Depression and antidepressant drugs: Beyond a purely neurotransmitter approach

Abstract

Previously proposed mechanisms of action of antidepressants have involved effects on biogenic amines, and glutamic acid systems to explain their clinically beneficial effects. These models of neurotransmission have also attempted to explain the pathophysiology of depression. These approaches, however, have failed to explain the long latency of clinical actions of classic antidepressants or provide a convincing explanation for the pathophysiology of depression. There is a need for a paradigm shift to understand the mechanism of action of antidepressants and pathophysiology of depression. Translational research on cerebral structures that are involved in depression and short-latency antidepressant actions of N-methyl-D-aspartate receptor antagonists have provided some insights on: (1) Neuroinflammatory processes in depression; (2) The participation of neuroglia and neurotrophic factors; (3) Alterations of the functional activity of corticolimbic structures; and (4) Common antiinflammatory actions among both old and new antidepressants. Our hypothetical model consists of the following: When there is a need to cope with adversity, allostatic processes allow natural functional recovery (resilient individuals), whereas allostatic overload results in a neuroinflammatory process that leads to anxiety and depression (vulnerable individuals). Such a model encourages the development of modulators of antiinflammatory processes that involve microglia, astrocytes, and neurotrophic factors.

Key Words: Stress; Anxiety; Depression; Neuroinflammation; Cerebral plasticity; Antidepressants

Core Tip: Antidepressant drug treatments have been classified based on their actions on neurotransmitters. However, these approaches have not reached solid conclusions about their mechanisms of action, their long latency to achieve clinical efficacy, or the pathophysiological processes that are involved in depression. Recent research suggests a continuum among stress, anxiety, and depression, in which the common denominator is neuroinflammation. Notably, N-methyl-D-aspartate receptor antagonists and classic antidepressants share actions on neurotrophic factors and glia-neuronal interactions. Therefore, one promising approach for the treatment of pathological anxiety and depression may lie in the search for antiinflammatory drugs with specific actions on neural tissue.

INTRODUCTION

Behavioral, molecular, and electrophysiological techniques suggest that clinical characteristics of depression are a consequence of maladaptive responses to stress that lead to functional changes in the neuroplasticity of neuronal circuits that are related to emotion regulation[1]. There are two main systems that mediate emotional responses: The ventral system and the dorsal system. The ventral system includes the amygdala, the insula, the ventral striatum, and ventral regions of the anterior cingulate gyrus and prefrontal cortex (PFC), which are in charge of identifying the emotional significance of a stimulus and the production of an affective state. The dorsal system includes the hippocampus, dorsal regions of the anterior cingulate gyrus, and the PFC, which are in charge of regulating affective states[2]. Therefore, the dysfunction of connections among the PFC and related areas, including the amygdala, ventral striatum, pallidum, medial thalamus, hypothalamus, and periaqueductal gray and other parts of the brainstem may be considered as the basis of several psychiatric disorders[3-5].

Antidepressants share multiple sites of action. At least three kinds of antidepressants-the tricyclic desipramine, the selective serotonin reuptake inhibitor (SSRI) fluoxetine, and the N-methyl-D-aspartate (NMDA) receptor antagonist ketamine-share the property of increasing c Fos expression in deep layers of the ventromedial PFC in rats[6]. The action of ketamine in the CA3 hippocampus-medial PFC (mPFC) circuit depends on prior exposure to a stressful situation. In a control group, CA3 stimulation produced an inhibitory response in the infralimbic region, and ketamine administration after stress abolished this inhibitory response[7]. Fluoxetine and ketamine produce similar actions on lateral septal nucleus-mPFC responsivity[8]. Therefore, the mPFC and its connections are a good place to explore antidepressant actions, and most of their actions cannot be explained purely by effects on neurotransmitter systems. Ketamine exerts its actions through insulin-like growth factor 1 in the mPFC, in addition to, and in parallel with, its other actions on brain-derived neurotrophic factor (BDNF)[9]. Fluoxetine and ketamine directly bind tyrosine kinase receptor 2, leading to the facilitation of BDNF’s actions[10,11].

ANHEDONIA

The discovery that some cerebral areas are susceptible to the phenomenon of intracranial self-stimulation (ICSS) opened new ways to understand essential aspects of behavior[12]. Initially, regions that were identified as “pleasure centers” were those that connected with the medial forebrain bundle, such as the septal nucleus and PFC, in addition to the entorhinal cortex and pyriform cortex, albeit with less sensitivity[13]. The concept of pleasure centers emerged in parallel with the emerging relationship between these centers and antidepressant drug effects[14-16].

However, the concept of pleasure centers is better understood by considering the ICSS phenomenon as an adaptive process that is directed toward survival[17]. Anhedonia, or the inability to feel or find pleasure, is one of the core symptoms of depression and anxiety. The presence of anhedonia is related to a lower quality of life, sleep problems, and negative cognition[18]. Anhedonia, or the loss of behaviors that are directed toward pleasure, represents a failure in the adaptive reward system, including the anticipation, consumption, and learning of reward[19].

Although an exact map of regions that are implicated in ICSS is still incomplete, the concept of pleasure or adaptive survival centers in the brain suggests a relationship between anhedonia and stress, anxiety, and depression. The concept of hotspots may be a good approach because the same structures that are defined as being involved in the anxiety process or pleasure-adaptive centers may produce opposite actions that depend on the subregion studied and the previous state of activity of such centers[20,21].

Hedonic behavior includes the construction of reward valuation, decision-making, anticipation, and motivation. The neural basis of anhedonia and depression may correspond to functional alterations of the cerebral reward circuit[22], involving the ventral striatum, PFC regions, their afferent and efferent connections[23], and the PFC connection to the anterior paraventricular thalamus[24]. In major depressive disorder, the dysfunction of striatal structures, mainly the bilateral putamen, is related to anhedonia as demonstrated by reward anticipation tasks[25]. In humans, high distress is related to ventral striatum-medial orbital frontal cortex connectivity during reward anticipation, and anhedonia is related to greater connectivity from the ventral striatum to amygdala regions during reward anticipation[26]. In a longitudinal study, the presence of anhedonia was shown to predict depression, whereas anxiety-depression comorbidity occurred when the level of anhedonia was high[27-29]. Therefore, when the systems that are responsible for stress management and brain reward systems lose their coordination, anhedonia becomes manifest[30].

The lateral habenula is strongly activated in the absence of reward or expectation of punishment[31], and basal forebrain connections to the lateral habenula display activation in the presence of aversive stimuli and inhibition during reward stimulation[32,33], with the participation of the ventral tegmental area[34]. Recent literature considers that dysfunction of the lateral habenula may explain components of the symptomatology of depression[35]. Ketamine blocks NMDA receptors particularly in the lateral habenula[36] but not in the hippocampus[37] in rats that are subjected to chronic unpredictable mild stress, and ketamine reduces c Fos immunoreactivity in the lateral habenula and increases it in the nucleus accumbens shell[38]. Some SSRIs reduce regional cerebral metabolic rates for glucose in the hippocampus and lateral habenula[39], and imipramine increases the phosphorylation of binding proteins in the nucleus accumbens and lateral habenula, supposedly contributing to long-term actions[40].

The PFC in humans corresponds to the mPFC in rodents[41,42], and its connections have been related to depressive-like behavior in rodents[43]. Translational research has yielded similar results as human studies. A reduction of glial fibrillary acidic protein (i.e., a filament protein that is involved in the function of both activated and non-activated astroglia in the mPFC) produces anhedonia-like behavior[44]. Social stress produces anhedonia concomitant with decreases in ICSS and neurotrophic factors in the ventral tegmental area, whereas administration of the antidepressant fluoxetine or desipramine restores these alterations in stress-susceptible animals[45].

STRESS AND A CHAIN OF EVENTS

The innate and adaptive immune systems interact with neurotransmitters and neurocircuits to cope with threatening situations. In dysfunctional cases, however, it may constitute a risk for psychiatric diseases[46]. Allostasis is a functional process that maintains homeostasis, but environmental situations in some cases overload the organism’s ability to maintain homeostasis in a process called allostatic load that, in resilient individuals, normally vanishes as soon as the stressful situation resolves. When frequent activation of the allostatic process occurs, when the allostatic process is longer than the stimulus, or when several allostatic processes occur simultaneously, allostatic overload can occur[47], in vulnerable individuals (Figure 1).

Figure 1 Hypothetical model of neuroinflammatory chained events leading to pathological conditions. Under normal conditions, any stressful situation produces a reaction of emergency systems (allostasis), accompanied by a period of adaptive anxiety that allows the individual to cope successfully with the situation and produce a recovery of function (green squares: Resilient individuals). When the stressful situation is prolonged or when the functional reactions are disproportionate, allostatic overload can lead to alterations of the function of microglia and astrocytes and the release of proinflammatory factors, leading to neuroinflammation and, in turn, pathological anxiety and depression (blue squares: Vulnerable individuals). Antidepressants with different latencies restore the function of corticolimbic circuits through actions on astrocytes, microglia, and neurotrophic factors. HPA: Hypothalamus-pituitary axis.

The continuous secretion of glucocorticoids and autonomic nervous system hyperactivity that accompanies allostatic overload lead to the delivery of inflammatory cytokines, promoting disease processes[48]. Stress normally activates the hypothalamic-pituitary-adrenal axis, leading to glucocorticoid release. Allostatic load may correspond to anxiety as an adaptive process[49], whereas the pathological process of anxiety may be considered a consequence of allostatic overload. One example of allostatic overload is represented by an inadequate response that leads to the compensatory long-term high secretion of glucocorticoids, which normally exert protective actions when coping with a threatening situation but can result in higher levels of cytokines that are normally counter regulated by glucocorticoids[50]. High allostatic load may predict a higher risk of later anxiety, depression, and suicide[51].

In the long-term when confronting a stressful situation, a vicious cycle may be established through the activation of microglial cells that release proinflammatory factors that stimulate the hypothalamic-pituitary-adrenal axis[52]. The secretion of stress hormones, including glucocorticoids, by the adrenal glands coordinates the body’s response to stress. In the brain, glucocorticoid receptors are expressed by various cell types, including microglia, which are its resident immune cells that regulate stress-induced inflammatory processes. Chronic unpredictable mild stress produces anhedonia- and anxiety-like behavior and increases serum corticosterone and the delivery of proinflammatory factors[53]. The participation of microglia leads to the production of excitotoxic substances, including excess glutamate, and lowers the release of 5-hydroxytryptamine (serotonin) and BDNF[54].

An excess of stress hormones or their dysfunction, mainly corticosteroids, appears to be related to an action that is opposite to their main function, from a protective antiinflammatory role to a dangerous inflammatory process. New-generation antidepressants and ketamine share similar actions on the stress response. Chronic corticosterone administration downregulates glucocorticoid receptors and reduces dendritic branching in the hippocampus, actions that are reversed by fluoxetine[55]. Chronic corticosterone stimulates noradrenergic activity through α₂-adrenergic receptors in the mPFC, and fluoxetine, through actions on presynaptic α₂-adrenergic receptors, reduces these actions[56]. Long-term corticosterone decreases dendritic spine density in the hippocampal CA1 region, which is also reversed by fluoxetine[57]. Meanwhile, ketamine decreases[58] or is able to antagonize glucocorticoid receptor activity[59], at least in the hippocampus, restoring neural plasticity. Ketamine also blocks serum concentrations of corticosterone after exposure to a stressful situation[60].

ANXIETY

There are experimental approaches to uncover the participation of neuroinflammation processes in anxiety. Emotional distress is associated with increases in cortisol and α-amylase in the acute phase, and long-term salivary examination indicates increases in interleukin (IL)-6 and tumor necrosis factor-α (TNF-α) cytokines and C-reactive protein, indicating chronic inflammation[61]. In patients who suffer from psychotic disorders, depressive disorder, or generalized anxiety disorder, there is a high content of plasma soluble urokinase plasminogen activator receptors and IL-6[62].

In a sample of psychiatric patients, abnormally high cerebrospinal fluid IL-6 Levels correlated with high trait anxiety scores[63]. IL-10 is elevated in patients who suffer from generalized anxiety disorder[64], and levels of this cytokine may predict anticipatory anhedonia and consummatory anhedonia[65]. In generalized anxiety disorder, high plasma levels of IL17A and IL23A are associated with the pathophysiology[66].

Under normal conditions, γ-aminobutyric acid-A (GABA_A) receptors modulate amygdala activity, but low levels of GABA activity are present during anxiety[67]. The metabotropic glutamate family of G protein-coupled receptors in the mPFC, basolateral amygdala, nucleus accumbens, and ventral hippocampus regulate anhedonia, anxiety, and fear[68]. Additionally, in rodents, exposure to prolonged periods of stress causes dendritic atrophy in the hippocampus and increases arborization in the amygdala[69] and mPFC[70]. Likewise, rodents that spend a long time immobile in the forced swim test express anxiety-like behavior, an increase in peroxidase activity, a decrease in catalase activity in the hippocampus, and an increase in plasma levels of IL-17 and interferon-γ (IFN-γ) inflammatory cytokines[71].

In young rodents that were exposed to severe stress conditions and later tested in adulthood, the abnormal function of fear circuits was detected, and anhedonia-like behavior occurred in parallel to high corticotropin-releasing factor-expressing neurons in the central nucleus of the amygdala[72]. The infiltration of monocytes that express IL-1β that derive from monocytes into the brain produces anxiety, and if this proinflammatory cytokine reaches the nucleus accumbens, depression-like behavior occurs[73]. Additionally, IL-17 receptors in the amygdala participate in the anxiogenic process[74], together with the extended amygdala (i.e., central nucleus of the amygdala and bed nucleus of the stria terminalis) cortical-subcortical circuit[75]. In children and adolescents who suffer from anxiety disorder, imaging studies indicate a larger volume of the amygdala[76].

Stress that is produced by long-term sleep deprivation produces anhedonia- and anxiety-like behavior, in parallel with greater serum corticosterone secretion, an upregulation of clock genes, and an increase in plasma proinflammatory cytokine levels[77]. Acute restraint stress produces anxiety-like behavior and high IL-19 and IL-20Rα expression in mPFC pyramidal neurons[78]. When an animal copes with a stressful situation, mPFC connections to the periaqueductal gray participate in anxiety regulation[79].

New-generation antidepressants that are effective in the treatment of generalized anxiety and depression also reduce anxiety scores, improve clinical aspects, and reduce proinflammatory cytokine levels[80,81]. Ketamine administration in a single dose was shown to also exert antiinflammatory effects[82] and produce anxiolytic actions together with a decrease in depression-like behavior, but these anxiolytic actions of ketamine and fluoxetine did not occur after a cholinergic lesion in the diagonal band of Broca[83].

DEPRESSION

Human imaging studies have shown a decrease in the functional connectivity of prefrontal-limbic circuits, which may explain some symptoms of depression[84], and increases in plasma C-reactive protein levels coincided with lower functional connectivity in the ventral striatum, the parahippocampal gyrus, the amygdala, orbitofrontal and insular cortices, and posterior cingulate cortex circuits, with a main effect on the PFC[85]. Postmortem analyses of depressed patients revealed an abnormal reduction of gray matter at the expense of a lower presence of glial cells in the subgenual PFC and its connections. Additionally, depressed patients exhibit alterations of the PFC, the amygdala, and related parts of the striatum and thalamus[86].

Positron emission tomography allows the observation of changes in glucose metabolism. In non-medicated depressed patients, this metabolism increased in the left and right lateral orbital and ventrolateral portions of the PFC, left amygdala, and posterior cingulate cortex and decreased in the subgenual anterior cingulate cortex and dorsomedial-anterolateral PFC. The antidepressant drugs treatment decreased metabolism in the left amygdala and left subgenual anterior cingulate cortex[87].

Depression in aged humans is related to a reduction of connectivity between the amygdala and dorsal PFC regions[88]. In younger depressed patients, a reduction of gray matter volume in the PFC, hippocampus, and striatum was also observed[89], as well as functional alterations of right hemisphere connections between the hippocampus and central nucleus of the amygdala[90]. Likewise, in major depressive disorder, abnormalities occur in circuits that involve the amygdala, medial thalamus, orbital frontal cortex, and PFC, as well as limbic-cortical-striatal-pallidal-thalamic structures, the hippocampus, and ventromedial parts of the basal ganglia[91,92]. Postmortem studies in depressed patients showed lower levels of BDNF and its tyrosine kinase receptor, tropomyosin-related kinase B, in the PFC and hippocampus[93-95], indicating the participation of forebrain structures in depression.

These observations have been replicated in animal models. Prelimbic and infralimbic mPFC regions are connected to various brain regions, and the infralimbic region and its midbrain connections appear to be more related to antidepressant actions than the prelimbic region[96]. Retrograde labeling after chronic restraint stress in rodents revealed a decrease in the distribution density of basal, proximal, and distal dendrites, as well as an increase in the loss of dendritic spines of distal dendrites in infralimbic regions of the mPFC, hippocampus, septum/basal forebrain, hypothalamus, and thalamus and an increase in the distribution density of fluorescence-positive neurons in the amygdala[97].

Among other actions, antidepressants produce their effects based on the balance between excitatory glutamatergic pyramidal neurons and GABA inhibitory interneurons [i.e., the functional excitation: Inhibition (E/I) model][98]. For example, dexamethasone administration in the long term produces depression-like behavior in animal models and a reduction of GABAergic interneurons in the mPFC, leading to an imbalance in the E/I process[99].

An initial action on glutamate activity is necessary for a later action of GABA[100]. Chronic stress-induced impairments in coping strategies reduced the E/I ratio, and an NMDA receptor antagonist restored E/I balance and behavior[101]. Long-term fluoxetine treatment reduced the expression of parvalbumin-GABA-positive neurons and restored the E/I process in cases in which inhibition predominated[102] and increased the interneuron density of dendritic spines to contribute to the restoration of E/I balance[103].

The difference in the onset of action of classic antidepressants and NMDA receptor antagonists could hypothetically be explained by considering that classic antidepressants establish initial changes in neurotransmitter systems that involve GABAergic receptors, among others, but a long latency may influence the action on cerebral plasticity. In turn, NMDA receptor antagonists exert a direct blocking action on NMDA receptors that can explain their anesthetic properties but not antidepressant actions. Instead, after NMDA receptor blockade, a secondary effect occurs, involving a cascade of events in which the activation of neurotrophic factors restores the natural functions of microglia and astrocytes that reestablish neuronal function through a plastic process. In this case, clinical effects appear within minutes or hours. The persistence of these antiinflammatory factors may explain long-term clinical effects.

NEUROINFLAMMATION

Neuroimaging studies show that neuroinflammation is associated with a decrease in activation of the ventral striatum and ventromedial PFC, leading to anhedonia[104]. Chronic stress produces neuroinflammatory responses through the activation of microglia and astrocytes, and then proinflammatory cytokines are released to aggravate symptomatology[105]. Additionally, proinflammatory cytokines may cross the blood-brain barrier and activate microglia[106].

The chronic unpredictable mild stress model produces depression-like behavior and glial cell activation but also inhibits IL-1β, IFN-γ, and TNF-α expression in the hippocampal dentate gyrus. All of these changes are reversed by fluoxetine, which produces reductions of neuronal apoptosis and downregulates apoptotic protein Bax, cleaved caspase 3, and cleaved caspase 9 levels[107]. Under normal conditions, microglia regulate synaptic plasticity and synaptic function. The dysregulation of cytokine activity and presence of neurotoxic elements may be precursors of depression through disruptions of microglial function[108], thereby affecting neuronal plasticity.

The inhibition of astrocyte activation may be a fundamental action of antidepressants[109]. Fluoxetine increases levels of BDNF as an initial effect[110-112] and reduces glycogen levels and increases glucose utilization and lactate release by astrocytes, without any relationship with serotonergic processes[113]. Fluoxetine is also able to promote autophagosome formation and increase the clearance of injured mitochondria in the hippocampus, consequently protecting astrocytes in rodents that are subjected to chronic mild stress[114] and attenuating astrocytic activity[115].

With regard to fast-acting antidepressant drugs, ketamine increases BDNF immunostaining in the PFC and hippocampal regions and increases glial fibrillary acidic protein in the PFC and striatum[116]. Ketamine, when showing a good clinical response, produces significant increases in BDNF levels compared with baseline levels or non-responders to treatment[117], and S-ketamine produces striking actions on astrocyte function, including stopping excessive pruning, proliferation, and activity[118-120].

In neurodegenerative processes, such as Alzheimer’s disease, targeting astrocyte function has been proposed, based on stem cell therapy and gene editing technology, with a focus on astrocyte transformation[121]. Additional models have been proposed, such as chemogenetic tools that are based on Designer Receptors Exclusively Activated by Designer Drugs in astrocytes and microglia[122]. There is a need to develop novel antidepressant drugs that act on perisynaptic astrocytes[123].

CONCLUSION

The discovery of areas in the brain that are susceptible to ICSS, dubbed the pleasure zone concept, sparked great enthusiasm for explaining anhedonia, a common symptom in various psychopathological processes, from stress to anxiety and depression and even other psychiatric conditions. However, these centers that are susceptible to ICSS could be better conceived as centers that regulate adaptive processes that are aimed at survival. ICSS centers, sites of antidepressant actions, and nuclei that are related to the pathophysiology of anhedonia have commonalities. Hedonia is necessary for the survival of species because it involves basic behaviors, such as eating and reproductive behavior. Mapping these structures that are related to hedonia and anhedonia has not yielded conclusive results because their neural activity responds differently depending on prior activity and experience (i.e., hotspots). Monoaminergic and catecholaminergic theories have not yielded expected results with regard to satisfactorily explaining either the mechanism of action of antidepressants or the pathophysiology of depression. In quite common cases, there is a continuum from the stressful situation to adaptive anxiety (allostatic load), which involves the participation of neuroinflammatory factors whose activity disappears when the stimulus that causes stress and adaptive anxiety is eliminated (resilient individuals). However, in other cases, even spontaneously, the neuroinflammatory process can be prolonged, and pathological anxiety and depression subsequently occur, suggesting a failure of neuroplasticity (vulnerable individuals). The participation of microglia and astrocytes is fundamental in these processes, which show a dysregulation of activity.

Notably, classic tricyclic and newer antidepressants and NMDA antagonists share antiinflammatory actions through processes that are beginning to be identified. A promising therapeutic approach consists of finding drugs with specific actions on neuroinflammation processes, particularly with actions on neurotrophic factors, microglia, astrocytes, and proinflammatory cytokines.

ACKNOWLEDGEMENTS

The authors thank Michael Arends for proofreading the manuscript.