Are There Connections between the Neuroendocrine and Dopamine Systems in Bipolar Disorder?

Abstract

Background

Bipolar disorder (BPD) represents a frequent and disabling disease, characterized by the occurrence of extreme mood swings leading to episodes of depression or mania. Although dysfunctions in dopamine (DA) neurotransmission are increasingly recognized as key determinants of BPD, little attention has been given to the biological factors which may shape such a cyclicity of mania and depression.

Summary

We propose that BPD may result, at least in part, from skewed connections between the neuroendocrine system, the DA system, and the hypothalamic clock center. First, we provide a brief description of the hypothalamic-pituitary complex, i.e., the core anatomical structure of the neuroendocrine system. We then review clinical data demonstrating the frequent onset of BPD at menopause, during postpartum or in peri-pubertal periods, suggesting that hormonal changes, under neuroendocrine regulation, may favor the clinical expression of BPD. Finally, we revisit the DA hypothesis of BPD and propose that both the hypothalamic clock center and the hypothalamic-pituitary axis exert rheostat effects on the regulation of mood by DA. Thus, in individuals with a genetically determined predisposition to BPD, an alteration of such rheostat functions may translate into a hyper- or hypo-activity of the DA system. Potential therapeutic implications and future research directions are discussed.

Key Messages

BPD may be related to altered connections between the DA system, the neuroendocrine system and the hypothalamic clock center. We hope this article will provide a basis for future interactions between endocrinologists, neurobiologists, and psychiatrists.

Introduction

Bipolar disorder (BPD) is a common, chronic, and serious disease with a prevalence estimated at 1–4% of the general population [1]. Symptoms generally appear in late adolescence, between the ages of 15 and 25 [2]. Due to its early onset and chronic evolution, BPD not only profoundly impacts the personal life of patients but represents a major public health issue with important socio-economic consequences. BPD is among the 10 most disabling diseases [3]. In particular, mortality is increased as compared to the general population, notably because the mortality rate due to suicide is 13–30 times higher in patients with BPD [4, 5].

BPD has been known since Greco-Roman antiquity. Hippocrates (460–377 BC) wrote, “If fear or sadness persists for a long time, it is due to melancholia” [6]. Since that time, this disorder has also been associated with creativity and genius. Indeed, Aristote (384–322 BC) raised the question: “Why is it that all those who have been exceptional men, in philosophy, state science, poetry or the arts, are obviously melancholic? He further added that “no great genius has ever existed without a strain of madness.” Similarly, Aretaeus of Cappadocia (100 or 200 AD) linked melancholia and mania and noted that melancholia precedes mania [7].

The nosology of BPD was clarified in the 19th century, with contributions by two French neuropsychiatrists: Esquirol (1772–1840), who established the role of heredity and the seasonal nature of mood disorders [8], and Falret (1794–1870), who further focused on the role of heredity and described “circular madness” [9, 10]. Later, a German psychiatrist, Kraepelin (1856–1926), proposed the expression “manic-depressive illness (madness)” in the sixth edition of his treatise on psychiatry [11]. Subsequently, through the influence of psychoanalysts, in particular Melanie Klein, herself depressive, the term “madness” was abandoned in favor of “thymic disorder,” which corresponds to a “global personality way of being.” In the reference book for psychiatric disorders classification (DSM) the term “manic-depressive psychosis” was abandoned starting DSM-IV [12]. Nowadays, different clinical forms of BPD are currently described in DSM-V [13]. To reflect such a clinical heterogeneity, some authors use the term “bipolar spectrum disorder.” While acknowledging the clinical heterogeneity of this group of conditions, we chose here to stick to the general and unifying term of BPD. Despite the advent of molecular psychiatry, the physiopathology of BPD is still poorly understood. Recent studies have focused on genetics and neuroinflammation. These aspects are excluded from the present work.

A role for neuroendocrinology in BPD was first mentioned in 2004, in a review entitled “Neuroendocrinology in mood disorders” [14]. The authors concluded that, “The neuroendocrine system, which plays an important role in regulation of mood, is dysfunctional in patients suffering from mood disorders…additional research is needed to define clinical implications of neuroendocrine dysfunction in mood disorders.” Indeed, the seasonal rhythms of mood, the cyclic characters of clinical signs, the onset at puberty or menopause, and the similarities between BPD and some of the psychiatric symptoms associated with endocrine pathologies, suggest that BPD may involve dysregulations of the neuroendocrine system.

Recently, endocrinologists appear to have rediscovered that patients with pituitary lactotroph tumors treated with dopamine (DA) agonists may experience episodes of mania [15]. Neurologists have also pointed out that 20–45% of patients with Parkinson’s disease treated by L-DOPA or DA agonists manifest mania or impulsive control disorders [16, 17], similar to the manic episodes seen in BPD.

In the present article, we first provide a general reminder regarding the anatomy and physiology of the neuroendocrine and DA systems. We then reviewed the main data showing that alterations of both systems are critically involved in the pathophysiology of BPD. Finally, we propose a data-driven hypothesis suggesting that a dysfunction of the circadian clock may underlie and connect the pathophysiological processes affecting the DA and neuroendocrine systems in BPD. Overall, this general review is intended to provide a basis for future and hopefully fruitful interactions between endocrinologists, neurobiologists and psychiatrists.

The Neuroendocrine System and the Hypothalamus-Pituitary Complex

The concept of neuroendocrinology emerged in the 1940s, notably with the founding works performed by Berta Scharrer on insect models [18]. The term “neuroendocrinology” refers to an integrative system which bears both endocrine and neural characteristics and exerts endocrine and brain functions via a specific set of molecules. Hypothalamic neuroendocrine cells secrete hypothalamic neuropeptides (also called hypothalamic neurohormones) and have characteristics of both neurons and endocrine cells. Such cells establish connections with neurons via classical neurotransmitters and with endocrine cells via neurohormones.

The hypothalamus-pituitary complex, shown in Figure 1a, b, is the core anatomical structure of the neuroendocrine system. Indeed, the hypothalamus and the pituitary form a unique entity, first described by Geoffrey Harris, who established the anatomical and functional bases of neuroendocrinology [19]. The pituitary gland is divided into two main parts: an endocrine part termed the anterior pituitary or adenohypophysis, and a nervous part termed the posterior pituitary (or neurohypophysis), which is an expansion of the hypothalamus. The hypothalamus-pituitary complex also comprises two small size areas endowed with endocrine functions: the intermediate part (inconstant), between the anterior and the posterior parts, and the tuberal part, around the stalk.

Fig. 1.

Hypothalamus-pituitary complex in humans: anatomical structure of the neuroendocrine system. a Hypothalamo-pituitary connections (modified from [21]). Hypothalamo-posterior pituitary connections: hypothalamic neuroendocrine cells of supraoptic nucleus (SON) secreting antidiuretic hormone (ADH) and oxytocin (OX); endings of the hypothalamo-hypophysial tracts within the arterio-venous capillary network of the posterior pituitary. Hypothalamo-anterior pituitary connections, via the arterial portal system: neuroendocrine cells secreting dopamine (DA) and 3′ gonadotrophin-releasing hormone (GnRH) in the ARN. Axon terminals in contact with the capillaries of the first capillary network. Long and short capillaries. Portal vessels between the first capillary network in the stalk and the second capillary network in the anterior pituitary. Red arrow indicates the direction of the blood flow in the arterial portal system. The intermediate and the tuberal parts are not shown. b Hypothalamic nuclei involved in BPD pathophysiology. PON, preoptical nucleus (GnRH secretion); SON, supraoptic nucleus (ADH/OX secretions); SCN, suprachiasmatic nucleus (clock center); PVN, paraventricular nucleus (CRH/TRH/SS secretions) in anterior hypothalamus; ARN, arcuate nucleus (DA/GHRH/GnRH secretions); DMN, dorsomedial nucleus (GnRH/TRH secretions); VMN, ventro medial nucleus in medial hypothalamus; MM, mammillary nucleus, PMN, premammillary nucleus in posterior hypothalamus; SM, sulcus monroe; HLP, hypothalamolimbic pathways (connections between hypothalamic nuclei and brain structures belonging to the limbic system); TRH, thyrotrophin-releasing hormone.

The human pituitary is located in the sella turcica, a depression of the sphenoid bone. Anatomically, the hypothalamus is quite a small region (around 1 cm³), located on the floor and the side walls of the third ventricle. It is connected to the pituitary by the stalk. Some of the neuropeptides synthesized by hypothalamic neuroendocrine cells are released in the general circulation via the capillary network of the posterior pituitary. This is the case for oxytocin (OX) and the antidiuretic hormone (ADH) which are thus acting at distance from the hypothalamus-pituitary complex. Other hypothalamic neuroendocrine cells synthesize neuropeptides, which are delivered via the capillaries of the arterial portal system and regulate the endocrine cells of the anterior pituitary. This vascular system, shown in Figure 1a, is divided into two networks: the primary or first network (or plexus) in the pituitary stalk and the secondary network, in the anterior pituitary itself, between endocrine cells.

The hypothalamic neuropeptides were isolated and synthesized by the research groups led by Andrew Schally and Roger Guillemin, between 1974 and 1983 [20, 21]. In 1977, these researchers, along with Rosalyn Yalow, were awarded the Nobel Prize for these discoveries. As early as 1974, Roger Guillemin noted that, “These peptides of hypothalamic origin will thus have to be carefully studied in psychiatric medicine” [20]. Of particular interest in the context of the present article, the first immunohistochemical identification of hypothalamic gonadotrophin-releasing hormone GnRH-secreting neuroendocrine cells was reported in 1973 by Julien Barry and Maurice Dubois [22]. Ten years later, the immunolocalization of cells secreting corticotropin-releasing hormone (CRH) was performed by Pelletier et al. [23].

The main hypothalamic nuclei formed by neuroendocrine cells, as well as their respective secretions, are shown in Figure 1b. The hypothalamic neuropeptides secreted by these neuroendocrine cells regulate the hormonal secretory activity of endocrine cells located in the anterior pituitary. Thus, hypothalamic CRH stimulates the secretion of adreno-corticotrophin hormone by corticotroph cells, thyrotrophin-releasing hormone stimulates the secretion of TSH (thyroid-stimulating hormone) by thyrotroph cells, and gonadotrophin-releasing hormone (GnRH) stimulates the secretion of follicle-stimulating hormone and luteinizing hormone (LH) by gonadotroph cells. In turn, the above-mentioned pituitary hormones regulate the following peripheral endocrine glands: adrenals, thyroid, and gonads. Of note, DA can be also considered a hypothalamic neurohormone since it regulates the pituitary synthesis of prolactin (PRL) in the postpartum period (see below). Interestingly, DA and GnRH in the arcuate nucleus (ARN) and CRH in the paraventricular nucleus (PVN) all display a cyclic secretion. In fact, the clock center, which drives the circadian and seasonal rhythms, is located in the suprachiasmatic nucleus (SCN), close to the ARN. DA neurons of the ARN exhibit a synchronized oscillating activity [24] and exert paracrine effects on neurons and neuroendocrine cells located in the hypothalamus [25]. Moreover, hypothalamic DA neurons establish synaptic connections with the OX-secreting neurons of the PVN [26], a neuronal subpopulation involved notably in social behavior [27].

Overall, hypothalamic neuropeptides are involved in a large array of physiological processes [28]. In this review, we will focus only on the hypothalamic neuropeptides potentially involved in the pathophysiology of BPD. Special emphasis will be put on the roles of GnRH and CRH.

Neuroendocrine Dysregulations in BPD

GnRH and Sex Hormones

The secretion of sex hormones is controlled by hypothalamic GnRH-secreting cells. During childhood, the GnRH system is quiescent and becomes functional only at puberty. In adults, GnRH and LH are released in a typical pulsatile fashion. In females, the pulse frequency varies across the ovarian cycle. During menopause, due to the deprivation of gonadal steroids and the absence of their negative feedback on hypothalamic GnRH-secreting cells, a high frequency of GnRH/LH pulse is constantly maintained [29]. Clinical observations listed below show that, in vulnerable individuals, physiological fluctuations in sex hormones can trigger episodes of BPD.

With regard to the menstrual cycle, 3–8% of all women and a subset of women with BPD experience varying degrees of premenstrual mood alterations [30, 31]. It should be underscored that premenstrual behavioral symptoms may negatively influence marital relationships and family life, in a larger proportion than social parameters [32].

There is also a strong link between BPD and the risk of postpartum episodes of mood disorder [33, 34]. The period of highest risk for the onset of such symptoms is the first 2–4 weeks after delivery [35, 36]. In patients with BPD, the risk of postpartum relapse varied from 35% [35] to 75% in a retrospective study performed on a cohort of treatment-naive patients [37]. Depression is more frequently reported [38], although (hypo)manic symptoms may occur in 9.6–20.4% of women in the early puerperium [39]. With regard to postpartum depression, it has been suggested that estrogen-responsive genes might be abnormally silenced by the physiological fall in circulating estrogen levels [40]. However, whether and how such a mechanism may also trigger (hypo)manic symptoms remains elusive.

Regarding menopause, it is well-established that the onset of BPD symptoms, especially depression, frequently occurs in the menopausal period, mainly in pre-and perimenopause [41], as confirmed in more recent studies [42, 43]. Again, these mood symptoms correlate with a dramatic fall in estrogen levels during the premenopausal period.

Finally, it is important to keep in mind that 75% of patients with BPD exhibit symptoms, mainly depressive episodes, during childhood in the peri-pubertal period [44]. Interestingly, peripartum events such as maternal stress, prematurity and caesarean delivery are known risk factors for BPD [45]. More generally, stressful events during infancy increase the risk of developing BPD in adolescence [46].

In conclusion, during the main periods of reproductive life, alterations of GnRH secretion and/or cyclicity may be implicated in BPD, which requires further investigations. In particular, the interplay between GnRH and the DA system should be explored.

CRH, Adreno-Corticotrophin Hormone, and Cortisol

In Cushing’s syndrome, hypercortisolism is due to an adrenal adenoma or to a corticotroph pituitary tumor. Since the 1980s, the presence of psychiatric symptoms in patients with Cushing’s syndrome has been underlined. In historical series, it was reported an incidence range from 35% [47] to 65% [48–51]. The most frequent symptoms are irritability (86%) and depression (76%) while mania and hypomania are rare (5–10%). The seasonal characters of such symptoms and their direct link with hypercortisolism have been reported [52]. Nevertheless, it should be noticed that in some patients, notably children, depression and irritability may remain unchanged after correction of hypercortisolism [53, 54].

Patients with BPD also display increased plasma cortisol levels [55, 56] and the severity of BPD-associated manic symptoms has been correlated to cortisol levels [56]. In fact, alterations of the hypothalamic-pituitary adrenal axis (HPA) have been previously documented in BPD [57] and the role of stress as a trigger of mood swing episodes is clearly established [58, 59]. Confirming and extending these findings, a recent study showed that in patients with BPD, hair cortisol levels exhibit abnormal months-scale fluctuations which are preceding the occurrence of mood swing episodes [60]. The authors notably concluded that patients with BPD harbor a high emotional reactivity translating into high and mood-modifying amplitudes of cortisol fluctuations. Further investigations are needed in order to integrate such observations with the current neurotransmission-based hypotheses of BPD. More specifically, as stated below, future research should aim at deciphering the impact of hormones, notably cortisol, on dopaminergic neurotransmission.

The DA System

Among the different neurotransmitters involved in the pathophysiology of BPD, i.e., mainly DA, serotonin and GABA, we will focus on DA. We provide literature-based evidence that, as compared to other neurotransmitters, DA is more specifically linked to the neuroendocrine systems and to the hypothalamic clock center, which both play key roles in BPD pathophysiology.

DA is both a neurohormone and a neurotransmitter. As a hypothalamic neurohormone, it is mainly synthesized by the DA neurons of the ARN, transported along axons and released into the circulation of the hypothalamo-pituitary portal system. DA is then transported to the anterior pituitary where it inhibits PRL secretion by the pituitary lactotroph cells. Of note, recent studies demonstrated that DA is also released within the ARN and acts there locally as a genuine neurotransmitter regulating the activity of both DA neurons, via autocrine mechanisms [25] and non-DA neurons, via paracrine mechanisms [61]. Furthermore, DA neurons of the ARN exhibit a synchronized oscillating activity [62] and establish conventional synaptic connections with the OX-secreting neurons of the PVN [24]. This point is of particular interest since OX is involved in social behavior [63] and the control of mood [64].

As a brain neurotransmitter, DA is synthesized by neurons localized mainly in the ventral tegmental area (VTA), the substantia nigra, the hippocampus and the accumbens nucleus of the mesolimbic system. DA is involved in multiple brain functions including reward, cognition, locomotion, and learning. The DA interneuronal synapse is well characterized. It should be underlined that, as observed for other neurotransmitters, presynaptic axon terminals of DA neurons may both release and reuptake DA. As shown in Figure 2, DA in the presynaptic terminal is loaded into cytoplasmic vesicles via the membrane vesicle monoamine transporter protein (VMAT). The release of DA into the synaptic cleft is triggered by an action potential. Of note, DA reuptake into the presynaptic bouton occurs via the membrane DA transporter (DAT), also known as solute carrier 6A3, encoded by the SLC6A3 gene [65]. Thanks to this reuptake, DA is rapidly renewed in the presynaptic bouton. DA-containing vesicles in presynaptic terminals are thus originating either from the soma of DA neurons or from a recycling reuptake of DA.

Fig. 2.

DA synapse. Main abnormalities suggested in BPD: abnormal expression or mutation of DAT, abnormal density of DRs, abnormalities of the release or recapture of DA. VMAT, vesicle monoamine transporter; DAT, dopamine transporter; DR, dopamine receptor.

Both as a neurohormone and a neurotransmitter, DA acts via its binding to DA receptors (DR). There are five types of DR. In the conventional interneuronal DA synapse, D₂-D₃R are located on both the pre-and post-synaptic membranes and D₁-D₄-D₅R only on the post-synaptic membrane. Irrespective of the neuronal dopaminergic circuitry, DA binding is generally not restricted to a single DR subtype. Importantly, the effects of DA on motor control, motivation, and reward are essentially mediated via D₂R/D₃R expressed in the striatal pathway and mesolimbic system [66].

The DA Hypothesis of BPD

The DA hypothesis of BPD emerged in the 1970s [67]. Ten years later, Cookson’s [68], Swerdlow’s [69], and Willner‘s groups [70] further supported the hypothesis that DA is implicated in the pathophysiology of mania. Diehl et al. [71] suggested that “DA-active treatments have some therapeutic efficacy in mood disorders”. This hypothesis was based on clinical observations as well as rigorous experiments performed by physiology researchers.

Lessons from the Therapeutic Use of DA Agonists

DA Agonist Treatment of PRL Hypersecretion

In the 1970s, the first DA agonist of the D₂R (bromocriptine) was synthesized by Arthur Stoll [72]. For over 40 years up to the present days, other D₂R agonists (quinagolide, cabergoline) have been used to inhibit PRL secretion and prevent tumoral progression of lactotroph tumors in rat models [73] and in patients [74–76]. D₂R agonists have also been used for inhibiting postpartum lactation. The first 2 cases of mania in patients treated by D₂R agonist were reported in the 1980s [77, 78]. Clinical studies are scarce and most are only case reports. In 1984, Turner et al. [79] reported 8 cases out of a total 600 patients (1.3%) treated with high doses of D₂R agonist for lactotroph tumors or acromegaly. Besides mania, which remains rare [80], psychiatric side effects of D₂R agonists, in the context of lactotroph tumors or postpartum galactorrhea [81], include impulse control disorders such as gambling, compulsive shopping and, especially in men, hypersexuality [15]. Such symptoms, reminiscent to BPD clinical signs, were usually reversible after stopping DA treatment [15]. A relationship with the dose and the duration of DA therapy was not clearly demonstrated [15]. This important point needs specific studies. Recently, the neuropsychiatric side effects of DA therapy, mainly impulse control disorders have been reevaluated [82, 83]. In particular, two recent multicenter studies demonstrated that, in patients with prolactinoma or non-functioning pituitary adenomas (NFPAs), DA agonist treatment was accompanied by impulsive control disorders (compulsive shopping and/or hypersexuality, especially in men) in 17%–52% of patients, with no manic episode reported [84, 85].

On a historical side, supporting the hypomania-promoting effects of DA agonists, the term “madness” was coined in the middle ages to designate a whole of neurological and psychiatric symptoms including hallucinations, mania and convulsions. At that time, such a set of symptoms has been associated with accusations of sorcery or bewitchment. Since then, it was demonstrated that most cases of “madness” were secondary to intoxication by rye flour contaminated with the fungus “Claviceps purpurea.” Indeed, such a fungus, also named “ergot,” contains the alkaloid “ergotamine,” a DA agonist [86, 87].

DA Agonist Treatment of Parkinson’s Disease

It is well known that Parkinson’s disease (PD) is related to a DA deficiency due to the degeneration of substantia nigra DA neurons. Untreated patients present frequently (16–42%) with depression and apathy [88, 89]. On another hand, recent studies unraveled that 20–45% of the patients with PD treated with DA agonists or L-DOPA exhibit symptoms of mania or impulsive control disorders [16, 17]. Interestingly, a recent meta-analysis demonstrated the impact of high doses of DA agonist on the occurrence of such side effects [90]. Similar findings were reported regarding the links between high doses of L-DOPA and the development of impulsive disorders in patients with Parkinson’s disease [91]. Finally, apathy is efficiently treated with the non-ergot D₂/D₃R agonist piribedil in patients with PD having undergone subthalamic nucleus deep brain stimulation [89].

Experimental Data Obtained in BPD Animal Models

In the 1990s, Willner and his group developed a very interesting BPD model in rats: CMS (chronic mild stress)-induced anhedonia [70]. In this rat model, mild stressors such as overnight illumination, cage tilt, and/or change of cage mate are applied chronically and induce symptoms of anhedonia, analogous to those observed in BPD. Under those experimental conditions, rats display a decreased sensitivity to rewards, which can be measured by a substantial decrease in their consumption of a weak sucrose solution [70]. Importantly, although CMS-induced behavioral deficits may be maintained for several months, Wilner’s team found that a normal behavior could be restored by a chronic treatment with the D₃R agonist pramipexole [92]. Moreover, in this model, an increased DA concentration was measured in the nucleus accumbens, a major component of the mesolimbic reward system [93]. Conversely, the intracerebroventricular administration of a D₂R agonist was shown to induce a mania-like behavior in mice [94]. Interestingly also, heterozygous Slc6a3 knock-down mice exhibiting a roughly 50% decrease of brain DAT levels were found to display a seasonal switch between mania-like and depressive-like symptoms [95]. The authors concluded that such a behavioral phenotype was stemming from an inability to adjust DAT levels to photoperiod-induced modifications of the neurotransmitter landscape, notably in hypothalamic neurons.

Data Obtained in Patients with BPD

In the last 20 years, many studies in humans have demonstrated that DA transmission increases during mania and decreases during depression in patients with BPD [96–98]. In vivo quantification showed an increase of D₂R [99, 100] and a decrease of DAT levels in the caudate nucleus [101]. Postmortem studies showed an increased density of D₂R [102, 103] and an up-regulation of D₁R mRNA levels [104] in the hippocampus of patients with BPD. A polymorphism of the D₁R gene has also been reported [105, 106], while gene mutations in the DR have not been confirmed. DAT (SLC6A3) plays a critical role in the regulation of DA availability [107]. It is the site of action of amphetamines, which increase synaptic DA by inhibiting DA reuptake [25]. Polymorphisms [108], mutations [109], a gene duplication, and a deletion in DAT have all been identified in depression and BPD. Importantly, similarly to DAT, the function exerted by the presynaptic D₂R is part of the inhibitory response to curb excessive DA release.

In addition to mood and behavioral changes, patients with BPD suffer from a variety of neurocognitive deficits, in domains that include executive functioning, vigilance, impulsivity, and decision-making [110–112]. Although the neurobiological basis of this persistent cognitive impairment is not fully understood, abnormal dopaminergic transmission is suspected. Furthermore, enhancement of dopaminergic activity has been shown to be a useful strategy aimed at improving cognition in BPD.

In summary, the experimental data obtained in animal models, the alterations observed in patients with BPD and the BPD-like symptoms induced by DA replacement therapies strongly suggest that the DA system plays a key role in BPD. Therapeutic implications are discussed below in the “future directions” paragraph.

Connections between the DA and Neuroendocrine Systems in BPD: Is the Hypothalamic Clock Center the Missing Link?

DA synapses are tightly regulated by various hormones including androgens [113, 114], estrogens [115, 116], and glucocorticoids [117]. In particular, estrogens were found to stimulate synaptic release of DA [118] and to decrease the expression of DAT by astrocytes, thus delaying DA reuptake [119]. Also of note, DRs are expressed within the PVN [120], and DA fibers not only connect CRH neurons in the PVN but regulate their secretory activity via D₁-D₂R [121]. While direct regulatory effects of hormones on the DA system need to be further studied in patients with BPD, we would like to propose that the hypothalamic clock center is underlying a major share of the interactions between the neuroendocrine and DA systems. Indeed, diurnal variations in mood, alterations of the sleep/wake cycle, cyclicity, and seasonality of recurrence are typically observed in BPD. This led to speculate that the circadian system, a key regulator of the HPA axis, may be a crucial pathophysiological determinant of BPD. Below, we review elements of the literature indicating that the circadian clock center, localized in close vicinity with the ARN dopamine nucleus, is disturbed in BPD and might be responsible for the concomitant and possibly synergistic alterations of both the DA and neuroendocrine systems in patients with BPD.

Circadian Rhythms and the Clock Regulation

Circadian rhythms regulating physiological, psychological, and behavioral processes are generated by the cyclic activity of a biological clock center, located in the hypothalamic SCN, as shown in Figure 1b. The light/dark cycle resets the rhythmicity of SCN activity which, in turns, regulates the rhythmicity of melatonin secretion by the pineal gland. Thus, in the presence of light, the output from the retino-hypothalamic tract inhibits melatonin synthesis. In that sense, melatonin should be considered the hormone of the night. By providing the organism with information regarding darkness, melatonin plays the role of an endogenous synchronizer, able to stabilize circadian rhythms, reinforce them, and maintain their mutual phase-relationship. Melatonin exerts a direct feedback effect on SCN neurons and instructs their firing rate, metabolic activity and expression of the clock genes CLOCK and BMAL1 [122]. Such feedback control is mediated via the G protein-coupled melatonin receptor (MT1). Importantly, besides the SCN, MT1 and the clock molecular machinery are expressed in various brain areas [123, 124], including those linked to dopaminergic neurotransmission (see below).

The sleep/wake cycle is also controlled by the clock center which regulates sleep characteristics, sleep time as well as sleep depth and duration [125]. A reduced need for sleep is a recognized symptom of manic episodes, while insomnia or hypersomnia is a symptom of major depression [13]. Importantly, alterations of the sleep-wake cycle are more than only symptoms of manic or depressive episodes. In particular, a peculiar circadian chronotype could be associated with BPD. The term “chronotype” refers to the individual trait-like preference (morningness vs eveningness) defined by a person’s preferred time to conduct daily activities or the timing of their sleep. Using the Composite Scale of Morningness (CSM) for circadian typology and Mood Disorder Questionnaire (MDQ), it was demonstrated in 8,562 students that the evening-type group was more likely to experience hypomanic symptoms [126]. Along this line, disruptions in the sleep/wake cycle were suggested to trigger manic or depressive episodes in patients with BPD [127]. In a recent authoritative review, it was concluded that the most prominent sleep-wake cycle abnormality involved in BPD could be a delayed circadian phase [128]. Finally, it has been also suggested that, in vulnerable individuals, life stress affects sleep/wake and social rhythms, leading to disruptions of the circadian rhythms and the subsequent triggering of depressive or manic episodes [129].

Connections between the DA System and Circadian Rhythms

As mentioned above, clock genes are not only expressed in the SCN, i.e., the brain master clock structure, but also in populations of neurons located outside the SCN. This is notably the case for midbrain dopaminergic neurons [130]. In these DA neurons, clock genes were shown to imprint autonomous oscillatory waves of activity. Thus, mice lacking a functional Clock gene display alterations of DR levels, an increased excitability of dopaminergic neurons and a higher susceptibility to experimental drug addiction [130, 131]. Experimental data including SCN ablation further showed that daily changes in DA system activity are likely to be both generated intrinsically and entrained by the master brain clock (i.e., the SCN) [132, 133]. Some of the structures forming the dopaminergic reward system appear relatively autonomous in terms of oscillatory activity, whereas others exhibit a weak endogenous rhythm more or less reinforced by the SCN [134]. Along this line, recent data showed that MT1 and MT2 receptors regulate the availability of DAT at the cytoplasmic membrane, thus limiting striatal DA reuptake capacity [135].

While the light/dark cycle is essential to the control of SCN neurons, light-independent neural circuits exert also a direct influence in the circadian regulation of such neurons. In particular, SCN neurons exhibit a dopaminergic innervation originating from the VTA [136]. Moreover, the expression of Drd1 by SCN neurons is indispensable for a properly timed synchronization between activity rhythms and light/dark cycles [137]. Further documenting the effects of DA on SCN functions, a recent key article showed that sleep/wake infradian cycles of 48–72 h can be induced in mice by experimentally increasing the levels of DA in the SCN [138]. Notably, the chemogenetic activation of VTA dopaminergic neurons, which project for part of them in the SCN, was sufficient to trigger such alterations of the circadian rhythms. Moreover, a subset of patients with BPD was shown to exhibit mood and sleep-wake infradian rhythms with periods of 48 h and beyond [138]. Supporting these findings, genetic investigations, including genome-wide association studies, demonstrated that polymorphisms linked to genes of the clock molecular machinery confer a higher risk of developing BPD [139, 140]. Thus, well-coordinated neuronal communications between the dopaminergic and circadian systems are likely necessary for appropriately timed behavioral responses and adaptation to the environment. In patients with BPD, an in-depth and pluridisciplinary reassessment of biological rhythms is needed. Longitudinal studies including parallel analyses of sleep architecture and hormonal cycles would be of particular interest.

Future Directions

Regulating the DA and Circadian Systems for the Treatment of BPD

DA Agonists for the Treatment of BPD

Soon after the discovery of pharmacological D₂R agonists, two original articles published, respectively, in 1979 and 1984 reported that, at a dose of 2.5 mg–40 mg/day, the D₂R agonist (bromocriptine) improved the symptoms of depression in 20/25 patients with BPD, within the first 2 weeks [141, 142]. Later on, in 2000, the efficacy of pramipexole, a D₂-D₃R agonist, in the treatment of patients with uni-and bipolar depression was reported [143]. Studies from Aiken [144] and Tundo [145] underlined the efficacy of pramipexole treatment and the low percentage of (hypo)manic switch. Two recent meta-analyses gathering data obtained from cohorts of 504 patients [146] and 281 patients [147] demonstrated that the overall response and remission rates under pramipexole treatment were 62.5% and 48%, respectively. There was no significant difference regarding the rates of short-term response or remission between patients with BPD and patients with major depressive disorder (MDD). It should be noticed that, besides open-label trials and observational studies, randomized clinical trials comparing pramipexole to placebo also demonstrated the efficacy of this treatment. This was shown in 22 patients (12 pramipexole and 10 placebo) with MMD [148], 21 patients (12 pramipexole and 10 placebo) with BPD in depressive phase [149], and 60 patients (30 pramipexole and 30 placebo) with MMD [150]. The results were similar, with a therapeutic response of >50% in 67-60-40% of patients under treatment in the three trials, respectively, and 20-9-27% of patients under placebo (p = 0.04; 0.02; 0.03). The mean doses were low (0.69 mg/day to 2.18 mg/day) and the mean duration of follow-up was 5 years. Pramipexole treatment was used in patients aged 40–60 years, with MDD or BPD, in association or not with various thymoregulators. Accordingly, in a yet unpublished study by one of the authors (H.E.), a cohort of 100 patients (71 unipolar and 29 bipolar patients) were treated with low doses of pramipexole (0.18–1.4 mg/day), associated with a thymoregulator. As shown in the online supplementary data (available at https://doi.org/10.1159/000547042), 75 patients were found to reach remission (49 unipolar and 26 bipolar). Such a finding suggests that for the treatment of BPD, as in endocrinology practice, DA agonists should be prescribed at low doses, adapted in the clinic status and, in many patients, associated with thymoregulators. This promising treatment needs to be tested in large cohorts of selected patients with typical BPDs, in comparison with placebo and/or other therapeutic agents.

DA Antagonists for Treatment of BPD

DA antagonists such as risperidone, quetiapine, aripiprazole, asenapine, ziprasidone and cariprazine are used not only for the resolution of acute mania but also as maintenance treatments [151, 152]. The fact that DA antagonists may also be used for the treatment of bipolar depression is somehow counterintuitive. The mixed activities exerted by such drugs on distinct neurotransmission systems, notably the serotonin and adrenergic systems, may explain, at least in part, such a matter of fact [98]. Moreover, some drugs referred to as DA antagonists, for e.g., cariprazine, are indeed DA partial agonists, which means that, depending on drug dosages and the actual DA levels in the synaptic cleft, such molecules may act either as D2/D3 agonists or D2/D3 antagonists [153–155]. Interestingly also, the mood stabilizers lithium and sodium valproate (VA) were shown to reduce DA neurotransmission [156, 157].

Overall, while alterations of DA neurotransmission are likely central to BPD pathophysiological dynamics, it appears that treatments should aim at reaching a state of physiological DA neurotransmission. As previously stated by Ashok et al. [98], “a failure of dopamine receptor and transporter homeostasis might underlie the pathophysiology of BPD.” Determining whether, when and at which doses DA agonists vs DA antagonists should be prescribed is undeniably a challenging task.

In conclusion, new biological approaches assessing the circadian system are urgently needed for the clinical management of patients with BPD, especially in the therapeutic follow-up and the choice of adjunct thymoregulators. The exploration of sleep architecture by polysomnography [158], under treatment with thymoregulators, may be of particular interest. Indeed, several works unraveled the effects of thymoregulators, in particular lithium and VA, on the circadian system. In mice, VA normalizes the circadian disturbances caused by elevated DA levels [159] and lithium ameliorates the mania-like behavior and HPA axis alterations induced by sleep deprivation [160]. These findings suggest that mood stabilizers are also regulators of the circadian rhythms. Finally, in the current therapeutic arsenal against BPD, a drug specifically aimed at restoring the circadian system is still missing. The use of melatonin in BPD has been promising but requires further investigations [161]. New generation pharmacological clock regulators come of age and should be tested in the context of BPD [162, 163].

Exploring the Connections between the Hypothalamus and the Mesolimbic System in Animal Models and BPD

The dopaminergic mesolimbic system includes multiple brain areas: hippocampus, VTA, nucleus accumbens, substantia nigra, cingulate cortex, mammillary bodies, amygdala. As mentioned above and further exposed below, alterations of the DA system have been documented in BPD. In particular, in animal models of BPD, increased DA levels were measured in the nucleus accumbens [93, 164]. In patients with BPD, in vivo quantification showed increased brain D₂R density [99, 100], an altered D₂R expression pattern in the nucleus accumbens and hippocampus [102, 165] as well as a decrease of DAT mRNA or protein levels in the caudate nucleus [100, 101]. Postmortem studies showed an increased D₂R density [102, 103] and an increased D₁R mRNA expression [104] in the hippocampus. Recently also, using in vivo MRI, the ENIGMA BPD Working Group showed a smaller hippocampal volume and altered white matter microstructure in the cingulate cortex [166].

However, while the DA system itself has been extensively studied in BPD, to our knowledge, the connections between the mesolimbic dopaminergic system and the hypothalamus have never been explored in BPD. This is all the more important that DA neurons of the VTA innervate the SCN [136–138] and that, conversely, hypothalamic neurons project toward mesolimbic dopaminergic targets [167, 168]. Future studies should aim at investigating if and how such connections are altered in patients with BPD.

Performing Neuropathological Investigations in Patients with BPD

There are relatively few studies assessing neuropathological alterations in the brain of patients with psychiatric disorders, including BPD. This is due to the limited number of autopsies including neuropathological assessment in this patient group [169]. The sizes of cohorts are thus small, leading to a significant selection bias. Based on this limited amount of data, Harrison et al. [170] in 2020 concluded, nevertheless, that no specific neuropathological lesions characterize BPD. It should be noted that neuropathological assessments carried out on postmortem brain tissues suffer from many pitfalls. The autopsy might have been carried out under treatment that alters the outcome. Furthermore, pre-analytic (postmortem time) and post-analytic (fixation time) alterations related to the neuropathological procedures might significantly influence the results. Postmortem assessment and analysis of brain tissues obtained from subjects that had suffered from a psychiatric disorder, including BPD, are certainly warranted but require rigorous sampling strategies, representative and substantial samples sizes, well-defined and well-characterized disease entities, and the choice of relevant analytical parameters. The neuropathological assessment of hypothalamus is certainly of major interest in psychiatric disorders and particularly in BPD. Being part of the basal forebrain, hypothalamus can easily be sampled, but it is not routinely performed. More specifically, studying the nucleus accumbens as well as the ARN with state-of-the-art molecular approached would be of great interest. In any case, we encourage the sampling and analysis of human brains with these studies in mind. Comparing observations obtained in animal models, especially those of Willner’s team [171] vs human pathology is indeed of crucial interest.

Conclusion

In this review, we revisited the DA hypothesis of BPD and propose that regulation of mood by DA is cyclically altered, in connection with cyclic alterations of the hypothalamic clock center and the neuroendocrine system. Thus, BPD pathophysiology may rely on cyclic, combined, and synergistic alterations of the clock center, the neuroendocrine system and the DA system.

More specifically, we suggest that both the hypothalamic clock center and the hypothalamic-pituitary axis exert rheostat effects on the regulation of mood by DA. In individuals with a genetically determined predisposition to BPD, alterations of these rheostat functions may result in abnormal amplitudes of DA-mediated mood oscillations. In turn, an abnormal DA neurotransmission is likely to translate in an imbalanced or improperly timed regulation of biological rhythms, leading notably to neuroendocrine dysfunctions. In this scheme, the occurrence of depressive or manic episodes could thus depend on triggers determined by dopaminergic, neuroendocrine and/or sleep/wake cycle cues.

In future studies aimed at assessing the efficacy of DA agonists as a preventive strategy in BPD, we suggest the prescription of low doses, as in endocrinology practice. To reach a state of physiological DA neurotransmission, the concomitant use of thymoregulators appears also crucial. In any case, this review should encourage an interdisciplinary approach of BPD in order to unravel new and clinically relevant links between endocrinology, neuroendocrinology, and psychiatry.

Acknowledgments

We would like to thank M.F. Belin and G. Raverot for the critical reading of the manuscript. M. Guillen and B. Valdenaire for formatting the references, and E. Josié for drawing the figures. This review is dedicated to the patients with BPD and to their family. We would like to share with them this hypothesis that bipolar disorders might be, in part, related to neuroendocrine dysfunctions. It is also dedicated to Roger Guillemin, who died recently (1924–2024) and supported the hypothesis that “neuroendocrinology may be involved in some psychiatric diseases.”

Conflict of Interest Statement

The authors declare that there is no conflict of interest regarding the content of this article.

Funding Sources

This research received no external funding.

Author Contributions

Conceptualization: J.T., I.A., S.N., and B.C.; figures: J.T.; drawn from the original articles: J.T., S.N., and B.C.; review and editing of the original draft: I.A., J.L., and E.H. All authors have read and agreed to the published version of the manuscript.

Funding Statement

This research received no external funding.