Several studies have attempted to better understand the mechanisms by which the behavioral, cardiovascular, and neuroendocrine effects of cholinomimetic drugs occur. In early studies, Janowsky et al. (43) and Modestin et al. (73, 74) noted that in contrast to centrally acting physostigmine, the peripherally acting cholinesterase inhibitor, neostigmine, like placebo, did not exert any behavioral effects. This indicated that the behavioral affects of physostigmine were probably central in origin. Later, Janowsky et al (51) noted that the increases in blood pressure, pulse rate, serum epinephrine, ACTH, cortisol, and prolactin, as well as the anergia and negative affect caused by physostigmine also occurred via a central mechanism, since no such changes occurred with neostigmine. Furthermore, Janowsky et al (51) noted that the behavioral, cardiovascular, and neuroendocrine effects of physostigmine can be blocked by the centrally acting anticholinergic drug scopolamine, but not by the non-centrally acting anticholinergic drug methscopolamine, again suggesting a central focus for the above effects.


There is evidence that centrally active anticholinergic drugs have mood-elevating properties, although this evidence is generally anecdotal and mostly uncontrolled. As noted by Jellinec in 1981 (56) and Smith in 1980 (112), anti-parkinsonian drugs used to treat drug-induced parkinsonian symptoms have been reported to cause feelings of euphoria associated with a sense of well-being, increased sociability, and a reversal of depressed mood. Furthermore, one report by Coid and Strang (12) suggested that the anticholinergic agent procyclidine caused a switch into mania in a bipolar patient. In addition, several reports suggest that high doses of atropine and other anticholinergics such as ditran may alleviate depression, and one report suggests that a tricyclic-antidepressant-induced central anticholinergic syndrome may alleviate depression. Also, Kasper et al. (57) observed antidepressant effects with the anticholinergic drug biperiden, especially in patients with endogenous depression who had a non-suppressing dexamethasone suppression test.

However, in contrast, several controlled studies have indicated a lack of therapeutic efficacy for anticholinergic agents. Fritze et al (31) could not demonstrate an advantage to adding centrally acting biperidin over a non-centrally acting anticholinergic drug, both added to existing non-anticholinergic antidepressant treatments. Gillin et al (36) showed no antidepressant effect of biperidin in depressives. Such information is not supportive of a cholinergic hypothesis of depression unless novel muscarinic receptors are involved in the chronic regulation of depression. It must also be stressed that many effective antidepressant medications, such as fluoxetine and related selective serotonin reuptake inhibitors, as well as mianserin and trazodone, lack significant direct muscarinic receptor-blocking properties, and yet can alleviate depression (Also see the "Acetylcholine and Stress" section below).


In addition to the mood-depressing and inhibitory effects of centrally active cholinomimetics, antidepressant and anti-parkinsonian agent-treated patients may develop some aspects of depression, including a depressed mood, anxiety, withdrawal, agitation, and insomnia soon after discontinuing these medications (20). These symptoms are preventable and/or reversible with anticholinergic treatment, and may represent the unmasking of muscarinic receptor hypersensitivity. This issue has been elaborated in detail by Dilsaver and Greden (20), who proposed the concept of "cholinergic overdrive," consisting of anxiety, nausea, agitation, and sometimes depression. Cholinergic overdrive is described as occurring after discontinuation of tricyclics with anticholinergic side effects and with anti-Parkinsonian drugs, and especially after anticholinergic drug withdrawal combined with marijuana intoxication. Dilsaver and Greden attribute cholinergic overdrive to the unmasking of up-regulated cholinergic receptors occurring secondary to the withdrawal of muscarinic blockade, and propose that such overdrive may also activate noradrenergic receptors, leading to induction of arousal and manic symptoms in some cases (20).


Several reports suggest that many sympatholytic-antihypertensive medications, including a-methyldopa, propranolol, clonidine, and reserpine can cause depression. These drugs also have significant cholinomimetic properties (43). In humans, the CNS side effects of reserpine and other antiadrenergic-cholinomimetic antihypertensives are similar to those of centrally active cholinergic agents, and include mood depression, vivid nightmares, lethargy, and sleepiness. With respect to antipsychotic drugs, these agents, which block central dopamine and increase acetylcholine turnover, can also cause some of the components of depression in selected patients, effects that can be reversed with centrally active anticholinergic medications.


Several studies optimistically have suggested a difference among affective disorder and normal subjects with respect to muscarinic receptor binding. Nadi et al. (77) reported that fibroblasts grown in culture from affective disorder patients and their mood-disordered relatives had more muscarinic binding sites than controls, and Meyerson et al. (72) noted that samples from the frontal cortexes of individuals who committed suicide had more muscarinic receptor binding activity than from matched brains from people dying from accidents or murder. However, Kelsoe et al (58) have not been able to replicate the fibroblast receptor findings described above. Generally, at present most evidence suggests that muscarinic binding in fibroblasts and brain is not altered in affective disorder patients. It should be pointed out, however, that the FSL rats previously mentioned as a cholinergic supersensitivity model of depression also do not exhibit changes in cortical muscarinic receptors (86), but do exhibit increases in such receptors in the hippocampus and striatum. It may be that previous workers have not examined the key regions of the human brain relevant to muscarinic receptor changes in depression.


The initial adrenergic-cholinergic balance hypothesis is supported by a considerable amount of preclinical and clinical evidence. A pharmacological–behavioral model for naturally occurring adrenergic and cholinergic regulation of mood may be found in the interactions and reciprocal effects of psychostimulants (which increase catecholaminergic activity) and cholinomimetics (which increase cholinergic activity). Methylphenidate-induced psychostimulation in rats and in humans is rapidly antagonized by physostigmine, but not by neostigmine, and, conversely, physostigmine's inhibitory-depressant effects can be reversed by methylphenidate (44, 46, 47). Possibly related to the ability of physostigmine to antagonize methylphenidate-induced psychostimulation is the observation that physostigmine caused a rapid, dramatic drop in the norepinephrine metabolite serum 3-methyoxy-4-hydroxyphenylglycol (MHPG) in a manic patient, presumably reflecting a drop in CNS noradrenergic activity. This phenomenon was associated with induction of a tearful depressed state and an improvement in the manic symptoms (85).

Furthermore, a reciprocal relationship apparently may exist between a subject's response to a psychostimulant and his or her separate response to a cholinomimetic agent. A negative correlation was noted between amphetamine-induced behavioral excitation and the ability of arecoline, given on another occasion, to decrease REM latency (79). Similarly, Siever et al. (105) showed that in a mixed group of affective disorder and normal subjects, those with the most dramatic physostigmine- and arecoline-induced anergy and negative affect showed a blunted growth hormone response to the noradrenergic agonist clonidine (presumably a reflection of decreased noradrenergic responsiveness).

Schittecatte et al. (102) have also demonstrated that human depressives are subsensitive to the REM sleep suppressing effects of clonidine, thereby supporting the a-adrenergic subsensitivity postulated to exist in depressives on the basis of neuroendocrine challenge studies (1). It is not clear at this stage whether the clonidine subsensitivity is a direct reflection of changes in the a-noradrenergic system or merely a consequence of the supersensitivity proposed to exist in the balancing cholinergic systems. It could be very illuminating to obtain both clonidine and cholinomimetic challenge data in the same subjects (107).

Preclinical studies have also provided evidence in support of the monoamine–acetylcholine interaction model of affective disorders. Hasey and Hanin (40) showed that the immobility-promoting effects of the anticholinesterase, physostigmine, could be modified by manipulating the b-noradrenergic system. Also, Ikarashi et al (41) found that dopamine D2 receptor stimulation in striatum led to an inhibition in acetylcholine release. In addition, Downs et al (24) demonstrated that depletion of brain dopamine caused a supersensitive ACTH response following physostigmine administration, a finding paralleling the exaggerated ACTH response to physostigmine found in affective disorder patients (96). Consistent with the above data, imipramine, a noradrenergic antidepressant causes decreased acetylcholinesterase activity in the hippocampus, suggesting decreased acetylcholine release (7).

The adrenergic-cholinergic balance hypothesis of affective disorders can be expanded to include serotonergic-cholinergic interactions. Although the selective serotonin reuptake inhibitors do not appear to block muscarinic receptors as such, Saito et al (111) have shown that acetycholine release appears to be regulated by inhibitory 5HT1B hetero-receptors located on cholinergic nerve terminals. Similarly, Crespi et al (14) found that 5HT3 receptor agonists decrease acetycholine release by effecting 5HT3 hetero-receptors, located on cortical cholmergic nerve endings. Consistent with the above, the 5-HTIA agonist 8-OH-DPAT turned off cholinergic REM - on neurons which activate REM sleep (121). Conversely, 8-OH-DPAT enhanced acetylcholine release from rat hippocampus and cerebral cortex (32), and MKC-242, another 5HT1A agonist likewise increased extracellular acetylcholine, but via another mechanism (110). Thus, it is at least possible that the selective serotonin reuptake inhibitors, such as fluoxetine, may in part exert their effects by decreasing acetylcholine availability, or by shifting acetylcholine-serotonin balance to a serotonergic predominance.

Furthermore, monoamine-acetylcholine interactions may be mediated through a range of second-messenger systems, including phosphotidyl inositol and cyclic AMP, and via calcium and other post-synaptic mechanisms. There is growing interest in the possibility that an abnormality in one or more of these second-messenger systems may indeed contribute to the biochemical changes underlying depression (2, 26). Such a revelation may help to reconcile the differing functional and biochemical results with respect to muscarinic mechanisms, including the question of why, in depression, functional muscarinic supersensitivity has been frequently reported, but changes in muscarinic receptors have not (54). An alteration in one or more of the second-messenger systems or other post-synaptic systems could account for the muscarinic supersensitivity observed in depressives that occurs without any obvious change in receptors.


Although virtually all the stress-sensitive neurohormones are released by centrally acting cholinomimetic drugs, and many can be decreased by anticholinergic drugs, a controversy exists as to the interpretation of these phenomena, especially with respect to human studies. Davis and Davis (16) observed that in their normal subjects, serum prolactin, cortisol, and growth hormone levels did not increase after physostigmine infusion unless other unpleasant symptoms of the physostigmine response occurred, such as dizziness, nausea, or emesis. They suggested that cholinomimetic-induced increases in stress hormones may be due to nonspecific stress effects (i.e., feeling sick or distressed), rather than to a direct cholinergic mediation of stress. Motion sickness, which appears quite similar to the physostigmine response at its extreme, and which includes nausea, dizziness, and vomiting, almost certainly involves central cholinergic mechanisms (53). Motion sickness is a potent stimulator of growth hormone, prolactin, and cortisol secretion.

However, contrary to the interpretation that nonspecific stress causes cholinomimetic-induced hormonal effects, Hasey and Hanin (40) demonstrated that physostigmine caused significantly greater increases in cortisol release in rats than did neostigmine even though the peripheral toxicity of both drugs was recorded to be severe and equal to one another. Janowsky et al. (49, 50, 52, 54, 55) have reported the occurrence of increases in serum prolactin and cortisol in physostigmine-treated patients and in normals who manifested no nausea, emesis, or dizziness. In addition, Risch et al. (98, 99) have observed that in arecoline-treated subjects in whom serum b-endorphin, ACTH, and cortisol levels significantly increased, a sizable proportion of subjects could not tell when active drug and when placebo had been administered. Furthermore, Janowsky et al. (49, 51) reported that physostigmine's behavioral-anergic effects almost always precede its nauseating effects, and Raskind et al. (90) noted that increases in serum ACTH, epinephrine, and cortisol occurred in aged controls and in Alzheimer patients, whether or not nausea occurred. Finally, Rubin et al (101) has observed that very low doses of physostigmine, associated with elevated cortisol and ACTH levels, did not cause significant subjective stress or nausea, and Steinberg et al (113) found no correlations between the mood response to physostigmine and changes in plasma cortisol, prolactin or growth hormone, or to nausea or other side effects in the patients they studied. Thus, it is quite possible that the nauseating, behavioral, and neurohormonal effects of centrally acting cholinomimetic drugs occur at similar dose thresholds, and that they occur in parallel with each other, rather than causing each other.

It is possible that acetylcholine as such actually moderates the body's stress responses. Stress is multidimensional, and includes gastrointestinal, cardiovascular, behavioral, analgesic, immunological, endocrinological, psychological-behavioral, and psychopathological changes (54). Information from animal studies has accumulated suggesting that the manifestations of stress are likely to be mediated by complex alterations in central neurotransmitters, including norepinephrine, dopamine, serotonin, and gamma amino butyric acid (GABA). Acetylcholine and neurotransmitter interactions with acetycholine may also have a role in the mediation of the stress response, a role that has not been widely investigated (54). For example, Zhang and Zheng (123) have found that paraventricular stimulation causes stress induced gastric ulcerations, an effect enhanced by intraventricular acetylcholine administration.

As with other neurotransmitters, there is growing evidence that stressors can cause changes in central acetylcholine activity. A dramatic demonstration of stress–acetylcholine linkages is the work of Gilad et al. (35), which shows that stress causes an increase in acetylcholine release and compensatory down-regulation of muscarinic receptors and that choline uptake, as well as the increase in acetylcholine release occurring after exposure to stress, is differentially exaggerated in stress-sensitive rats. Other investigators have shown that in the hypothalamus, acetylcholine turnover increases after 1, 4, or 24 hr of stress. Furthermore, central acetylcholine receptor sites are increased during uncontrollable stress (53), as are acetylcholine levels. A number of more recent studies have extended the above early observations. Mizuno and Kimura (76) observed that hippocampal acetylcholine release, as well as cortisol release, was increased following stress in young, but not aged rats. Mark et al (65) using microdialysis techniques, demonstrated that inescapable stress selectively enhances extracellular acetylcholine release in rat hippocampus and prefrontal cortex. Significantly, acetylcholine release increased even more when the stress was lifted. Consistent with the above results, Day et al (19) have observed, using microdialysis techniques, that prenatally stressed rats as adults, showed greater release of hippocampal acetylcholine when exposed to a mild stress or being administered corticotropin releasing factor (CRF). The effect of CRF on hippocampal acetylcholine release was found to be due to a central, rather than a peripheral or adrenal effect of the CRF (19). Thus, there is considerable evidence that acute stress can increase central acetylcholine activity.

Consistent with stress activating central acetylcholine activity, cholinomimetic drugs cause many of the same effects as do naturally occurring stressors. These include increases in negative affect, the induction of affective symptoms, increases in stress-sensitive neuroendocrines including ACTH, cortisol, b-endorphin, growth hormone, prolactin, epinephrine, and possibly norepinephrine, increases in blood pressure and pulse rate, and increases in analgesia. These cholinomimetic effects, combined with the effect of stress on central acetylcholine activity, suggest a stress–acetylcholine linkage.

Although it has been hypothesized that increases in acetylcholine activity are fundamental to the multiple manifestations of acute stress, acetylcholine does not exert its effects in a vacuum. It is likely that the final manifestations of stress involve complex, multiple interactions involving inhibitory and excitatory neurotransmitters, such as gaba norepinephrine, serotonin, and the opioid polypeptides, which are impacted upon by acetylcholine. Also, the possibility exists that acetylcholine may be a secondary, rather than a primary, mediator of stress. However, it is probably significant that there is little or no evidence that any neurotransmitter or neuromodulator other than acetylcholine is simultaneously able to activate the cardiovascular, sympathetic, adrenal– medullary, behavioral–affective, analgesic, and neuroendocrine systems involved in the stress response. Thus, a primary role for acetylcholine in the modulation of stress is possible, and even likely, and this possibility is worthy of further investigation.


Although the vast majority of observations linking the cholinergic nervous system to affective disorders have focused on muscarinic mechanisms, there is also evidence that nicotinic cholinergic mechanisms may also be linked to depression. As described previously, the cholinesterase inhibitor, physostigmine and cholinergic agonists such as oxotremorine produce a range of behavioral and physiological effects in animals. Reduction in locomotor activity is commonly seen after physostigmine, oxotremorine, and arecoline administration and these changes are generally thought to be muscarinic in nature. In contrast, nicotine has biphasic effects, stimulating locomotor behavior at lower doses and depressing it at higher doses (11, 116). On the other hand, nicotinic and muscarinic compounds both produce dose-dependent decreases in core body temperature and may enhance memory under the appropriate experimental conditions. These drugs also have fairly similar neuroendocrine effects, particularly with respect to a stimulation of the HPA axis leading to cortisol and ACTH release.

One of the questions about the behavioral and physiological effects of nicotine is whether they are the consequence of a direct interaction of nicotine on postsynaptic nicotinic receptors or due to an indirect interaction with presynaptic nicotinic receptors leading to stimulation of (i.e., disinhibition of) the release of acetylcholine onto muscarinic receptors. The antinociceptive and hypothermic effects of nicotine (42, 86) have been attributed, at least in part, to the latter mechanism. Some of the other effects of nicotine, such as locomotor stimulation and nicotine's reinforcing effects have been proposed to be from nicotine-induced release of dopamine or other neurotransmitters (114). With respect to nicotine's effects in humans, Glassman's (38) extensive review indicates that a high rate of cigarette smoking is associated with current major depression and current depressive symptoms. He also noted that a lifetime history of major depression, even if not active at the time that a person starts to smoke, increases the chances of a person trying nicotine and becoming addicted to it and has a significant negative impact on smoking cessation efforts. This latter deleterious effect appears more pronounced in women than in men. There is also evidence that in predisposed individuals with a history of major depression, smoking cessation may precipitate severe depressive symptoms, which also appear to counteract smoking cessation efforts. Of course, the linkage between cigarette addiction, smoking, and emotional disorders is not exclusive for depression, since there are observed linkages between smoking and alcoholism, anxiety disorders, and especially schizophrenia.

Another interesting linkage between smoking and depression has been noted. Using monozygotic and dyzygotic twin pair data, Kendler et al. (59) have found that smoking and depression are indeed linked, but that smoking doesn't necessarily cause depression, and that depression doesn't necessarily cause smoking. It appears that the association of depression and smoking are linked through genetic factors that influence vulnerability to both conditions.

As noted above, nicotine's effects on dopaminergic reward mechanisms have been offered as an explanation for why depressives, as well as schizophrenics, smoke and have more trouble stopping smoking (38). Indeed, buproprion (Wellbutrin, Zyban) is an effective antidepressant agent and is effective in facilitating smoking cessation; and buproprion has dopamine agonist properties. However, it is also possible that a relationship between muscarinic and nicotinic mechanisms may be important in causing nicotine addiction. Obviously, a common neurotransmitter, acetylcholine, underlies muscarinic and nicotinic behavioral and physiological effects. It is conceivable that stress and/or depression, by activating muscarinic mechanisms leads to an overall muscarinic predominance. This may conceivably be overcome by nicotinic stimulation, leading to a relief of symptoms. Withdrawal from nicotine, allowing an unmasking of down-regulated nicotinic receptors, could lead to a muscarinic predominance and the induction of depression and dysphoria. Furthermore, there is evidence that activation of nicotinic receptors can lead to muscarinic outflow. Specifically, nicotine-induced changes in temperature and nociception can be blocked by the antimuscarinic agent, atropine (42, 86). Therefore, it is possible that nicotinic stimulation(possibly due to causing dopamine release) could lead to muscarinic activation, which would become the predominant effect when nicotine administration was stopped. A logical adjunct to utilization of nicotine replacement therapy would thus be the addition of a centrally acting anticholinergic drug to the treatment. Conversely, a drug with dopamine agonist properties, such as buproprion, would be expected to decrease cholinergic outflow.


Applications of the adrenergic-cholinergic balance hypothesis to the treatment of affective disorders has been sporadically attempted over the past several decades. As described above, anticholinergic drugs may or may not have antidepressant efficacy. Cholinomimetic treatment of mania has been more consistently rewarding. The choline precursor lecithin was used by Cohen and colleagues in the early 1980’s to treat mania, with promising results. Stoll et al (116) demonstrated that choline augmentation of lithium therapy in six rapidly cycling Bipolar Disorder patients caused a substantial reduction in manic symptoms in five, and a marked reduction of all symptoms in four. Similarly, Leiva (62) reported efficacy for phosphatidal choline in the treatment of mania. Finally, in an uncontrolled study, Burt et al (in press, Biological Psychiatry, 1999) has noted that donepezil (Aricept), 5 mg each day, proved useful in treating six of eleven treatment resistant manic patients. Thus, the use of cholinomimetic agents to treat mania appears to have therapeutic potential.


As reviewed above, considerable evidence suggests that the cholinergic nervous system, alone or acting in concert with other neurotransmitters, may have an important role in the regulation of affect. Nevertheless, as with all currently proposed biological hypotheses of the etiology of affective disorders, there exist alternative explanations, as well as some data that are inconsistent with the cholinergic hypothesis.

One question concerns the specificity of the cholinergic alterations: Are they confined to patients with affective disorder? A number of studies have failed to detect any evidence for cholinergic supersensitivity among patients with anxiety disorders (89). In contrast, several studies suggest an involvement of cholinergic mechanisms in schizophrenia. Early investigators used cholinesterase inhibitors to treat schizophrenia. Others proposed a dopaminergic/cholinergic balance and/or interactive model for schizophrenia (45, 119), and it has been suggested that cholinergic over-activity may underlie the negative symptoms of schizophrenia, such as affective flattening, anhedonia, asociality, and apathy (118, 119). Reduced REM sleep latency has also been observed in schizophrenics (92). Thus, symptoms of schizophrenia that bear some similarity to key symptoms of depression may also be mediated by cholinergic over-activity.

It is possible that pharmacologically induced changes in acetylcholine may cause perturbations in depression-relevant systems other than the cholinergic nervous system. Pharmacologically induced acetylcholine alterations could cause a "model depression" by perturbing other potential governing neurotransmitters (for example, GABA, serotonin, dopamine, or norepineprine) or second messengers in affect disorder patients. Furthermore, obviously, the fundamental biochemical changes in depression could be due to a relatively low level of central norepineprine or serotonin activity, a situation that could explain all of the above observations under the scope of a balance hypothesis. It is also conceivable that the cholinergic supersensitivity observed in many depressives may be a reflection of altered second messengers, including G proteins (2, 26, 63).

However, even if cholinomimetics can only cause a "model depression" (with such components as low mood; psychomotor depression; elevated ACTH, cortisol, b-endorphin, and epinephrine levels; as well as sleep architecture changes and increased pulse rates and blood pressure), understanding how this presumed pharmacological phenomenon occurs may ultimately offer a window into understanding the pathophysiology of affective disorders and may have useful treatment implications. Thus, at the least, understanding the implications of the mood-depressant and other depression like effects of cholinomimetics may give clues as to the actual neurobiology of affective disorders. Alternatively, it is not beyond possibility that acetylcholine actually is directly involved in the etiology and the expression of the affective disorders, alone or acting through other relevant neurotransmitters and/or second messengers.

published 2000