The HPA axis is the primary neuroendocrine system mediating the stress response and includes the hormones and structures mediating the production of glucocorticoids. Corticotrophin-releasing hormone (CRH), also known as corticotrophin-releasing factor (CRF), is produced in the paraventricular nucleus of the hypothalamus. It acts on CRF1 and CRF2 receptors in the central nervous system and anterior pituitary (34). The CRF1 receptor mediates anxiety and depression behaviors and the stress response. The role of CRF2 is not known, but has been hypothesized to counter the actions of CRF1. Alternatively, it may be that CRF1 is activated by escapable stressors and CRF2 is activated by inescapable stressors. It is a major regulator of basal and stress-induced release of proopiomelanocortin (POMC) and POMC-derived peptides, such as adrenocorticotrophic hormone (ACTH) and beta-endorphin, from the anterior pituitary. ACTH acts on the adrenal cortex to promote synthesis and release of cortisol and other glucocorticoids. Glucocorticoids inhibit subsequent release of CRF and ACTH. Gamma-aminobutyric acid (GABA) inputs from the hippocampus inhibit the stress response by decreasing CRF synthesis in the central nucleus of the amygdala (cnAmy) (35). Serotonin, norepinephrine, and acetylcholine inputs from the amygdala and hippocampus stimulate secretion of ACTH. Serotonin neurons terminate on inhibitory GABA neurons to block GABA inhibition of CRF synthesis (36). Dampened GABAergic tone in rats exposed to maternal separation enhances CRF expression in the amygdala and activation of the NE system (37). Thus, it appears that GABA might play a tonic regulatory role on the HPA axis.
The mechanisms underlying disturbance in the HPA axis include increased secretion of any or all of the hormones in the cascade or decreased sensitivity to negative feedback at any or all levels of the axis (38). CRF antagonists reduce stress-induced increases in plasma catecholamines, tyrosine hydroxylase mRNA in the locus coeruleus (LC), and CRF mRNA and Type 1 CRF receptor mRNA in the paraventricular nucleus (PVN) (39), giving evidence of a tonic regulatory role of CRF in specific brain regions in animal models.
Cortisol is elevated over 24-h periods in severely depressed patients (40) consistent with increased stress as part of the syndrome. Dexamethasone, a synthetic glucocorticoid, suppresses ACTH release in most healthy individuals at a standard dose (41, 42). Depressed patients have a significantly higher rate of nonsuppression than controls, although rates of nonsuppression are still not that high (43). This is one example of considerable overlap between patients with and without depression in a measure that distinguishes some, but not most, patients meeting the broad criteria for the diagnosis of depression.
CRF, which is increased in cerebral spinal fluid (CSF) and plasma in some depressed patients, activates the sympathetic nervous system and inhibits gastric emptying as well as gastric acid secretion. CRF also inhibits the secretion of growth hormone (35). After injection of CRF, the amount of ACTH released is less in depressed patients than in normal subjects (44, 45). This blunted ACTH secretion suggests that there is increased central CRF release (46, 47), since, in animals, stress and adrenalectomy lead to hypersecretion of CRF and downregulation of receptors in the anterior pituitary (48).
HPA Axis, Anxiety, and Stress
Acute stress leads to release of CRF, ACTH, and cortisol (HPA axis activation). With continued stress, adaptive changes occur. Most studies to date have focused on various animal models of stress. These reveal feedback inhibition by glucocor-ticoid receptors in the hippocampus and pituitary, downregulation of postsynaptic norepinephrine receptors as well as upregulation of inhibitory autoreceptors and heteroreceptors on presynaptic NE neurons.
In some types of anxiety, adaptive changes during chronic stress lead to lower levels of corticosterone and ACTH than seen acutely (49). In other types of anxiety, there are enhanced increases in corticosterone (50), and prior stress experience can lead to augmentation of subsequent stress response. The multiple forms of stress and anxiety that can be associated with depression and multiple inter-related possible physiological responses render any simple generalizations inappropriate. For instance, some relatively time-limited stressors lead to long-term HPA axis effects. Severe prenatal stress or early maternal deprivation stress leads rats to have higher corticosteroid concentrations with exaggerated glucocorticoid responses to stress persisting to adulthood (51, 52). A review of how this may account for the great impact of early neglect and abuse as well as its potential role in the etiology of depression is available elsewhere (20).
Limbic-Cortical-Striatal-Pallidal-Thalamic (LCSPT) Tract, Stress, and Depression
The LCSPT tract consists of several extensively interconnected brain structures: hippocampus, amygdala, caudate nucleus, putamen, and frontal cortex. These regions have glucocorticoid receptors (53, 54) and thus may be affected by variations in glucocorticoid concentrations. Most imaging studies, e.g., 3D MRI, show measurable, but relatively small, changes in volumes of LCSPT tract structures between depressed and control subjects; and postmortem brain studies have also noted volume loss. The hippocampus, the most studied of these structures, most consistently shows volume loss. Since these LCSPT brain structures are interconnected, they mutually influence each other; and effects, such as volume loss, in one structure might be expected to be reflected in structural or functional changes in the other structures (33).
Nevertheless, evaluation of volume reduction in the other LCSPT structures has lacked consistency with volume loss observed in some, but not all, studies. The lack of consistent findings in such studies has led to hypotheses related to subsets of patients who have reduction in structure volume rather than the alternative hypothesis that there is a significant overlap of LCSPT tract size between depressed and normal subjects. It should also be noted that compensatory changes, such as the presence of increased neurons in the paraventricular nucleus (PVN) of the hypothalamus (55, 56), may possibly obscure detection of volume loss. It has been noted that there is an apparent association of greater hippocampal atrophy with depression subtypes that are more likely to have hypercortisolemia (57). MRI studies have shown that the magnitude of hippocampal loss is associated with frequency of depressive episodes and the duration of symptoms prior to treatment (58). Another possibility that might have led to the lack of consistency in findings across studies is that the volume loss is small and may not be detectable using the techniques and technologies utilized by all evaluators. It should also be noted that volume loss does not necessarily imply cell loss which, when observed, may involve glia rather than neurons (see below).
The cause of the reported hippocampal volume loss is unknown. Various proposals include the following: (1) Depression susceptibility is associated with stress-related volume loss, precedes the onset of depression, and is central to the development of depression (22, 25). (2) Neuronal loss occurs secondary to exposure to hypercortisolemia (59). (3) Glial cell loss results in increased vulnerability to glutamate neurotoxicity since glia are responsible for most glutamate removal from the synapse and the production of brain-derived neurotrophic factor (BDNF). Thus, glial loss results in increases in synaptic glutamate and decreases in BDNF in the LCSPT tract, both potentially resulting in neuronal loss. (4) Stress results in reduction in neurotrophic factors (60), such as BDNF and glial-derived neurotrophic factor which tonically suppress apoptosis, the latent biochemical (suicide pathway) leading to cell death (61). (5) Stress results in reduced neurogenesis (62, 63). (6) Genetic polymorphisms decreasing activity-dependent release of BDNF, perhaps working synergistically with a polymorphism of the gene encoding the serotonin transporter and stress combine to produce depression (16).
Animal models provide support for the ability of many, but not all antidepres-sants to induce adult hippocampus neurogenesis, and when this effect is blocked, the signs of antidepressant response in rodent models are reversed. On the other hand, several rodent models of stress reduce hippocampal neurogenesis and that alone is insufficient for production of depressive-like signs. Considering these contradictory findings, it has been proposed that antidepressants act through neurogenesis-dependent and neurogenesis-independent processes (64). Furthermore, most evidence suggests that the reduction in adult hippocampal neurogenesis is not responsible for volume reductions in depression, although may be responsible for cognitive deficits observed in clinical depression. There is an increasing focus on the ventral hippocampus, which has connections to the prefrontal cortex and limbic system. The ventral hippocampus has hilar mossy cells and interneurons that are modulated by dopamine, serotonin, and norepinephrine, possibly linking structure and function in depression.
Additional evidence supporting a role for the LCSPT tract in depression is that late-onset depression is more common in age-associated medical and neurological disorders that cause damage to the LCSPT tract (65). Prolonged maturation and stabilization of neural elements and synapses in the prefrontal cortex (PFC) continues into adulthood. This neural plasticity may make the PFC more susceptible to reductions in neuronal density (66).
If the state of depression produces or increases reductions in critical brain structures, then the ability of antidepressants to increase neurotrophic factors such as BDNF may prove therapeutically important for the relief of symptoms. Some but not all antidepressants increase BDNF and neurogenesis, suggesting that this may be one of several therapeutic mechanisms of antidepressants.
Hippocampus: Possible Pivotal Role Among LCSPT Tract Structures?
During stress, normal feedback mechanisms in the HPA axis fail to operate, leading to damage to hippocampal neuronal cells (67). Stress is associated with damage to the hippocampus in animals (60). Sustained fetal social stress in vervet monkeys causes neuronal degeneration of the CA3 region (68). Chronic restraint stress in rats causes atrophy of apical dendrites of CA3 pyramidal neurons which could lead to decreased volume without loss of neurons themselves (69). Cold water immersion stress in rats causes structural damage to the CA2 and CA3 fields and decreases CRF in hippocampus (70, 71). Chronic exposure to corticosterone also leads to loss of CA3 region neurons (59, 72) and decreased dendritic branching and length of hippocampus (73). For example, in Cushing's syndrome, an endocrinopathy manifest by overproduction of cortisol leads to reduced hippocampal volume (74).
In man, those studies reporting hippocampal volume loss show it to persist over years and after depression has resolved. The amount of volume loss appears best related to the total lifetime duration of depression, not the age of the patient (57, 75). Whether or not hypercortisolemia is related to findings of decreased hippocampal volume remains, however, to be demonstrated. The close relationship that might have been predicted from preclinical studies has not, to date, been established.
Nonetheless, other lines of evidence point to linkages between glucocorticoids and hippocampal volume. For instance, hippocampal lesions lead to increased release of glucocorticoids during stress (76, 77), and this release may lead to further damage of the hippocampus (71). Hippocampal atrophy may result in impaired cognition, a feature of depression. Patients with hippocampal atrophy may be more treatment resistant (78); however, because the amount of hippocampal atrophy tends to be related to the duration of depression, hippocampal atrophy may be a surrogate marker for earlier onset and more frequent recurrence. This brings us back to the potential of restorative processes that may prove important in the long-term treatment and management of depression.
Neuronal Plasticity and Brain-Derived Neurotrophic Factor (BDNF)
BDNF is a downstream target of the c-AMP pathway. It regulates neuronal survival and synaptic plasticity both during development and in adult brain (79). Stress is associated with decreased BDNF (80). Serum BDNF concentrations have been reported to be decreased in depression and continue to be explored as a potential biomarker of the depressed state (81). When BDNF is infused into the midbrain, it produces an antidepressant-like effect in two behavioral models of depression, learned helplessness and forced swim tests, suggesting that BDNF may be involved in depression (82).
Consistent with this possibility, the cascade of events which follow antide-pressant treatment can produce increased BDNF according to a series of studies in animal models, although many other studies have failed to replicate these findings. Chronic antidepressant treatment increases Gs coupling to adenyl cyclase which results in increased cyclic adenosine monophosphate (c-AMP) which increases Ca2+-dependent protein kinases and leads to increased expression of the transcriptional regulator c-AMP response element binding protein (CREB) (83, 84) which increases both BDNF expression in limbic structures, including hippocampus, and the BDNF receptor, TrkB (85). Chronic administration of antidepressants and electroconvulsive seizures increases proliferation and survival of new neurons consistent with the effects shown after activation of the cAMP-CREB cascade or incubation with BDNF which increases differentiation of new cells into neurons (86). Taken together, these findings suggest that some treatments of depression enhance neurotrophic factor activity in specific brain regions (22, 24).
How Strong Is the Case for a Major Role of Stress and the HPA Axis in Depression?
As reviewed above, multiple lines of preclinical and clinical evidence argue that depression is associated with functional and/or structural alterations in the brain which are consistent with HPA dysfunction. Furthermore, whatever the primary biochemical effects of antidepressant treatments, pathways exist whereby long-term effects impinge on components of the HPA axis (87). What is not addressed by recent formulations is the failure to translate the finding of hypercortisolemia in depression reported three decades ago (40) into a convincing diagnostic tool and/or predictor of treatment response despite diverse and sustained efforts (41). As more sensitive methods have become available to document region-specific changes in structure or in function in the brains of patients with depression or effects of antidepressants on glucocorticoid receptor function in preclinical models, there has been a new wave of circumstantial evidence to support statements such as "...disturbed regulation of CRF neuronal circuitry plays a causative role in producing cardinal signs and symptoms of depression." (88). The problem for the clinician or neuroscientist focused on providing or developing the best treatments is that no measure or combination of biochemical and physiological measures has allowed for a stable, reasonably replicable, and robust means of distinguishing a depressed from a normal individual, or for predicting an individual patient's response to different classes of antidepressants.
A primary focus on the HPA axis and, more recently, LCSPT tract risks subsuming findings of alterations in other measures as merely secondary. As will be succinctly reviewed in what follows, investigators have reported that other neuroendocrine or neurotransmitter systems are just as consistently dysregulated in depression as the "primary" HPA one. As cataloged in Table 1 and conceptualized in Figs. 1 and 2, these constitute a multitude of complex and potentially inter-related findings relevant to the pathophysiology and treatment of unipolar depression(s). As noted at the outset, trying to fit manic-depressive illness and unipolar depression into a common pathophysiologic model is an even more difficult task, particularly when one considers the differences in spectrum of efficacy between putative mood stabilizers and antidepressants. We will, therefore, continue to restrict our focus and only occasionally refer to those studies on bipolar disorder that help to elucidate investigations of unipolar depression.
Given the complexity of findings, even within the broad category of patients with unipolar depression and the spectrum of marketed antidepressants with highly variable efficacy, it is not surprising that researchers look for unifying hypotheses. Unfortunately, those that have been proposed and tested such as definable norepi-nephrine or serotonergic types of depression have not been supported and those, such as the primacy of HPA axis dysfunction, have not been testable in the absence
Fig. 1 Interaction of neuroendocrine system involved in the depression cascade
of appropriate pharmacologic agents. Reasoning that we should remain open to all lines of evidence, we will highlight reports of many other classes of abnormalities in depression that may or may not ultimately prove to be related to those of the HPA axis. In the absence of a compelling scientific case to narrow one's focus, we may best achieve therapeutic advances by targeting each of the systems implicated in depression and evaluating the potential advantage of selective interventions either alone or in combination (see below).
The Hypothalamus-Pituitary-Thyroid (HPT) Axis, Growth Hormone, Somatostatin, and Prolactin in Depression
It has been noted for many decades that many behavioral symptoms of hypothyroidism -dysphoria, anxiety, fatigue, and irritability - overlap those of depression. This observation plus the clinical finding that small doses of thyroid may potentiate the effects of antidepressants (89) has sustained an interest in the relevance of this system to depression. Thyrotropin-releasing hormone (TRH) released from hypothalamus stimulates TRH receptors in the pituitary to release TSH which stimulates specific receptors in the pituitary to release triiodothyroxine (T3) and thyroxine (T4) hormones. A subset of depressed patients show a blunted TSH response to TRH, others symptomless autoimmune thyroiditis (46), and still others an exaggerated
TSH response to TRH (88). Preclinical studies on the modulation of multiple neurotransmitter functions in the brain coupled with clinical observations on rates of mood switches in bipolar disorder point to the possibility that to understand certain forms of depression, it will be necessary to understand altered function of components of the HPT axis (90).
Growth hormone (GH) and somatostatin, the hypothalamic GH suppressing factor, regulation have also been found to be altered in depression. A change in the diurnal rhythm of GH may be reflected by increased plasma concentrations (91), a finding that is opposite in direction to what would be provided if CRF were exerting control (see below). It is here worth recalling that cortisol abnormalities are also best described in terms of the diurnal pattern with elevations only observed at certain times of the day (92). GH increases to a2 agonists (e.g., clonidine) are blunted in depressed patients (93, 94). This blunted GH response has been consistently replicated and complemented by findings of blunted responses to uptake inhibitors, such as desmethylimipramine, which increase the intrasynaptic concentrations of the endogenous a2 agonist norepinephrine (95).
Interestingly, somatostatin concentrations are reported to be reduced in the CSF of depressed patients compared with controls, although this finding is not specific to depression and may be related to elevated cortisol concentrations (44, 96, 97). A reduction of the inhibitory factor is also consistent with the previously described elevation of GH in blood but not the blunted response to a2 stimulation. The latter is most consistent with several lines of evidence implicating altered a2 function in depression (98). The complex inter-relationships of neuroendocrine and monoam-ine function are not well enough understood to allow us to test for primary causality of any single abnormality.
Another highly replicated neuroendocrine abnormality in depression is that of blunted prolactin responses to serotonergic stimulation. For instance, there is a blunted release of prolactin to a fenfluramine challenge in depressed patients (99, 100). Prolactin responses to intravenous tryptophan, a precursor of serotonin (101), or clomipramine, a serotonin uptake inhibitor (102, 103), are also blunted. Since abnormalities of unstimulated prolactin have not been reported, these responses would appear to best reflect altered serotonin function.
As already noted, the inter-relatedness of catecholamine and serotonin systems in the brain with modulation of neuroendocrine function makes it difficult to address cause vs. effect as reflected in the above examples. An additional issue is that many of the observed abnormalities involve a circadian component, which, in other words, may only show differences at certain times of day, which leads to an interest in a pathophysiologic role of altered circadian regulation (104), particularly in terms of seasonal affective disorder (105). Melatonin secretion varies over the 24 h period in a circadian pattern related to light and darkness. Its secretion is partly under norepinephrine control and exogenous melatonin and/or using light to shift the phase of endogenous melatonin may have a role in the treatment of circadian disorders under which seasonal affective disorder can be subsumed (105). It has also been suggested that blunted circadian variation in natural killer cell activity in depression may reflect some underlying chronobiological rhythm (106).
All these reports of altered neuroendocrine and possible circadian regulation in depression need to be considered in light of the extensive work on the monoamine neurotransmitters in brain which have been shown to be involved in the action of established antidepressant treatments. Despite the theoretical attractiveness of other approaches, no intervention derived from neuroendocrine or circadian hypotheses has yet led to a treatment which, by itself (e.g., light therapy), shows sustained efficacy in a substantial proportion of patients diagnosed with depression. Considerable effort has gone into identifying CRF antagonists which will ultimately allow for a test of whether excess CRF tone plays a pathologic role in patients with evidence of hypercortisolemia. Disappointingly, the most recent large study with a CRF antagonist in depression was negative (107), although it is not known whether the doses employed significantly altered function in the brain or the extent to which CRF1 receptors were blocked.
Was this article helpful?