Depression has been likened to a state of “accelerated aging,” affecting the hippocampus and the cardiovascular (CV), cerebrovascular, neuroendocrine, metabolic, and immune systems,1–3 and depressed individuals have a higher incidence of various diseases often associated with aging, such as Type II diabetes, metabolic syndrome, osteoporosis, CV disease, stroke, and pathological cognitive aging, including Alzheimer's disease and other dementias.2, 4–8 Depression is also associated with significantly worse outcomes in a number of medical conditions, and depression is an independent risk factor for early mortality (even after accounting for sociodemographic factors, suicide, and biological and behavioral risk factors, such as smoking, alcohol, and physical illness).9–13 Various explanations for “accelerated aging” in depression have been proposed, such as the “glucocorticoid cascade” hypothesis14, 15 and “allostatic load.”16 In this review, we explore the additional possibility that “accelerated aging” in depression occurs at the level of the individual cell and that it can be traced to specific biochemical mediators that are altered in depression. Discovering pathological processes in depression at the cellular level could help identify novel targets for treating depression and its comorbid medical conditions.
We propose a depression model that accounts for certain linked pathogenic processes, which occur in the brain and in the periphery, and which can culminate in cellular aging and damage and disease (Fig. 1). There is widespread recognition that certain physical stressors, such as oxidative and inflammatory stress, can accelerate aging in cells.17–21 It has recently been appreciated that psychological stress can also prematurely age cells, possibly by invoking similar physical processes.15, 20, 22–32 Major depression and its associated biological perturbations are the focus of this review. To the extent similar processes are seen in other conditions (e.g., chronic psychological stress, posttraumatic stress disorder, schizophrenia, certain neurodegenerative disorders, etc.), aspects of this model might also be applicable. Indeed, some of the data supporting this model were derived from chronically stressed, but not necessarily depressed, populations; such data will be identified in the text. In brief, stress-related dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis, as moderated by genetic33–36 and epigenetic37 factors and by cognitive appraisal,38, 39 social support,40, 41 and coping styles,42, 43 leads to cortisol-induced changes in gene expression (including genes related to monoaminergic and peptidergic neurotransmission), neuroendangering or neurotoxic effects in certain brain areas (e.g., prefrontal cortex and hippocampus), excitotoxicity, oxidative stress, immune alterations leading to a proinflammatory milieu (or “neuroinflammation”), and accelerated cell aging (via effects on the telomere/telomerase maintenance system), as described below. In this context, normal compensatory or reparative processes are diminished, e.g., decreased counterregulatory neurosteroids (e.g., dehydroepiandrosterone [DHEA]) and allopregnanolone), decreased antioxidant compounds (e.g., Vitamin C or E), diminished anti-inflammatory/immunomodulatory cytokines (e.g., IL-10), decreased activity of neurotrophic factors (e.g., brain-derived neurotrophic factor [BDNF]) and decreased activity or effectiveness of the telomere-lengthening enzyme, telomerase. The juxtaposition of enhanced toxic processes with diminished protective or restorative ones can culminate in cellular damage and physical disease44 (Table 1). The presentation of this model here will be relatively concise, but related reviews of this and similar models are published elsewhere.20, 22, 23, 36, 45–53 This review represents an update and refinement of models we have presented earlier.45–48 This model is not meant to be complete or all-encompassing, nor is it meant to apply to all individuals with major depression, because different endophenotypes of depression could well have different underlying pathologic components.45, 54, 55 Instead, this broadly sketched model is meant to highlight and connect certain interesting new findings in the study of stress and depression, and to provide testable hypotheses that could guide research and treatment in new directions.
Table 1. Possible detrimental changes seen in depression and/or chronic stress
↑Potentially damaging mediators
↓Potentially protective or restorative mediators
aEvidence is mixed as to whether major depression is characterized by excessive or diminished levels of DHEA, but it is often low with psychological stress.
bEvidence is mixed as to whether the anti-inflammatory/immunoregulatory cytokine, IL-10, is elevated or diminished in major depression.
cTelomerase activity has been reported as low or high (albeit less effective in preserving telomere length) in chronic stress; there are as yet no published data on telomerase activity in major depression.
Hypercortisolemia (with hyper- or hypocortisolism)
The physiological significance of increased circulating GC levels remains unknown, and it is even debatable whether “hypercortisolemia” results in net hypercortisol-ism at the cellular level, or rather in net hypocortisolism, perhaps due to downregulation of the glucocorticoid receptor (GR) (referred to as “GC resistance”).45, 56 It is possible but not proven that hyper- and hypocortisolism identify different subtypes of depression or map onto different symptom clusters.45, 54, 57–59 It should also be recognized that the effects of either state are likely to differ, depending on the target tissue involved and that “relative” conditions of either hyper- or hypocortisolism may exist at the same time within organisms, making any global statements a simplification of the underlying endocrine state.60–63 For example, different GR polymorphisms can significantly affect individuals' responses to GCs,35, 64 and alternative splicing of the GR mRNA can lead to different GR isoforms with different actions in different tissues.65, 66 Furthermore, early life events, such as childhood abuse, can epigenetically reprogram GR expression and splicing, leading to important inter-individual differences in GC responsivity.67 The “hypocortisolism” hypothesis is supported by findings that proinflammatory cytokine levels (e.g., tumor necrosis factor [TNF]-α, IL-1β, and IL-6) tend to be increased in the plasma of depressed patients, and that proinflammatory cytokines can contribute to depressive symptomatology. Because cortisol typically has anti-inflammatory actions and suppresses proinflammatory cytokines (although there are instances to the contrary68–71), the coexistence of elevated cortisol and proinflammatory cytokine levels suggests an insensitivity to cortisol at the level of the lymphocyte GR.72 This possibility is supported by the finding that peripheral GR sensitivity in depressed individuals (assessed by cutaneous vasoconstrictive responses to topically applied GCs) is inversely correlated with TNF-α concentrations.73 The “hypocortisolism” hypothesis is also supported by recent genome-wide expression microarray analyses on monocytes from stressed (but not necessarily depressed) caregivers compared to controls.74 Despite having similar cortisol secretory patterns, the caregivers in that study showed diminished expression of glucocorticoid response element transcripts and heightened expression of transcripts with response elements for NF-kappaB, a key proinflammatory transcription factor.
On the other hand, the “hypercortisolism” hypothesis is supported by phenotypic somatic features suggestive of cortisol excess and of increased end-organ cortisol signaling in depression, e.g., osteoporosis, insulin resistance, Type II diabetes, a relative hypokalemic alkalosis accompanied by neutrophilia and lymphocytosis, hypertension, metabolic syndrome and visceral/intra-abdominal adiposity (reviewed in Ref.45:). Further support of net GC over-activation is provided by evidence of altered expression of target genes such as BDNF, which are believed to be under negative regulatory control by cortisol.75 It remains debatable whether hypercortisolism is causally related to hippocampal atrophy often reported in depression.32, 76–82
Pathologically elevated or diminished GC activity could, via genomic mechanisms, alter transcription of genes involved in synthesis and degradation of monoamine neurotransmitters and other substances,83–87 and could have neurobehavioral sequellae.45 Chronic hypercortisolemia, in particular, has been proposed by Sapolsky and others,14 to result in a biochemical “cascade,” which can culminate in cell endangerment or cell death in certain hippocampal cells. In the simplest description of this model, GC excess engenders a state of intracellular glucoprivation (insufficient intracellular glucose energy stores) in certain cells, impairing the ability of glia and other cells to clear synaptic glutamate. The resulting excitotoxicity results in excessive release of calcium into the cytoplasm, which can contribute to oxidative damage, proteolysis, and cytoskeletal damage.88–90 Unchecked, these processes can culminate in diminished cell viability or cell death. For example, GCs can, via non-genomic mechanisms, directly modulate mitochondrial calcium and oxidation in an inverted U-shaped manner, with chronically elevated levels leading to cellular damage.91 In the present model, we expand upon these earlier GC models by integrating effects on neurotrophic factors, neurosteroids, inflammation, and the telomere/telomerase maintenance system, an important aspect of cell aging.
Although circulating cortisol concentrations are frequently elevated in depression, plasma and CSF concentrations of the GABA-A receptor agonist neurosteroid, allopregnanolone, are decreased in unmedicated depressives, plasma and CSF levels of allopregnanolone increase with selective serotonin reuptake inhibitor (SSRI) treatment in proportion to their antidepressant effect.92, 93 SSRI antidepressants rapidly increase allopregnanolone synthesis, and this may contribute to their anxiolytic effects.92, 94, 95 Another neurosteroid, DHEA, which can have “anticortisol” effects (reviewed in96), and which promotes psychological resilience,51, 97 has been reported to be both high and low in depression,96 but DHEA treatment is generally reported as having significant antidepressant effects.96 Notably, both these neurosteroids modulate HPA,96 BDNF96, 98, 99 and immune3, 100–102 activity, antagonize oxidative stress103, 104 and have neuroprotective or neuroregenerative effects.96, 98–101 Allopregnanolone also inhbits stress-induced corticotropin-releasing hormone release.99 Endogenous decreases in these neurosteroids or exogenously produced increases in their effects would be expected to have damaging or beneficial effects, respectively, in the context of depression or chronic stress.48, 92, 95, 96, 105, 106
Stress-related dysregulation of HPA axis and of GC activity also contributes to immune dysregulation in depression,107 and proinflammatory cytokines further alter HPA axis activity.108, 109 Immune dysregulation may be an important pathway by which depression and chronic stress heighten the risk of serious medical comorbidity.20, 30, 102, 110, 111 Several major proinflammatory cytokines, such as IL-1β, IL-2, IL-6, and TNF-α, are elevated in depression, either basally or in response to mitogen stimulation or acute stress.107, 112, 113 Conversely, certain anti-inflammatory or immunomodulatory cytokines, such as IL-1 receptor antagonist and IL-10, may be increased or decreased or may be dysregulated relative to proinflammatory cytokines.112, 114, 115 In particular, the ratio of proinflammatory to anti-inflammatory/immunomodulatory cytokines may be heightened in depression and could result in increased inflammation112 and, subsequently, in increased free radical production and oxidative stress.116 Converging findings suggest that high peripheral levels of inflammatory cytokines, such as IL-6, are associated with the activation of central inflammatory mechanisms that, under some circumstances, adversely affect the hippocampus, where IL-6 receptors are abundantly expressed.117 Hippocampal neurogenesis is also suppressed by microglial activation, which leads to brain inflammation,118 and high proinflammatory cytokine concentrations can contribute to hippocampal neurodegeneration.119 In wild-type mice, stress increases hippocampal IL-6 concentrations, but IL-6 (−/−) knockout mice are resistant to stress-induced learned helplessness, an animal model of depression.120 In healthy humans, plasma IL-6 concentrations are inversely correlated with hippocampal gray matter,121 and elevated pretreatment inflammatory cytokine levels predict poorer response to antidepressant medications in individuals with major depression.122 High proinflammatory cytokine levels also directly contribute to monoamine dysregulation, HPA axis stimulation, depression, and cellular and organismic senescence.119, 123 It should be noted, however, that due to the complexity of cytokine actions in neurons and glia, the end effect of individual cytokines can be either detrimental or protective, depending on the circumstances.112
Stress and altered HPA axis activity can also increase oxidative damage and decrease antioxidant defenses.20, 29, 46, 124 Oxidative stress, together with inflammatory cytokines, often increase with aging and in various disease states, whereas antioxidant and anti-inflammatory activities paradoxically decrease, resulting in a heightened likelihood of cellular damage and of a senescent phenotype.20, 125 Oxidative stress occurs when the production of oxygen-free radicals exceeds the capacity of the body's antioxidants to neutralize them. Oxidative stress damages DNA, protein, lipids, and other macromolecules in many tissues, with telomeres (discussed below)126 and the brain90 being particularly sensitive. Elevated plasma and/or urine oxidative stress markers (e.g., increased F2-isoprostanes and 8-hydroxydeoxyguanosine [8-OHdG] along with decreased antioxidant compounds, such as Vitamins C and E) have been reported in individuals with depression and in those with chronic psychological stress,27, 29, 127, 128 and the concentration of peripheral oxidative stress markers is positively correlated with the severity and chronicity of depression,29, 129, 130 as well as with evidence of accelerated apoptosis in polymorphonuclear blood cells.131 Furthermore, the ratio of serum oxidized lipids (F2-isoprostanes) to antioxidants (Vitamin E) is directly related to psychological stress, and is inversely related to telomere length and telomerase activity (both discussed below) in chronically stressed caregivers.22 Conversely, antidepressants decrease oxidative stress.132 Because cellular oxidative damage is an important component of the aging process, prolonged or repeated exposure to oxidative stress could accelerate aspects of biological aging and promote aging-related comorbid diseases in depression.29 For example, oxidative stress potentiates TNF-α-induced activation of the cell death cascade.133 Stress- or depression-related increases in oxidative stress additionally blunt certain protective or reparative processes, because oxidative stress is inversely correlated with telomerase activity as well as telomere length (discussed below),126, 134 and because increased oxidative stress (and lower antioxidant protection) is associated with lower BDNF activity135, 136 (discussed below).
BRAIN-DERIVED NEUROTROPHIC FACTOR
The “neurotrophic model” of depression75 posits that diminished hippocampal BDNF activity, caused by stress or excessive GCs, impairs the ability of stem cells in the subgranular zone of the dentate gyrus (as well as cells in the subventricular zone, projecting to the prefrontal cortex) to proliferate into mature cells that remain viable. It is not known whether such processes can cause depression and whether they are relevant to the mechanism of action of antidepressant drugs; evidence is somewhat stronger for BDNF involvement in antidepressant effects than in the etiology of depression.137, 138 Furthermore, unmedicated patients with depression have decreased hippocampal (at autopsy) and serum concentrations of BDNF.137, 139, 140 A role of BDNF in antidepressant mechanisms of action is supported by findings that hippocampal neurogenesis (in animals) and serum BDNF concentrations (in depressed humans) increase with antidepressant treatment,137, 140 and that hippocampal neurogenesis is required for behavioral effects of antidepressants in animals.141 Apart from its direct neurotrophic actions, BDNF also has anti-inflammatory and antioxidant effects and improves the efficiency of brain mitochondrial oxygen utilization, which may contribute to its neuroprotective efficacy,142, 143 BDNF attenuates glucocorticoid-induced neuronal death,144 and BDNF activity synergizes with telomerase activity (discussed below) in promoting the growth of developing neurons.145
CELL AGING: TELOMERES AND TELOMERASE
Telomeres are DNA-protein complexes that cap the ends of linear DNA strands, protecting DNA from damage.146 When telomeres reach a critically short length, as happens when cells undergo repeated mitotic divisions without adequate telomerase activity (e.g., immune cells and stem cells, including neurogenic stem cells in the hippocampus), cells become susceptible to apoptosis and death. Even in nondividing cells, such as mature neurons, telomeres can become shortened by oxidative stress, which preferentially damages telomeres to a greater extent than nontelomeric DNA.126, 147 This non-mitotic type of telomere shortening also increases susceptibly to apoptosis and cell death. Telomere length is a indicator of “biological age” (as opposed to just chronological age) and represents a cumulative log of the number of cell divisions and a cumulative record of exposure to genotoxic and cytotoxic processes, such as oxidation.20, 22, 23, 126, 146, 148 Telomere length may also represent a biomarker for assessing an individual's cumulative exposure to, or ability to cope with, stressful conditions. For example, preliminary data point to accelerated leukocyte telomere shortening, a sign of cellular aging, in chronically stressed22, 23 and in depressed149 individuals. The telomere shortening may, at least in part, be related to increases in stress-related cortisol and catecholamine output.23, 150 The estimated magnitude of the acceleration of biological aging is not trivial; it was estimated as approximately 9–17 additional years of chronological aging in the stressed caregivers and as much as 10 years in the depressed individuals. It should be noted that the subjects in the depression study had very chronic courses of depression (an average of nearly 26 years of lifetime depression).149 Preliminary data suggest that telomere shortening is a function of the duration of the lifetime exposure to depression (Wolkowitz et al., unpublished) and may not be present in individuals with short lifetime exposures to depression. In nondepressed populations, shortening of leukocyte telomeres is associated with atherosclerosis and CV disease,151–153 osteoporosis154 and cognitive impairment,155 and with increased medical morbidity and earlier mortality from a number of causes, including CV and infectious disease, and dementia.156 For example, shortened telomeres are associated with a greater than three-fold increase in the risk of myocardial infarction and stroke, and with a greater than eight-fold increase in the risk of death from infectious disease.157 In a more recent study, baseline telomere length (in women) and prospective rate of change in telomere length over a 2.5 year period (in men) predicted CV mortality over a 12-year period.156 Thus, cell aging (manifest as shortened telomeres), associated with any of the mediators discussed above, provides a conceptual link between depression and its associated medical comorbidities and shortened life span.20, 102, 148
Telomerase is a reverse transcriptase enzyme that rebuilds telomere length, thereby delaying cell senescence, apoptosis, and cell death.146 Telomerase also has antiaging or cell survival-promoting effects independent of its effects on telomere length by regulating transcription of growth factors, synergizing with the neurotrophic effects of BDNF, having antioxidant effects and intrinsic antiapoptotic effects, protecting cells from necrosis, and stimulating cell growth in adverse conditions.145, 158, 159 Telomerase activity has not yet been characterized in individuals with major depression, but it has been reported to be diminished22 or increased160 in stressed caregivers compared to low stress controls. Several of the mediators discussed above can contribute to diminished telomere length and/or telomerase activity (e.g., cortisol,150 oxidative stress,147 and inflammatory cytokines160, 161), highlighting the interlinked nature of cell-damaging and cell-protective mediators in stress and depression. Important moderators of telomere length are rapidly being discovered (e.g., childhood maltreatment,162 socioeconomic status,163 race,164, 165 physical exercise,166 and dispositional pessimism,39 among others).
DO STRESS AND DEPRESSION ACCELERATE CELL AGING?
We have briefly reviewed evidence of biochemical abnormalities in depression, some of which are consistent with an aged phenotype that could contribute to certain medical comorbidities seen with depression. They could also contribute to the depressive state itself, but that has not been adequately tested. In particular, depression (and perhaps chronic stress, as well) may be associated with increased cell damaging processes and decreased cell protective or restorative ones (Table 1). We propose a model in which these abnormalities are causally interlinked and may derive, directly or indirectly, from altered HPA axis and GC activity seen in depression (Fig. 1). It remains uncertain whether the brain in depressed individuals is subject to net hypo- or hypercortisolism, and even within the brain, individual component tissues, such as neurons and glia, may differ in their response to altered circulating GC levels as a result of differing receptor expression or metabolic enzymes.
We have couched this model in terms of “accelerated aging” at the cell level, although whether cell aging is actually accelerated in depression remains to be determined in prospective trials. It is important to recognize that this model is unlikely to apply to all individuals with depression (many of whom do not have discernible HPA axis dysregulation), and that many of these changes are not specific to major depression. Also, various genetic and epigenetic moderators, not discussed here, are undoubtedly important.39, 52, 67, 167–169 The major importance of this hypothetical model is that it identifies certain nontraditional targets for pharmacological and nonpharmacological treatment, and thus could lead to new theory-driven therapies. In particular, treatments directed at the targets identified here have the potential not only to treat depression but also to treat certain medical comorbidities that occur alongside depression.47, 49, 170 Interestingly, even traditional antidepressant medications, which putatively work via monoaminergic actions, affect many of the novel targets described here95, 109, 128, 171–177 (see Fig. 1), even though they were not developed with those purposes in mind. Last, the identification of novel biomarkers of depression may discriminate separate endophenotypes of depression that respond differently to different treatments,54, 55, 122, 174 although some of the endocrinological and neurochemical differences reported may be dependent more on the target tissue examined than reflective of a global endophenotype. This will hopefully accelerate the era of personalized antidepressant treatment.
THEORETICAL MODEL AND NOVEL TREATMENT POSSIBILITIES
A schematic overview of our model is presented in Figure 1. The condensed and simplified nature of this schematic precludes depiction of numerous other mediators and moderators and interactions that are involved. Therefore, this depiction should be viewed as a “broad brush stroke” theoretical model. The bracketed numbers in Figure 1 are keyed to potential sites of therapeutic intervention described below. In this model, elevated cortisol levels are associated with downregulation of GRs (“GC resistance”); the “net” GC activity remains uncertain and could even differ in different tissues. A deficit in GR function can precede or result from the hypercortisolemia. To the extent that lymphocyte GRs become GC resistant, immune function is altered and excessive proinflammatory cytokine effects can occur. Changes in cortisol activity also result in multiple genomic changes, e.g., altered levels of certain neurotransmitters (e.g., decreased serotonin and increased dopamine activity in certain brain regions, which could contribute to depressive or psychotic symptoms). To the extent GC activity is “excessive” in certain brain regions, a cascade of events can follow, characterized by diminished insulin signaling, intraneuronal glucoprivation and diminished energy availability, defective clearance of intrasynaptic glutamate, excitotoxicity, intracellular buildup of calcium, generation of oxygen-free radicals (oxidative stress), diminution of telomerase activity and cellular damage or cell death. Increased oxidative stress can damage the enzyme telomerase and shorten telomeres, at least in certain cells in the body. In nondepressed individuals, leukocyte telomere shortening is associated with a host of physical illnesses and premature mortality. If this occurs in depressed individuals as well, it could help explain the surfeit of medical illness and the shortened life expectancy seen with chronic depression. Chronic stress and depression and/or excessive cortisol exposure can also be associated with underproduction of certain counterregulatory neurosteroid hormones, e.g., DHEA and allopregnanolone, which could further dysregulate HPA axis activity, hamper antioxidative function, and reduce neuroprotective capacity. Additionally, prolonged stress and/or increased cortisol activity can downregulate BDNF activity, which further diminishes neuroreparative capacity and attenuates neurogenesis.
To the extent this theoretical model is accurate, several potential treatment loci emerge, as indicated numerically in Figure 1: (1) traditional antidepressants have several novel functions apart from increasing intrasynaptic monoamine concentrations: they up-regulate GR function,172 increase allopregnanolone synthesis (certain SSRIs),95 increase BDNF levels,75 and have anti-inflammatory174, 178 and antioxidant179, 180 effects; (2) CRH antagonists;181 (3) stress reduction, meditation, and other behavioral and lifestyle interventions;20, 182, 183 (4) antiglucocorticoids47, 184–186; (5) energy supplementation or insulin receptor sensitizers;187–189 (6) glutamate antagonists;190–194 (7) calcium blockers195, 196 and antioxidants197; (8) DHEA;96 (9) 3-α-hydroxy–steroid dehydrogenase (3-α-HSD) stimulators (including SSRIs), which increase allopregnanolone synthesis94, 95; (10) environmental enrichment, exercise198–201; (11) BDNF administration via novel routes of administration202–204; (12) telomerase activation,205, 206 and (13) anti-inflammatory drugs, TNF-α antagonists, etc.109, 174, 207–211 It is possible that, by targeting such “upstream” mediators of the biochemical milieu, additional therapeutic leverage might be gained. Already, preliminary studies are testing many of these strategies, with preliminary signs of success.
The authors acknowledge the generosity of the O'Shaughnessy Foundation, which supplied major funding. Additional funding was supplied by the University of California, San Francisco, Academic Senate. The authors are also grateful to Dr. Jue Lin, who has provided expert advice and technical aid in the field of cell aging and Dr. Elizabeth Blackburn, a pioneer in the field of cell aging, whose guidance has been indispensible.
Financial disclosures: Dr. Wolkowitz has received lecture honoraria from Jazz Pharmaceuticals and Merck Pharmaceuticals, and has served on an Advisory Board for Pfizer Pharmaceuticals. No other authors have financial ties to these or any other pharmaceutical companies.