Among elderly individuals, depressive symptoms are common and burdensome. Approximately 1% of individuals over the age of 65 meet criteria for major depressive disorder (MDD), as defined by the diagnostic and statistical manual of mental disorders, fourth edition, text revision (DSMIV-TR; American Psychiatric Association, 2000), a prevalence lower than that in younger adults (Kessler et al., 2003). Another 15–25%, however, experience depressive symptoms that, while not meeting criteria for MDD, do cause significant distress and interfere with daily functioning (Koenig and Blazer, 1992). In this article, the term late-life depression (LLD) will be used to refer to individuals over the age of 65 who for the first time in their lives meet criteria for MDD or display clinically significant depressive symptoms. Individuals with LLD experience greater functional disability (Dombrovski et al., 2007) and cognitive decline (Lenze et al., 2005) than those without. Further, they are at increased risk of morbidity and mortality from medical illness (Ganguli et al., 2002). LLD also appears to contribute to increased rates of suicide among older individuals (Van Orden and Conwell, 2011).

The biological substrates of LLD are being characterized and several hypotheses for the etiology and pathophysiology of LLD have been proposed, including the vascular hypothesis (Alexopoulos et al., 1997), inflammation hypothesis (Alexopoulos and Morimoto, 2011), and dementia prodrome hypothesis (Byers and Yaffe, 2011; reviewed in McKinney and Sibille, 2012). Here, we propose an alternative and complementary hypothesis, which we termed the age-by-disease biological interaction hypothesis of LLD. Central to this hypothesis is the concept of molecular aging of the human brain. An earlier version of this hypothesis has been described elsewhere (McKinney and Sibille, 2012).

Despite its critical importance to a population that is growing older, “normal” brain aging is understudied. This may be due to the often expressed, but false belief held by many that aging is inescapable, broad-ranging and non-specific. Studies that have investigated biological aging have revealed specific changes and thus avenues for intervention. At the cellular level in the human brain, morphological and stereological studies reveal a decrease in neuron volumes, a small loss or no change in cell numbers (Morrison and Hof, 1997; Pakkenberg and Gundersen, 1997), and a progressive thinning of cortical thickness, affecting both gray and white matter (Resnick et al., 2003; Sowell et al., 2003). Similar structural changes with age have been demonstrated in the brains of animal models (Jucker and Ingram, 1997; Peters, 2002). At the molecular level in animal models, less than 10% of brain-expressed genes exhibit age-related changes in gene expression (Lee et al., 1999, 2000; Jiang et al., 2001; Blalock et al., 2003; Sibille et al., 2007). Similar numbers have been reported in studies of human tissue (Lu et al., 2004; Avramopoulos et al., 2011). In one such study of human tissue from the prefrontal cortices of subjects aged 13–79, our group used gene microarray technology to investigate age-related changes in gene expression and reported that approximately 7.5% of genes changed significantly with aging (Erraji-Benchekroun et al., 2005). Other studies have confirmed these results, identifying a maximum of ~10% of all detected genes, depending on sample size and analytical power of the respective studies (Yankner et al., 2008; Glorioso et al., 2011). Of note, not only is the identity of the genes and gene classes that are affected with aging consistent among studies, but so are the directions of change.

Interestingly, the identity of the genes whose expression changes with age suggests that specific cellular populations and biological processes are selectively affected by the aging process. For instance, expression of genes playing a role in glial-mediated inflammation, oxidative stress responses, mitochondrial function, synaptic function and plasticity, and calcium regulation and neuropeptide signaling have consistently been shown across multiple studies to be affected by aging, while numerous other neuronal and glial genes remain apparently unchanged (Yankner et al., 2008; Glorioso and Sibille, 2011). Overall, age-upregulated genes are mostly of glial origin and related to inflammation and cellular defenses, while downregulated genes display mostly neuron-enriched transcripts relating to cellular communication and signaling (Erraji-Benchekroun et al., 2005).

Using the expression levels of the age-dependent genes and their expected trajectories with age, we have generated predicted ages of individual subjects from which the brain tissue was sampled, and demonstrated that this predicted age is highly correlated with the chronological age of that individual (Erraji-Benchekroun et al., 2005; Glorioso and Sibille, 2011; Glorioso et al., 2011). We have termed this predicted age the “molecular age”. These findings suggest that gene expression changes with age can be used as biomarkers for brain aging.

This correlation between molecular and chronological ages is robust in individuals without neurodegenerative and neuropsychiatric disorders (Erraji-Benchekroun et al., 2005), but investigations of individual genes suggest that the molecular age or individual gene trajectory can deviate from chronological age in individuals with these disorders. To illustrate this phenomenon, one can look at somatostatin (SST), a signaling neuropeptide that is expressed in a subpopulation of gamma-aminobutyric acid (GABA)-positive inhibitory interneurons (Viollet et al., 2008). We have demonstrated that expression of SST decreases progressively with age in individuals without neurodegenerative and neuropsychiatric disorders such that expression levels at 70 years of age are approximately 40–50% of those at 20 years of age (Erraji-Benchekroun et al., 2005; Tripp et al., 2011; Figure 1A). SST is also downregulated in individuals with MDD. Interestingly, the magnitude and direction of change in SST expression with age is similar to that of control individuals, but the absolute values of SST expression are lower at most ages in individuals with MDD compared to those without (Sibille et al., 2011; Tripp et al., 2011; Figure 1A, top panel). Similar findings have been observed in subjects with schizophrenia (Morris et al., 2008; Figure 1A, bottom panel).

The relationship between gene expression of a disorder-associated gene and aging is not limited to SST. In fact, genome-wide investigations have reported that up to a third of age-regulated genes in the human brain have been at some point associated in the literature with neurodegenerative (Alzheimer’s, Parkinson’s, and Huntington’s diseases, amyotrophic lateral sclerosis) and neuropsychiatric disorders (bipolar depression, major depression, and schizophrenia; see Figure 3 in Glorioso et al., 2011 for details). Not only do the genes relevant to these brain disorders show age-related changes, but the direction of the changes that occur with age is almost always in the direction thought to cause or promote diseases. See Table 1 in Glorioso et al. (2011) for a list of approximately 40 top candidate disease genes that exhibit agreement between age- and disease-related changes. Studies of gene expression in Alzheimer’s disease (AD) and normal aging find that most of the changes in gene expression that occur in normal aging also occur in AD, only in greater magnitude (Miller et al., 2008; Avramopoulos et al., 2011). The difference in the magnitude of change in gene expression between individuals with brain disease and those without during aging is attributable to the disease. Our hypothesis posits that it is these changes that drive the disease process; however, it cannot be ruled out that these additional changes in magnitude are the result of the disease process rather than the cause. That said, there are certainly genes for which it is known that the age-related direction of change in gene expression directly causes, rather than is a consequence of, the disease, for example, the Parkinson’s disease genes PINK₁ and DJ₁/PARK₇. The expression of these genes progressively declines with age and it is known that individuals with loss of expression/function mutations in these genes develop Parkinson’s disease (Schapira, 2008). In contrast, relatively fewer of the larger pool of genes that do not display age-dependent changes (<5%), are associated with neurodegenerative and neuropsychiatric diseases.

Recently, we have further investigated the relationship between age, gene changes, and neuropsychiatric disorders, specifically in the context of MDD (Douillard-Guilloux et al., 2012). Results demonstrated that most MDD-related genes were frequently age-regulated in both MDD and in control subjects, and that the effects of MDD and age were positively correlated. Moreover, most genes that are age-dependent in control subjects displayed greater age effects in MDD subjects, and the overall increased prevalence of age effects on genes in MDD subjects corresponded to similar trends in controls, rather than representing de novo age effects. This systematic correlation between age-dependent and depression-related changes, with greater effects in depressed subjects, further suggests that normal brain aging is a risk factor for biological changes observed in MDD subjects.

One interpretation of these observations is that age-dependent changes (i.e., molecular aging) are on an earlier trajectory in individuals who develop MDD and potentially other neuropsychiatric disorders. However, it is important to note that these studies are cross-sectional and do not follow the longitudinal progression of gene changes within individuals, so it is not known whether age-dependent changes are on an earlier trajectory, or whether changes occurred at earlier time points and were fixed at lower levels, for instance in the case of SST. So while we hypothesize that age may be pushing the expression of genes in disorder-causing directions, an alternate scenario is that of earlier and fixed changes, which then act as latent vulnerability factors that are revealed with advancing age, resulting in increased vulnerability to develop neurodegenerative and psychiatric disorders, including LLD.

While molecular and chronological ages are highly correlated, we have also reported individual deviations from predicted ages (Erraji-Benchekroun et al., 2005; Glorioso et al., 2011). The fact that molecular age can deviate from its chronological age suggests that modulating factors exist and may contribute to one’s vulnerability to brain aging and to developing late-life brain disorders, such as LLD. In the age-by-disease biological interaction hypothesis we have proposed that those individuals with older predicted molecular ages compared to their chronological age may not only display greater biological brain aging, but may also be at greater risk of age-gated brain diseases, because gene expression of disease-related genes would have proceeded further in disease-promoting directions. Conversely, subjects with younger age-dependent gene trajectories and lower predicted molecular ages would be at lower risk, and may in fact display resiliency against LLD and other late-life disorders (Figure 1B). Environmental and genetic factors represent obvious candidates to modulate the trajectory of biological aging.

In a proof-of-principle study, our laboratory sought to demonstrate a genetic role in modulating the aging process. The above-described “molecular age” assay was used to characterize the brain tissue of individuals carrying different polymorphisms of the sirtuin genes (Glorioso et al., 2011), a family of genes previously demonstrated to modulate age and longevity in nematodes, insects, and rodents. We focused on SIRT5, due to prior report of altered expression for that gene in a rodent model of anticipated brain aging (Sibille et al., 2007). This study found that subjects carrying a low-expressing polymorphism of the SIRT5 gene had molecular ages that were older than actual chronological age, as measured in the three different areas of the cerebral cortex (i.e., BA9, BA24, and BA47) of human postmortem samples (Glorioso et al., 2011). Interestingly, this effect was not demonstrated in the amygdala, a brain area in which the low-expressing polymorphism of the SIRT5 gene does not appear to affect expression levels of SIRT5 as it does in the cerebral cortex. These findings at the molecular level are consistent with findings from studies of brain structure with age that show robust decreases in gray matter in the cerebral cortex but more variable decreases in the amygdala (Good et al., 2001). The effect of the low-expressing SIRT5 polymorphism on molecular age was accompanied by expression changes for a set of genes whose products are localized to the mitochondria, including PINK-1 and DJ-1, two Parkinson’s disease-associated genes, in ways that would promote mitochondrial dysfunction-related diseases, including Parkinson’s disease. This (correlative) proof-of-principle study suggests that factors that affect brain aging can potentially place an individual at higher risk for disease, through a mechanism by which it accelerates brain biological aging, and thus promotes changes in expression of disorder-relevant genes in disease-causing directions. With respect to Figure 1B, the low-expressing polymorphism of the SIRT5 gene can be thought of a modulator that puts one on the “at risk” trajectory. The converse of this model is that factors delaying age trajectories of gene changes may lead to younger brain molecular aging and potential resiliency toward developing functional declines and age-related disorders, including LLD (Figure 1B).

The mechanisms by which age-related changes in gene expression occur are unknown. Candidate mediators include among others, loss of telomere integrity, increased oxidative stress, and epigenetic modifications.

We propose a novel framework for investigating the development of late-life brain disorders, including LLD, which we term the age-by-disease biological interaction hypothesis. This paper expands upon an earlier version described elsewhere (McKinney and Sibille, 2012). This hypothesis posits that symptoms of LLD and other late-life brain disorders are the emerging properties of underlying biological changes, which in turn are supported by normal changes in the expression of multiple genes with age, including disease-related genes changing in disease-causing directions. Here, in addition to presenting the gene expression data on which the hypothesis is based, we discussed molecular mechanisms that may account for age-dependent gene expression changes, including loss of telomere integrity, increased oxidative stress, and epigenetic modifications. Importantly, this hypothesis complements existing hypotheses, including the vascular, inflammatory, and dementia prodrome hypotheses of LLD, but it differs in that it positions age-dependent gene expression changes as the mechanism potentially driving dysfunctions in multiple biological pathways, including vascular, inflammatory, and neurotrophic functions. A potential sequence of events is summarized in Figure 2. The purpose of this paper was to discuss the general framework. Examples of gene changes at the intersection of depression and aging were provided (e.g., SST and BDNF), but the exact nature and complexity of changes in multiple genes and pathways and their relevance to disease pathways will be described in details elsewhere.

The implications of this hypothesis for the prevention and treatment of LLD and other late-life brain disorders are exciting. Understanding the mechanisms mediating age-related changes in gene expression is expected to provide insight into pathophysiological mechanisms and potential targets for intervention into these disorders. Identifying key upstream hub genes mediating patterns of altered age-dependent changes would provide novel targets for further investigations. Although sirtuins and BDNF may represent obvious candidates, the large set of age-dependent genes (~10% of all genes; Erraji-Benchekroun et al., 2005) and its overlap with genes previously implicated in brain disorders (Glorioso et al., 2011) should be viewed as an enriched pool of candidate genes. Targeting such upstream factors (transcription or function) should represent productive research avenues. Early candidate interventions may include known interventions such as antidepressant medications, psychotherapy, exercise, and others. Investigating how these known interventions affect age-dependent changes in the function of critical genes may help in optimizing their implementation with respect to timing and duration of intervention for age-dependent disorders. Further, identifying genetic and environmental factors that slow or accelerate age-related changes in gene function may lead to individualized strategies aimed at promoting resilience and successful aging. Other entry points and targets for intervention are likely to arise out of understanding the mechanisms by which gene expression changes with age, including determining the role of telomere integrity, oxidative stress, and epigenetic modifications. Finally, for the broader fields of aging and gerontology, the implication of this hypothesis is that it brings together research on normal aging more closely with the investigation of neuropsychiatric and neurodegenerative diseases. Indeed, our data firmly support the assertion that they may in fact be related facets of similar biological processes, and also provide the basis for a putative mechanism of age-by-disease biological interactions.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.