Like all biological processes, genes play an important role in aging. Within humans and other species, variation both in susceptibility to individual diseases of aging and in overall longevity has a heritable component (Finch and Tanzi, 1997). Across species, overall lifespan can vary by many orders of magnitude, and the shape of the relationship between mortality risk and age is similarly diverse (Jones et al., 2014). Such differences are indicative of a long history of evolutionary change in the aging process, which requires underlying genetic variation on which to act. Finally, targeted manipulations of individual genes in laboratory model organisms have pinpointed specific mutations that have strong effects on longevity. Mutations in the insulin-like growth factor receptor gene daf-2 in the worm Caenorhabditis elegans, for example, result in a more than 2-fold increase in lifespan (Kenyon et al., 1993). Effects of a similar magnitude have been identified in studies of fruitfly (Drosophila melanogaster) and mouse (Mus musculus) mutants (reviewed in Kenyon, 2010). Together, the cumulative evidence from quantitative genetic analysis, cross-species comparisons, and model systems research argues that understanding aging will necessarily involve, at some level, understanding the role played by genes.
How can a gene-centered perspective be integrated into aging biodemography? Unlike studies of the genetics of aging in lab models, which take advantage of large effect mutations, genetic clonality, and environmental homogeneity to isolate aging-related genes and pathways, biodemographers
are fundamentally interested in aging in natural populations—in which individual genetic effects on aging will generally be modest, and individual study subjects will be genetically diverse. In addition, biodemographers are often explicitly interested in the consequences of natural environmental heterogeneity, much of which cannot be replicated in the lab. For example, social environmental effects are risk factors of substantial demographic importance in humans and other socially complex animals, because social adversity both occurs at high frequency and confers high relative risk for mortality and many of the major diseases of aging (Berkman and Syme, 1979; House et al., 1988; Sapolsky, 2004; Sapolsky, 2005; Marmot, 2006; Holt-Lunstad et al., 2010). Thus, gene-focused approaches will be most relevant to biodemographers of aging not when they exclude environmental variance, but when they contribute to a mechanistic or predictive understanding of how environmental risk factors act to influence aging-related phenotypes.
Recently, the field of genomics has shifted from a primary emphasis on the static sequence composition of the genome to an emphasis on its functional organization and potential (e.g., Dunham et al., 2012, for an impactful recent example of this shift). This “functional genomic” perspective suggests how studying genes—and particularly gene regulation—could yield valuable insight into the comparative biodemography of aging. The functional genomic perspective treats gene regulation as a dynamic process influenced by a combination of intrinsic genetic effects, extrinsic factors, and demographic variables like age and sex. Hence, functional genomic data promise to shed new light onto the mechanistic basis of biodemographically important environmental effects, including the enduring puzzle of how adverse social environments “get under the skin” to influence health (Taylor et al., 1997; Adler and Ostrove, 1999; Hyman, 2009). For example, such approaches can be used to investigate which genes are mutually affected by social adversity and age, how gene regulation contributes to social environmental effects on health, and whether gene regulatory responses to social stressors vary across species in a predictable manner.
Functional genomic approaches have already been embraced for aging research in laboratory model systems (Vijg and Suh, 2003; Partridge and Gems, 2006; de Magalhães et al., 2012). They have been less explored in human population studies or in nonmodel systems, which include many of the species in which social environmental effects on health most closely parallel those in humans. However, recent studies have established a strong link between social conditions and gene regulation in both of these contexts (reviewed in Slavich and Cole, 2013; Tung and Gilad, 2013). Using social environmental variation as a focal point, this paper considers the potential for a closer integration of functional genomics and biodemography in studies of populations outside the lab. To do so, I first review the evidence that the functional genomics of gene regulation is important for understand-
ing aging, and I outline the arguments in favor of collecting genome-scale data. I then consider how functional genomic studies and biodemography could inform one another and, jointly, an understanding of social environmental influences on aging. Finally, I close by suggesting future directions and discussing the potential prospects for this approach.
THE RATIONALE FOR ADOPTING A FUNCTIONAL GENOMIC PERSPECTIVE
Gene Regulation in Aging
Functional genomics focuses on the biochemical potential and activity of the genome, including the molecular changes that influence gene activation, the determinants of RNA and protein synthesis and decay, and the factors that affect the binding and conformation of nucleic acids and proteins. Hence, functional genomics is closely tied to the study of gene regulation, and adopting a functional genomic strategy to study aging relies on the assumption that changes in gene regulation are an important component of the aging process.
Support for this assumption comes from several independent sources. First, shared pathways are involved in aging in a broad set of species, yet produce widely variable life history outcomes. For example, although C. elegans can live for days, fruitflies for weeks, and mice for years, genes involved in the stress response and nutrient sensing are linked to control of aging in all three species (Partridge and Gems, 2002; Kenyon, 2010; López-Otín et al., 2013). This remarkable conservation suggests that evolutionary shifts in expected lifespan are not due to changes in the identity of the genes involved, but instead stem from changes in the regulation of these genes—in other words, when and to what degree the molecular species that mediate aging are employed, as opposed to which molecules are used. This argument echoes well-established arguments made for the evolution of development, in which striking differences in body plan can arise from regulatory changes in genes that are otherwise highly conserved at the protein-coding sequence level (Carroll, 2005). Indeed, in support of a parallel pattern for aging, cross-species comparisons have revealed that the protein-coding regions of genes linked to aging tend to be evolutionarily conserved (de Magalhães and Church, 2007).
Changes in gene regulation also account for dramatic changes in aging and longevity that can arise within a single species in response to environmental cues. This pattern is most clearly demonstrated in taxa that are capable of adopting discrete, alternative life histories, such as eusocial insects. For example, honey bees are capable of developing into either non-reproductive worker bees, with an expected lifespan of several months, or
reproductively active queen bees, which can live up to five years (Amdam, 2011). They do so based on nutritional cues provided in early life, which activate distinct gene regulatory programs without requiring changes in the DNA sequence itself (Evans and Wheeler, 1999; Kucharski et al., 2008; Elango et al., 2009). Similarly, isogenic (i.e., genetically clonal) lines of the nematode Strongyloides ratti can develop into either a short-lived, free-living form or a long-lived parasitic form, which are differentiated by an over 80-fold difference in expected lifespan (Gardner et al., 2006). Less dramatic shifts can be induced in species without distinct alternative life histories. In lab model systems, environmental manipulation of temperature or diet can substantially change how long animals live: For example, C. elegans subjected to dietary restriction can live up to 150 percent longer than those fed ad libitum (Greer and Brunet, 2009). Such shifts clearly involve variation in gene regulation as opposed to variation in DNA sequence, as no genetic variation is present in the sample. Together, they have given rise to the observation that animals in general likely have “the latent potential to live much longer than they normally do” (Kenyon, 2010).
Finally, direct empirical evidence ties gene regulation to aging. Environmental interventions that influence aging and longevity often do so via their effects on gene regulation (e.g., Pletcher et al., 2002; Fok et al., 2014), and longevity-associated mutations are often components of gene regulatory pathways. For example, the C. elegans gene daf-16 (as well as its fruitfly and mammalian homologues) is a transcription factor that regulates hundreds of downstream genes (Murphy et al., 2003). Age is also closely tied to variation in gene expression levels in unmanipulated populations. For example, chronological age explains variation in the expression levels of approximately half the genes expressed in the human prefrontal cortex, a pattern that is qualitatively recapitulated in other tissues and species (Lund et al., 2002; Lu et al., 2004; Fraser et al., 2005; Göring et al., 2007; Berchtold et al., 2008; Hong et al., 2008; Cao et al., 2010; Somel et al., 2010; Yuan et al., 2012). While other gene regulatory phenotypes have been less well studied in relation to age, where data are available, they show similar patterns (Christensen et al., 2009; Fraga, 2009; Han et al., 2012). In particular, epigenetic marks—chemical modifications to the genome that can influence transcription rates and gene expression levels—are closely associated with the aging process (Fraga, 2009). DNA methylation, histone acetylation, and histone methylation marks all have been shown to either covary with age or modify aging-related phenotypes (reviewed in Fraga, 2009; López-Otín et al., 2013). Indeed, epigenetic patterning is so consistently altered during aging that it is now widely considered to be one of the major molecular hallmarks of aging (López-Otín et al., 2013).
Genomic Approaches to Gene Regulatory Studies of Aging
Taken together, the cumulative evidence strongly supports the importance of gene regulation in aging. However, variation in gene regulation also reflects the effects of important environmental variables, and for humans and other socially complex species, social status, social support, and social competition are central components of the environment. Gene regulatory phenotypes—measurements that capture variation in gene regulation, including gene expression levels and epigenetic marks—can thus provide a useful common currency for studying how social experiences influence aging-related genes. While studies of single genes or small sets of genes can take advantage of this approach as well, it can be particularly valuable when applied to genome-scale datasets (see Box 3-1) for two key reasons (see also Robinson et al., 2005, 2008; Boyce et al., 2012).
First, genome-wide datasets allow the generality of relationships between the social environment and aging to be assessed. For example, social stressors have been hypothesized to influence biological targets that are also affected by age, potentially accelerating the aging process (Bauer, 2008). This idea predicts that physiological changes associated with social stress should also be associated with age, and that the direction of these effects should be positively correlated. Studies of a few individual biomarkers, such as IL-6 cytokine levels and telomere length, have lent support to this argument (Epel et al., 2004; Cherkas et al., 2006; Juster et al., 2010; but see Dowd and Goldman, 2006, who report few consistent relationships between social adversity and biomarkers of chronic stress, challenging a major assumption of the argument). However, these individual cases are not sufficient to reveal a general pattern, especially given that age is associated with a large set of biological pathways (identification of an overlap for a single biomarker could thus be attributable to chance alone). Because
Genome-Wide Measures of Gene Regulation in Population Studies
In recent decades, population studies have pioneered collection of physiological and molecular biomarkers of health, and such data are now a routine component of many population-based analyses of aging. This paper outlines how genome-wide measures of gene regulation might serve to further advance such a synthesis; here, I address several practical considerations linked to this possibility.
What is “genome-wide?”
“Genome-wide” is a flexible term and depends in part on the aspect of gene regulation under study. For example, a human genome-wide gene expression profile might involve ~ 10,000–20,000 genes, but a complete survey of DNA CpG methylation could involve ~ 450,000 sites (the number measurable using current off-the-shelf methods) or > 25 million sites (all sites in the genome). Regardless of numbers, genome-wide studies of gene regulation are unified by the attempt to generate enough measurements to capture the true distribution of trait values across the genome. This approach allows investigators to test whether a given association is unusual or widespread in comparison to the rest of the genome (as opposed to whether it is unusual in comparison only to a theoretical null). Genome-wide approaches thus can provide valuable biological insight unavailable from smaller-scale studies and help to identify potential systematic sources of bias.
How are genome-wide gene regulatory phenotypes measured?
Genome-wide assays of gene regulatory phenotypes vary depending on the phenotype of interest. However, most genome-wide assays now converge on one of two strategies: array-based methods or high-throughput sequencing methods. The basic difference between these methods depends on whether measurements are made based on matches to a pre-specified set of sequences (on arrays) or based on the counts of de novo-generated sequences associated with a regulatory feature of interest (sequencing-based approaches). The choice of one approach versus the other depends on cost (often favoring arrays), sensitivity (favoring sequencing), and feasibility (commercial arrays are not available for most nonmodel species). See Marioni et al., 2008; Mortazavi et al., 2008; Wang et al., 2009; Hawkins et al., 2010; and Metzker, 2010, for recent comparisons and reviews.
Are genome-wide approaches realistic for population studies?
Three major considerations are at play here: (1) Sample quantity. Many approaches for measuring genome-wide gene regulatory phenotypes are able to utilize very small amounts of sample (e.g., < 50 uL of blood). For some types of analyses, studies that already collect biological samples are therefore well positioned to either modify existing protocols or utilize previously collected samples. (2) Cost. Generating genome-wide data, especially using sequencing-based approaches, remains prohibitively expensive for large sample sets (i.e., in the hundreds to thousands of individuals). Investigators will therefore often have to choose informative subsets of individuals, rather than the full sample, to interrogate (potentially banking additional samples for later analysis). Notably, sample sizes in functional genomic studies are often small by population survey standards. (3) Cell/tissue type. Biological samples that can be obtained using minimally invasive approaches (e.g., blood draws) will be the most feasible to study in population surveys. While other tissues, such as the brain, are clearly also of interest, changes in the periphery, including in blood, have consistently proven to be important in the response to social adversity, suggesting that this sample type will provide important first insights into social environmental effects on aging. Over time, opportunistic sampling may make it possible to eventually investigate other tissues as well (e.g., McGowan et al., 2009).
genomic approaches generate data for many gene regulatory traits simultaneously, they facilitate more comprehensive tests of parallels between social environmental variation and aging. At the same time, genomic datasets provide important information about whether the effects of a particular social environmental variable are concentrated in a handful of pathways versus broadly distributed across the genome.
Second, genome-wide datasets facilitate the use of analysis tools targeted above the level of individual genes. For example, gene set enrichment approaches help test whether genes that share a particular property (e.g., involvement in a specific biological function or known association with a disease trait) appear more often than expected by chance among a group of genes identified in the focal analysis (e.g., among genes for which gene expression covaries with age). Such tools allow researchers to utilize prior knowledge about gene function, pathway membership, or other associations to investigate whether genes linked to a particular environmental effect are biologically coherent in other ways (Subramanian et al., 2005; Backes et al., 2007; Huang et al., 2008). Gene set approaches have been useful, for instance, in revealing that genes that are differentially expressed in response to social adversity are often related to inflammation and glucocorticoid signaling (reviewed in Slavich and Cole, 2013). Importantly for comparative studies, because the units of analysis are collections of genes, enrichment analyses do not require exactly the same loci to be measured in different datasets. Especially when individual studies are low powered, similarities between different populations, species, or environmental conditions may be more detectable at the level of gene categories than at the level of individual genes. However, pathway-level, rather than gene-level, similarities may also reflect real biological patterns.
Research on aging in model systems supports the latter explanation. Pathway-level effects on aging are remarkably conserved across species: Changes in the insulin/insulin-like growth factor signaling pathway, for example, affect the aging process in species from C. elegans to humans—taxa that have been independently evolving for hundreds of millions of years (Partridge and Gems, 2006; Kenyon, 2010). In contrast, changes in gene expression levels associated with chronological age, longevity-enhancing mutations, or longevity-enhancing environmental treatments are rarely replicable at the individual gene level (McElwee et al., 2007; de Magalhães et al., 2009). Instead, similarities are primarily observable at the level of functionally related gene sets: advanced age is consistently associated with downregulation of genes involved in energy metabolism and upregulation of genes involved in the immune response, apoptosis, and insulin-like growth factor binding (de Magalhães et al., 2009). Focusing on one or a handful of genes is likely to miss these similarities, as well as the opportunity to assess where differences truly lie.
FUNCTIONAL GENOMICS AND THE BIODEMOGRAPHY OF SOCIAL ENVIRONMENTAL VARIATION
Functional genomic studies of aging and studies of social environmental effects on health and senescence have largely proceeded along parallel lines. They are bridged, however, by recent evidence that social environmental variation also impacts gene regulation. In humans, self-perceived loneliness, early and adult socioeconomic status, and provision and receipt of social support have each been linked to variation in the expression levels of hundreds of gene transcripts (Cole et al., 2007; Miller et al., 2008; Chen et al., 2009; Lutgendorf et al., 2009; Miller et al., 2009). The causal nature of at least some of these relationships is supported by work in animal models. For example, social status (i.e., dominance rank) and early rearing environments can be experimentally manipulated in captive rhesus macaques, which, like humans, have evolved to navigate complex social group environments (Cole et al., 2012; Tung et al., 2012). These manipulations lead to extensive changes in gene expression patterns in the blood, as well as accompanying changes in DNA methylation (Provencal et al., 2012; Tung et al., 2012). Further, in rodent models, targeted changes in epigenetic gene regulatory mechanisms have been successfully demonstrated to reverse some of the negative behavioral consequences of social adversity (Weaver et al., 2004; Tsankova et al., 2006). For example, chemical repression of HDAC5, a gene involved in epigenetic patterning, restored social interaction rates in socially defeated mice to normal (higher) levels (Tsankova et al., 2006).
Combined, these findings suggest that gene regulatory phenotypes (including, but not limited to, gene expression levels) may act as useful biomarkers for tracking the association between social environmental variation, health, and aging. In addition, they suggest potential research avenues aimed at understanding the mechanistic basis of this relationship. To do so will require forging stronger links between population studies of health and aging on one hand, and functional genomic approaches for studying the social environment on the other. Below, I outline two broadly defined strategies for pursuing this research agenda. The first involves using biodemographic observations as a motivation for functional genomic studies, with the aim of understanding how, and in what contexts, gene regulatory changes with age intersect with gene regulatory effects of the social environment. The second involves development of a tighter integration between functional genomics and biodemography, in which demographic concepts of mortality rates and senescence can be leveraged to investigate the role of gene regulation in aging.
Biodemographically Motivated Studies
In biodemographically motivated studies, social environmental variables of known biodemographic importance can be tested for their effects on gene regulation using standard functional genomic methods, without explicit reference to demographic parameters or theory. Current research on the relationship between gene regulation and social adversity implicitly uses this approach (Cole et al., 2007; Miller et al., 2008; Chen et al., 2009; Lutgendorf et al., 2009; Miller et al., 2009; Cole et al., 2012; Tung et al., 2012): The rationale for studying the effects of social status and social integration in this work derives in large part from population-based research highlighting their importance as predictors of disease susceptibility and mortality risk (Berkman and Syme, 1979; House et al., 1988; Holt-Lunstad et al., 2010). Jointly, these studies have established that social environmental variation influences both health and longevity and the molecular control of gene regulation. The next steps forward for biodemographically motivated studies will involve, first, establishing the degree to which these observations reflect a shared phenomenon, and second, investigating how the type and timing of social adversity affects its downstream consequences.
Does the Social Environment Affect Aging-Related Pathways?
Circumstantially, it makes sense that gene regulatory changes in response to social adversity are related to aging. Aging-related genes and pathways are often involved in the response to environmental stressors. Altering the levels of environmental stress by manipulating ambient temperature or caloric intake represent some of the most robust ways to alter life expectancy in lab model systems (Kenyon, 2010; López-Otín et al., 2013), and dietary restriction also may confer health and life-extending benefits in primates (Colman et al., 2009; Mattison et al., 2012). Importantly, for highly social species like humans, the social environment acts as one of the most potent sources of environmental stress. Consequently, the effects of social adversity on gene regulation may also mediate aspects of the aging process. In general, however, the effects of socially mediated stress on aging have not been evaluated in the lab environment. This is probably in part because the types of social adversity important in humans often have no direct parallels in the major lab models for aging (e.g., C. elegans, yeast, and Drosophila, although mouse and rat models represent a partial exception). Hence, the degree to which social adversity affects the regulation of genes that are also affected by aging remains unclear. It thus remains conceptually possible that social environmental effects on gene regulation and social environmental effects on aging in fact tap into independent pathways.
Genome-wide datasets on both age-associated and social environment-associated gene expression patterns provide one approach for differentiating between these two alternatives. Specifically, they facilitate an exploration of whether, and for what genes, socially induced changes in gene expression match shifts in gene expression linked with aging. While such overlaps provide only indirect evidence for a mechanistic relationship, they can at least exclude the possibility that changes with age and changes with the social environment are broadly independent. They also have the potential to greatly enrich the small set of existing biomarkers known to be influenced by both effects. As illustrated by telomere length in blood cells (Epel et al., 2004; Cherkas et al., 2006; Boonekamp et al., 2013), which both decays faster in chronically stressed women and is a predictor of mortality risk (Epel et al., 2004), such markers can serve as useful measures for capturing aspects of biological “age” that do not map well onto chronological age (but see Boonekamp et al., 2013, who argue that telomere length is a better marker of somatic redundancy than biological age).
Two analyses suggest that gene regulatory responses to the social environment are indeed similar to those observed during aging. As part of a study of sex differences in aging in the prefrontal cortex, Yuan and colleagues observed a pattern of accelerated age-related change in gene expression levels in human women relative to men (Berchtold et al., 2008; Yuan et al., 2012). They hypothesized that this pattern might reflect greater vulnerability to social stressors in women. Suggestively, genes that exhibited the female-biased pattern were significantly more likely to be altered in the same direction by social isolation stress, based on data from the prefrontal cortex of spider monkeys (Karssen et al., 2007). Likewise, blood expressed genes for which gene expression levels were associated with both age (in humans: Göring et al., 2007) and social status (in rhesus macaques: Tung et al., 2012) were significantly more likely to be changed in the same direction (Snyder-Mackler et al., 2014). Genes that were more highly expressed in low social status macaques tended to be more highly expressed in older humans, and vice-versa. Consistent with comparisons between disparate aging datasets, similarities between gene expression changes with age and the gene expression effects of social status were stronger for functionally coherent gene sets than for single loci. For example, genes involved in insulin growth factor I signaling and genes involved in inflammation—two processes tightly associated with aging—were identified in both the human age-associated (Göring et al., 2007) and rhesus macaque social status-associated (Tung et al., 2012) datasets. Overall, the number of these overlapping pathways was substantially greater than expected by chance (Snyder-Mackler et al., 2014). Such findings thus provide initial support for the argument that social environment-induced changes in gene regulation may also explain something interesting about aging. Additional datasets
would greatly improve the resolution of these comparisons, however, and studies in which the effects of age and the social environment are studied in the same population will be particularly important.
Does the Social Environment Impact Known Hallmarks of Aging?
A second way to investigate whether social environmental effects on gene regulation are involved with aging is to test whether they impact biological phenomena believed to be fundamental to the aging process. Recently, López-Otín and colleagues identified nine basic hallmarks of aging that have received broad support in aging research on nonmodel organisms and humans (López-Otín et al., 2013). Several of these hallmarks, particularly changes in epigenetic patterning and altered intercellular communication, provide natural bridges between aging and social environment-associated changes in gene regulation.
The first, epigenetic patterning, plays a fundamental role in determining the three-dimensional configuration of DNA and hence its accessibility to the cell’s transcriptional machinery. Epigenetic marks, including DNA methylation, histone methylation, and histone acetylation levels, show marked patterns of age-related change, and targeted manipulation of some epigenetic regulatory mechanisms affects longevity in model organisms (reviewed in Fraga, 2009; López-Otín et al., 2013). Changes in the epigenome also arise as a consequence of social environment-mediated life experiences. Early rearing environment is associated with differential DNA methylation levels in rodent models, nonhuman primates, and human populations (Champagne et al., 2003; Weaver et al., 2004; Champagne et al., 2006; McGowan et al., 2009; Murgatroyd et al., 2009; Roth et al., 2009; Bagot et al., 2012; Borghol et al., 2012; Provencal et al., 2012). Further, experimental manipulation of social status in rhesus macaques is linked to a large, broadly distributed set of differentially methylated regions across the genome (Hansen et al., 2012; Tung et al., 2012). The degree to which these changes overlap with those involved in aging remains unknown. However, in the macaque prefrontal cortex, age-associated changes in histone H3K4me2 (histone 3, lysine 4, dimethylation) methylation levels, which are markers of transcriptional activation, are preferentially observed near inflammation and environmental stress-related genes (Han et al., 2012). This observation suggests that social adversity may influence at least some aging-related epigenetic pathways.
Similarly, studies of the gene regulatory response to social adversity frequently identify genes involved in intercellular communication, a second major hallmark of aging. These genes include the cellular targets of steroid hormone and sympathetic nervous system signaling and, most consistently, genes involved in the inflammatory response. For example,
multiple studies of social adversity-linked gene expression have highlighted genes that are likely regulated by the inflammation-related transcription factor NFkB (e.g., Cole et al., 2007; Chen et al., 2009; Miller et al., 2009). NFkB-regulated genes tend to be upregulated in poor quality social environments, suggesting that social adversity produces a chronic condition of low-grade inflammation—a state also associated with aging (Larbi et al., 2008; López-Otín et al., 2013). Strikingly, targeted repression of NFkB in mouse skin has been shown to reduce markers of cellular senescence, increase cell proliferative potential, and change genome-wide gene expression profiles to match characteristics of young mice (Adler et al., 2007; see also Tilstra et al., 2012). NFkB signaling is thus a promising pathway for investigating shared mechanistic links between social adversity and aging. Indeed, NFkB is repressed by hypothalamic-pituitary-adrenal axis-mediated glucocorticoid (GC) signaling, another pathway associated with social adversity. Although chronically stressed individuals may exhibit elevated GC levels, they also can become insensitive to GC signaling at the cellular level, perhaps in part due to downregulation of glucocorticoid receptor transcript levels (Tung et al., 2012).
A focus on the regulation of specific aging-related pathways therefore promises to improve the understanding of how social environmental effects influence the aging process. In natural populations, the kinds of invasive manipulations possible in lab model systems, or even captive primates, will generally be impossible. However, assays of the gene regulatory response to experimental stimuli in primary (i.e., not immortalized) cells outside the body can be a useful way to investigate how environments experienced by the whole organism impact the function of individual cells. For example, white blood cells collected from study subjects who were low socioeconomic status early in life present a more pro-inflammatory phenotype following experimental stimulation than individuals who were high socioeconomic status in early life (Miller et al., 2009). These observations suggest potentially powerful approaches to directly test, using experimentally controlled conditions, whether aging-related signaling pathways can be differentially activated in low social adversity versus high social adversity individuals (including in a cross-species comparative context: see Barreiro et al., 2010 for an example investigating species differences in the innate immune response). Such assays facilitate a number of downstream gene regulatory trait assays. For example, in addition to measuring changes in gene expression levels, NFkB and glucocorticoid receptor protein-DNA binding can be directly measured using chromatin immunoprecipitation techniques to compare stimulated versus unstimulated cells (e.g., Reddy et al., 2009; Wang et al., 2012; Luca et al., 2013).
The Types and Timing of Social Environmental Effects
Work on the genomic response to social environmental variation is still in its early days. Consequently, the tendency is to emphasize similarities across studies, which reinforce their reproducibility and generalizability. Indeed, studies of social environmental effects on gene expression do tend to identify more overlapping genes than expected by chance (Tung and Gilad, 2013). These commonalities suggest that different studies may be pointing to a set of shared gene regulatory pathways, which some investigators have termed a “conserved transcriptional response to adversity” (Slavich and Cole, 2013). However, even in highly controlled lab settings, the consequences of aging-related environmental treatments are highly dependent on the nature of the intervention. For example, at least nine different dietary restriction protocols have been studied in the context of aging in C. elegans, and their effects range widely from about a 25 percent to a 150 percent increase in expected lifespan (Greer and Brunet, 2009). The timing and magnitude of environmental and molecular manipulations also matter. Mild heat shock, which is lifespan extending in C. elegans, is more protective if applied intermittently throughout life than during a single event early in life (Wu et al., 2009). In contrast, the protective benefits of lowering cellular respiration rates seem to be specific to larval development (Rea et al., 2007).
It seems reasonable to expect that any effects of social environmental variation on aging in natural populations will be at least as nuanced. In addition to investigating conserved social environmental signatures (either between studies or with signatures of age), functional genomic studies can also be used to dissect context-specific effects. Several types of context-dependency are already suggested in the literature. For example, both monocytes (Cole et al., 2011, 2012), which primarily function as part of the innate immune system, and cytotoxic T cells (Tung et al., 2012), which function in adaptive immunity, have been implicated as important in the response to social adversity. Similarly, observations that gene expression changes with age seem to be accelerated in human women compared to men, potentially due to differences in exposure or susceptibility to stress, suggest the potential importance of sex differences (Yuan et al., 2012). Finally, timing-dependent effects are indicated by the observation that early life social adversity can lead to long-term, stable changes in gene regulation. These changes may often be mediated by epigenetic patterning, as in the case of the long-term effects of maternal rearing environment on the regulation of brain- and blood-expressed genes. Interestingly, sites responsive to maternal rearing environment in the rat arginine vasopressin promoter appear to be protected from a general pattern of aging-related hypomethylation elsewhere in the region (Murgatroyd et al., 2010). Under-
standing timing-dependency will thus be key for understanding plasticity in, and the reversibility of, social environmental effects (including, as some have argued, whether differing levels of plasticity are adaptive: Boyce and Ellis, 2005; Ellis et al., 2011). Importantly, some studies indicate that social environment-associated epigenetic marks do not irreversibly crystallize in early life. For example, changes in social status in adulthood alter genome-wide DNA methylation levels in rhesus macaques, and perceived stress in adult humans is associated with a similar number of changes in DNA methylation levels as childhood socioeconomic status (Lam et al., 2012; Tung et al., 2012).
Finally, functional genomic analyses can contribute to a better understanding of how multiple types of social adversity, or social conditions acting at different periods in the lifecourse, jointly act to influence health. The degree to which social environmental effects act independently has important ramifications for understanding whether they act through independent molecular mechanisms. Other environmental effects linked to aging sometimes do affect discrete pathways. In C. elegans, for instance, dietary restriction and lowered temperature act additively to produce more dramatic effects on lifespan extension than either alone (Yen and Mobbs, 2010). It will be interesting to learn whether different social environmental factors also act additively. A good place to start would be in discriminating the gene regulatory consequences of social status versus social isolation—two distinct sources of social adversity that are often discussed together, but that need not be correlated. Functional genomic data would help resolve whether, and to what degree, social status and social integration act additively or nonadditively, as well as whether additive effects arise because they affect different molecular mechanisms.
Biodemographically Integrated Studies
Functional genomic studies have primarily defined aging-related gene regulatory phenotypes as those that change linearly with chronological age. This approach has been very useful for demonstrating the pervasiveness of associations between age and gene regulation. However, it leaves a great deal unanswered about how, exactly, a given regulatory phenotype is linked to the aging process—that is, the pace at which health declines and mortality rates increase over time. In demographic terms, aging is not described by the clocklike passage of time, but instead by a progressive loss of physiological function that results in an increase in age-specific mortality rates across the lifecourse. Biodemographic approaches allow the shape of this relationship to be quantitatively parameterized, and differences in these parameters across environmental conditions, populations, and species are of fundamental interest to the field. Integrating a biodemographic perspec-
tive on aging with functional genomic approaches therefore promises to provide novel insights into how social environmental variation affects the aging process. In this section, I detail potential avenues through which this integration could progress, recognizing that, at this early stage, these possibilities remain speculative.
Using Biodemographic Concepts to Identify Biomarkers of Social Environmental Effects on Aging
In humans, as well as in the animal species generally used as models for social adversity, the relationship between mortality rate and age tends to have a characteristic “bathtub” shape (see, for example, Figure 1 in Bronikowski et al., 2011, and Figure 1 in Jones et al., 2014, for the period following maturity). This relationship reflects a high risk of mortality in the post-natal period that reduces to a period of low risk during development and adolescence. Senescence begins after reproductive maturity and is characteristically marked by an exponential increase in mortality rate over time. In this last stage, mortality rate can be modeled as a function of age using one of a family of generalized logistic functions, most commonly the Gompertz function.
This relationship states that mortality rate at any given age is a consequence of both the initial mortality rate—a constant offset—and the pace at which mortality rates increase with age. When age-specific mortality rates are log-transformed, these two parameters become, respectively, the intercept and slope of a linear model. The Gompertz model provides a convenient way of expressing the relationship between the linear progression of chronological age and the non-linear increase in mortality rate (or physiological decline) associated with getting older. Genetic variation or environmental conditions that affect lifespan can do so by changing the Gompertz intercept, slope, or both. Importantly, though, because the intercept affects individuals of all ages equally, only the slope captures age-dependent changes in mortality risk. Thus, some investigators have emphasized that environmental conditions that alter the Gompertz intercept may be unrelated to aging, because they do not change the rate of senescence over time (Jacobson et al., 2010). Perhaps counterintuitively, though, they actually can reverse an individual’s risk of death to the level associated with being chronologically younger. Dietary restriction in fruitflies, for example, seems to operate in this manner (Mair et al., 2003; Jacobson et al., 2010).
Defining whether a given social environmental variable influences the Gompertz intercept or the Gompertz slope can therefore provide a useful framework for testing the relationship between social adversity, aging, and putative functional genomic biomarkers of both processes. In model systems, genetic and environmental effects on the Gompertz intercept are
often independent of effects on the Gompertz slope, implying that these two components of mortality rate are affected by discrete molecular mechanisms (Mair et al., 2003; de Magalhães et al., 2005; Wu et al., 2009; Yen and Mobbs, 2010; Sarup et al., 2011; Kelly et al., 2013). Several investigators have formalized this prediction with respect to biomarkers of aging (Jacobson et al., 2010; Kelly et al., 2013). According to this logic, aging is an irreversible process, as no interventions have successfully produced a decrease in mortality rates with age in senescing organisms (although this phenomenon, termed “negative senescence,” does seem to occur in some species in nature: Jones et al., 2014). Hence, biomarkers of aging, which reflect the Gompertz slope, should themselves be irreversible even if the environment improves: change in these biomarkers reflects uncorrectable, accumulated molecular damage. In contrast, biomarkers that reflect the Gompertz intercept should be capable of reverting to the state associated with younger individuals, but should continue to change with age at the same rate.
These predictions have been used to identify advance glycation end products (AGEs) as a biomarker of the rate of aging in fruitflies (Jacobson et al., 2010). Uniquely among a set of putative biomarkers, the rate of AGE accumulation slowed, but AGE levels did not reverse, in concert with temperature shifts known to influence the Gompertz slope. However, AGE accumulation did not change in association with the imposition of dietary restriction, which only influenced initial mortality rate. As the number of putative gene regulatory biomarkers of both the social environment and aging accumulates, it should be possible to perform similar kinds of analyses to support or refute the likelihood that individual markers of social adversity are mechanistically connected to aging. Unlike in fruitflies, experimental approaches that alter social context will often be practically and ethically unfeasible (although interventions in captive social primates could in principle be conducted). However, studies in both human and animal populations often document cases in which an individual’s social environment changes over time. Indeed, some of these studies have been highly influential in making the case for the long-term effects of early social adversity. For example, even among highly educated, socioeconomically privileged adults, early-life socioeconomic status can still predict rates of cardiovascular disease later in life (Kittleson et al., 2006).
Understanding whether social environmental conditions influence the Gompertz slope or the Gompertz intercept may also motivate the types of gene regulatory phenotypes investigators wish to investigate. Epigenetic marks, in particular, have been associated with the kind of long-term “memory” of past insults that might suggest a closer relationship with the rate of aging than with initial mortality rate. Importantly, studies in model systems have demonstrated that variation in both Gompertz parameters is
clearly reflected in functional genomic data (Pletcher et al., 2002; Sarup et al., 2011; Fok et al., 2014). For instance, artificial selection for increased longevity in Drosophila resulted in a marked change in expected lifespan primarily accounted for by a change in the Gompertz intercept. This change was mirrored by shifts in gene expression levels for genes known to be correlated with age, in which old flies from long-lived lines exhibited gene expression patterns that clustered with control flies at a younger age (Sarup et al., 2011).
Functional Senescence in Gene Regulation
While senescence often refers specifically to the pace at which mortality rates change with age, physiological function also declines with age. Understanding the properties of this decline, which can be usefully distinguished as functional (as opposed to demographic) senescence (Grotewiel et al., 2005), may also shed light onto how social environmental effects on gene regulation impact aging. For example, Bronikowski and Promislow (2005) suggested that mechanistic links between a given trait and the aging process could be indicated when the shape of the relationship between age and that trait mirrored the shape of the relationship between age and mortality rate.
To investigate this possibility will require moving away from testing simple linear relationships between age and gene regulatory traits towards investigating whether they, too, can be modeled using non-linear functions like the Gompertz. This seems likely, as exponential decay is a general property of many physical and biological systems, including protein decay and the regulation of RNA decay within cells (Levinthal et al., 1962). If so, then it will be particularly interesting to examine cases in which functional senescence in gene regulation recapitulates the parameters describing demographic senescence in mortality rates in the same population. A mechanistic link between the social environment and aging would then be supported if variation in social conditions (e.g., for low versus high social status individuals) produced parallel changes in both functions.
To my knowledge, analyses of this type have not been conducted. However, several studies have investigated non-linear relationships between gene expression levels and age. For example, Somel et al. (2010) showed that age-gene expression relationships in the brains of humans and rhesus macaques can be described by a relatively small set of clusters of discrete non-linear shapes. These clusters appear to be biologically meaningful, as genes within clusters tend to be functionally related. Genes involved in DNA repair, for instance, were enriched in a cluster that exhibited progressive stages of downregulation from birth through early life, stability during adolescence, and then upregulation after adolescence and into old age; metabolism-related genes tended to show the inverse pattern. Interest-
ingly, the two transition points for these curves fell around ages 4 and 25 in humans, suggestively close to the demographic transition points that mark the fall in mortality rate after the early high-risk period and the onset of senescence.
In a second analysis, Yuan and colleagues examined the age trajectories of genes that were significantly differentially expressed with both age and sex (Yuan et al., 2012). They were able to identify distinct sets of genes for which sex predicted differences in log-transformed gene expression for either the onset of aging-related change or the rate of that change. These two patterns are again suggestively similar to the biodemographic concepts captured by the Gompertz intercept and slope parameters. In both studies, age-gene expression relationships were modeled using somewhat arbitrary, but highly flexible functions (polynomial regression and cubic spline fitting), as opposed to generalized logistic functions with a more interpretable meaning vis-à-vis aging. Nevertheless, they indicate that taking a similar approach for genes that are affected by both the social environment and age would help distinguish between genes that are affected by the social environment in age-dependent versus age-independent manners.
The ability to investigate the functional genomic correlates of both age and the social environment in natural populations, outside of the lab, is very recent. Nevertheless, research in the last 10 years has clearly established that (1) both chronological age and demographically defined senescence are associated with variation in genome-wide gene regulatory phenotypes; (2) key aspects of the social environment associated with health and longevity also impact gene regulation, particularly gene expression and DNA methylation levels; and (3) gene regulatory signatures of age and the social environment reflect functionally coherent, and probably intersecting, biological pathways and processes, some of which are related to the known hallmarks of aging.
These findings are important from the perspective of understanding the biodemographic impact of social adversity. However, they are also very broad-brush and do not yet address many basic questions important to identifying which individuals in a population are most vulnerable, or the possibilities for ameliorating the effects of past adversity. For example, while sex differences in the relationship between social adversity and gene regulation probably exist, studies to date have focused either on a single sex, have been too small to investigate differences by sex, or simply have not asked the question. Reanalysis of existing datasets may contribute to a better understanding of these differences (e.g., Yuan et al., 2012). More likely, though, is that a synthetic treatment of differences by sex will
require the accrual of additional studies in more systems, and targeted investigations of both sex and social environmental effects within the same population. Basic differences in vulnerability to social stressors in men and women (Rohleder et al., 2001; Kudielka and Kirschbaum, 2005), as well as sex differences in the relationship between social status and age in many animal models (e.g., in baboons: Altmann et al., 2010), suggest that such studies will be illuminating.
Similarly, comparative studies across species and environmental contexts are sorely needed. To date, there are no published studies of social environmental effects on genome-wide gene regulation in a natural population of any nonhuman species. While captive populations and lab models have thus provided important proof of principle that social environmental variation can impact gene regulation, they explain little about its effects in naturally aging populations. What social status and social integration mean to wild animals may differ from their meaning to captive animals; indeed, some types of social adversity tested in captivity, such as extended maternal separation or peer-rearing conditions, have no analogue in natural populations. Further, gene expression patterns can themselves reflect a signature of captivity, particularly at stress response and inflammation-related genes (Kennerly et al., 2008). Finally, initial mortality rates (the Gompertz intercept) and potentially the rate of aging as well (the Gompertz slope) have been shown to differ between captive and wild primates of the same species (Bronikowski et al., 2002). Thus, all three dimensions of the relationship between social adversity, gene regulation, and aging could differ between captive and wild animal populations. Similar variability may affect studies in humans as well, in which subjects have been recruited from clinical and patient populations as well as from population-representative samples.
Future work targeting an expanded set of systems will thus be important for understanding the generality of proposed links between social environmental variation and gene regulation. For example, they will help reveal whether, as in aging, pathways involved in this relationship are conserved across species, and highlight the contexts in which specific types of social environmental variables matter. Comparative research on the relationship between glucocorticoid levels and social status provides a useful model here. Such comparisons have helped to replace the concept of an invariable relationship between low status and elevated glucocorticoid levels with a model in which this association is expected primarily in species with strictly enforced hierarchies, and for classes of individuals who have few sources of social support (Abbott et al., 2003).
Similar patterns may hold in the case of social environmental effects on gene regulation, and it will be interesting to learn if genes associated with social conditions are therefore more closely linked to aging for some species than others. Because genomic data are becoming increasingly easy
to collect, high-quality phenotypic data will ultimately be the limiting factor in expanding research on social effects on gene regulation to new systems. Populations for which high quality sociobehavioral and demographic data already exist, including large-scale population studies of humans and long-term field studies of social mammals, should therefore be prioritized. Notably, some of these populations provide the opportunity to track social environment-related longitudinal changes in gene regulation as well, something that is yet to be accomplished except at very limited (e.g., single gene) scales (Murphy et al., 2013).
In humans, genotype explains approximately one-third of variation in longevity, with much of the remaining variance accounted for by environmental effects. Variation in the social environment clearly accounts for some of these effects, and probably exerts its influence in part through altering the regulation of genes. Functional genomic approaches suggest strategies for investigating this relationship through either biodemographically motivated or biodemographically integrated approaches. Further, they provide a path forward that can take advantage of the rich phenotypic and demographic data already available in human population studies and field studies of wild social mammals.
Abbott, D.H., et al. (2003). Are subordinates always stressed? A comparative analysis of rank differences in cortisol levels among primates. Hormones and Behavior, 43(1), 67-82.
Adler, N.E., and Ostrove, J.M. (1999). Socioeconomic status and health: What we know and what we don’t. Annals of the New York Academy of Sciences, 896, 3-15.
Adler, A.S., et al. (2007). Motif module map reveals enforcement of aging by continual NF-kB activity. Genes & Development, 21(24), 3244-3257.
Altmann, J., et al. (2010). Life history context of reproductive aging in a wild primate model. Annals of the New York Academy of Sciences, 1204(1), 127-138.
Amdam, G.V. (2011). Social context, stress, and plasticity of aging. Aging Cell, 10(1), 18-27.
Backes, C., et al. (2007). GeneTrail—advanced gene set enrichment analysis. Nucleic Acids Research, 35(Suppl. 2), W186-W192.
Bagot, R.C., et al. (2012). Variations in postnatal maternal care and the epigenetic regulation of metabotropic glutamate receptor 1 expression and hippocampal function in the rat. Proceedings of the National Academy of Sciences of the United States of America, 109(Suppl.), 17200-17207.
Barreiro, L.B., et al. (2010). Functional comparison of innate immune signaling pathways in primates. PLOS Genetics, 6(12), e1001249.
Bauer, M.E. (2008). Chronic stress and immunosenescence: A review. Neuroimmunomodulation, 15(4-6), 241-250.
Berchtold, N.C., et al. (2008). Gene expression changes in the course of normal brain aging are sexually dimorphic. Proceedings of the National Academy of Sciences of the United States of America, 105(40), 15605-15610.
Berkman, L.F., and Syme, S.L. (1979). Social networks, host resistance, and mortality: A nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 109(2), 186-204.
Boonekamp, J.J., et al. (2013). Telomere length behaves as biomarker of somatic redundancy rather than biological age. Aging Cell, 12(2), 330-332.
Borghol, N., et al. (2012). Associations with early-life socio-economic position in adult DNA methylation. International Journal of Epidemiology, 41, 62-74.
Boyce, W.T., and Ellis, B.J. (2005). Biological sensitivity to context: I. An evolutionary-developmental theory of the origins and functions of stress reactivity. Development and Psychopathology, 17(2), 271-301.
Boyce, W.T., et al. (2012). Toward a new biology of social adversity. Proceedings of the National Academy of Sciences of the United States of America, 109(Suppl.), 17143-17148.
Bronikowski, A.M., et al. (2002). The aging baboon: Comparative demography in a nonhuman primate. Proceedings of the National Academy of Sciences of the United States of America, 99(14), 9591-9595.
Bronikowski, A.M., et al. (2011). Aging in the natural world: Comparative data reveal similar mortality patterns across primates. Science, 331(6022), 1325-1328.
Bronikowski, A.M., and Promislow, D.E. (2005). Testing evolutionary theories of aging in wild populations. Trends in Ecology & Evolution, 20(6), 271-273.
Cao, J.N., et al. (2010). Age-related alterations of gene expression patterns in human CD8+ T cells. Aging Cell, 9(1), 19-31.
Carroll, S.B. (2005). Endless Forms Most Beautiful: The New Science of Evo Devo and the Making of the Animal Kingdom. New York: W.W. Norton & Company.
Champagne, F.A., et al. (2003). Natural variations in maternal care are associated with estrogen receptor alpha expression and estrogen sensitivity in the medial preoptic area. Endocrinology, 144(11), 4720-4724.
Champagne, F.A., et al. (2006). Maternal care associated with methylation of the estrogen receptor-alpha1b promoter and estrogen receptor-alpha expression in the medial preoptic area of female offspring. Endocrinology, 147(6), 2909-2915.
Chen, E., et al. (2009). Genome-wide transcriptional profiling linked to social class in asthma. Thorax, 64(1), 38-43.
Cherkas, L.F., et al. (2006). The effects of social status on biological aging as measured by white-blood-cell telomere length. Aging Cell, 5, 361-365.
Christensen, B.C., et al. (2009). Aging and environmental exposures alter tissue-specific DNA methylation dependent upon CpG island context. PLOS Genetics, 5, e1000602.
Cole, S.W., et al. (2007). Social regulation of gene expression in human leukocytes. Genome Biology, 8, R189.
Cole, S.W., et al. (2011). Transcript origin analysis identifies antigen-presenting cells as primary targets of socially regulated gene expression in leukocytes. Proceedings of the National Academy of Sciences of the United States of America, 108(7), 3080-3085.
Cole, S.W., et al. (2012). Transcriptional modulation of the developing immune system by early life social adversity. Proceedings of the National Academy of Sciences of the United States of America, 109(50), 20578-20583.
Colman, R.J., et al. (2009). Caloric restriction delays disease onset and mortality in rhesus monkeys. Science, 325(5937), 201-204.
de Magalhães, J.P., et al. (2005). The influence of genes on the aging process of mice a statistical assessment of the genetics of aging. Genetics, 169(1), 265-274.
de Magalhães, J.P., and Church, G.M. (2007). Analyses of human-chimpanzee orthologous gene pairs to explore evolutionary hypotheses of aging. Mechanisms of Ageing and Development, 128(5), 355-364.
de Magalhães, J.P., et al. (2009). Meta-analysis of age-related gene expression profiles identifies common signatures of aging. Bioinformatics, 25(7), 875-881.
de Magalhães, J.P., et al. (2012). Genome-environment interactions that modulate aging: Powerful targets for drug discovery. Pharmacological Reviews, 64(1), 88-101.
Dowd, J.B., and Goldman, N. (2006). Do biomarkers of stress mediate the relation between socioeconomic status and health? Journal of Epidemiology and Community Health, 60(7), 633-639.
Dunham, I., et al. (2012). An integrated encyclopedia of DNA elements in the human genome. Nature, 489, 57-74.
Elango, N., et al. (2009). DNA methylation is widespread and associated with differential gene expression in castes of the honeybee, Apis mellifera. Proceedings of the National Academy of Sciences of the United States of America, 106, 11206-11211.
Ellis, B.J., et al. (2011). Differential susceptibility to the environment: An evolutionary—neurodevelopmental theory. Development and Psychopathology, 23(1), 7-28.
Epel, E.S., et al. (2004). Accelerated telomere shortening in response to life stress. Proceedings of the National Academy of Sciences of the United States of America, 101, 17312-17315.
Evans, J.D., and Wheeler, D.E. (1999). Differential gene expression between developing queens and workers in the honey bee, Apis mellifera. Proceedings of the National Academy of Sciences of the United States of America, 96(10), 5575-5580.
Finch, C.E., and R.E. Tanzi (1997). Genetics of aging. Science, 278(5337), 407-411.
Fok, W.C., et al. (2014). Mice fed rapamycin have an increase in lifespan associated with major changes in the liver transcriptome. PLOS ONE, 9(1), e83988.
Fraga, M.F. (2009). Genetic and epigenetic regulation of aging. Current Opinion in Immunology, 21(4), 446-453.
Fraser, H.B., et al. (2005). Aging and gene expression in the primate brain. PLOS Biology, 3(9), e274.
Gardner, M.P., et al. (2006). Extraordinary plasticity in aging in Strongyloides ratti implies a gene-regulatory mechanism of lifespan evolution. Aging Cell, 5(4), 315-323.
Göring, H.H., et al. (2007). Discovery of expression QTLs using large-scale transcriptional profiling in human lymphocytes. Nature Genetics, 39(10), 1208-1216.
Greer, E.L., and Brunet, A. (2009). Different dietary restriction regimens extend lifespan by both independent and overlapping genetic pathways in C. elegans. Aging Cell, 8(2), 113-127.
Grotewiel, M.S., et al. (2005). Functional senescence in Drosophila melanogaster. Ageing Research Reviews, 4(3), 372-397.
Han, Y., et al. (2012). Stress-associated H3K4 methylation accumulates during postnatal development and aging of rhesus macaque brain. Aging Cell, 11(6), 1055-1064.
Hansen, K.D., et al. (2012). BSmooth: From whole genome bisulfite sequencing reads to differentially methylated regions. Genome Biology, 13(10), R83.
Hawkins, R.D., et al. (2010). Next-generation genomics: An integrative approach. Nature Reviews Genetics, 11(7), 476-486.
Holt-Lunstad, J., et al. (2010). Social relationships and mortality risk: A meta-analytic review. PLOS Medicine, 7, e1000316.
Hong, M.-G., et al. (2008). Transcriptome-wide assessment of human brain and lymphocyte senescence. PLOS ONE, 3(8), e3024.
House, J.S., et al. (1988). Social relationships and health. Science, 241, 540-545.
Huang, D.W., et al. (2008). Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nature Protocols, 4(1), 44-57.
Hyman, S.E. (2009). How adversity gets under the skin. Nature Neuroscience, 12, 241-243.
Jacobson, J., et al. (2010). Biomarkers of aging in Drosophila. Aging Cell, 9(4), 466-477.
Jones, O.R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505(7482), 169-173.
Juster, R.-P., et al. (2010). Allostatic load biomarkers of chronic stress and impact on health and cognition. Neuroscience & Biobehavioral Reviews, 35(1), 2-16.
Karssen, A., et al. (2007). Stress-induced changes in primate prefrontal profiles of gene expression. Molecular Psychiatry, 12(12), 1089-1102.
Kelly, M.A., et al. (2013). Effect of temperature on the rate of ageing: An experimental study of the blowfly Calliphora stygia. PLOS ONE, 8(9), e73781.
Kennerly, E., et al. (2008). A gene expression signature of confinement in peripheral blood of red wolves (Canis rufus). Molecular Ecology, 17(11), 2782-2791.
Kenyon, C., et al. (1993). A C. elegans mutant that lives twice as long as wild type. Nature, 366(6454), 461-464.
Kenyon, C.J. (2010). The genetics of ageing. Nature, 464(7288), 504-512.
Kittleson, M.M., et al. (2006). Association of childhood socioeconomic status with subsequent coronary heart disease in physicians. Archives of Internal Medicine, 166(21), 2356-2361.
Kucharski, R., et al. (2008). Nutritional control of reproductive status in honeybees via DNA methylation. Science, 319(5871), 1827-1830.
Kudielka, B.M., and Kirschbaum, C. (2005). Sex differences in HPA axis responses to stress: A review. Biological Psychology, 69(1), 113-132.
Lam, L.L., et al. (2012). Factors underlying variable DNA methylation in a human community cohort. Proceedings of the National Academy of Sciences of the United States of America, 109(Suppl.), 17253-17260.
Larbi, A., et al. (2008). Aging of the immune system as a prognostic factor for human longevity. Physiology, 23(2), 64-74.
Levinthal, C., et al. (1962). Messenger RNA turnover and protein synthesis in B. subtilis inhibited by actinomycin D. Proceedings of the National Academy of Sciences of the United States of America, 48(9), 1631.
López-Otín, C., et al. (2013). The hallmarks of aging. Cell, 153(6), 1194-1217.
Lu, T., et al. (2004). Gene regulation and DNA damage in the ageing human brain. Nature, 429(6994), 883-891.
Luca, F., et al. (2013). Genetic, functional and molecular features of glucocorticoid receptor binding. PLOS ONE, 8(4), e61654.
Lund, J., et al. (2002). Transcriptional profile of aging in C. elegans. Current Biology, 12(18), 1566-1573.
Lutgendorf, S.K., et al. (2009). Depression, social support, and beta-adrenergic transcription control in human ovarian cancer. Brain, Behavior, and Immunity, 23(2), 176-183.
Mair, W., et al. (2003). Demography of dietary restriction and death in Drosophila. Science 301(5640), 1731-1733.
Marioni, J.C., et al. (2008). RNA-seq: An assessment of technical reproducibility and comparison with gene expression arrays. Genome Research, 18(9), 1509-1517.
Marmot, M.G. (2006). Status syndrome: A challenge to medicine. Journal of the American Medical Association, 295(11), 1304-1307.
Mattison, J.A., et al. (2012). Impact of caloric restriction on health and survival in rhesus monkeys from the NIA study. Nature, 489(7415), 318-321.
McElwee, J.J., et al. (2007). Evolutionary conservation of regulated longevity assurance mechanisms. Genome Biology, 8(7), R132.
McGowan, P.O., et al. (2009). Epigenetic regulation of the glucocorticoid receptor in human brain associates with childhood abuse. Nature Neuroscience, 12, 342-348.
Metzker, M.L. (2010). Sequencing technologies—the next generation. Nature Reviews Genetics, 11(1), 31-46.
Miller, G.E., et al. (2008). A functional genomic fingerprint of chronic stress in humans: Blunted glucocorticoid and increased NF-kappaB signaling. Biological Psychiatry, 64, 266-272.
Miller, G.E., et al. (2009). Low early-life social class leaves a biological residue manifested by decreased glucocorticoid and increased proinflammatory signaling. Proceedings of the National Academy of Sciences of the United States of America, 106, 14716-14721.
Mortazavi, A., et al. (2008). Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nature Methods, 5(7), 621-628.
Murgatroyd, C., et al. (2009). Dynamic DNA methylation programs persistent adverse effects of early-life stress. Nature Neuroscience, 12, 1559-1566.
Murgatroyd, C., et al. (2010). The Janus face of DNA methylation in aging. Aging, 2(2), 107.
Murphy, C.T., et al. (2003). Genes that act downstream of DAF-16 to influence the lifespan of Caenorhabditis elegans. Nature, 424(6946), 277-283.
Murphy, M.L.M., et al. (2013). Targeted rejection triggers differential pro- and anti-inflammatory gene expression in adolescents as a function of social status. Clinical Psychological Science, 1, 30-40.
Partridge, L., and Gems, D. (2002). Mechanisms of aging: Public or private? Nature Reviews Genetics, 3(3), 165-175.
Partridge, L., and Gems, D. (2006). Beyond the evolutionary theory of ageing, from functional genomics to evo-gero. Trends in Ecology & Evolution, 21(6), 334-340.
Pletcher, S.D., et al. (2002). Genome-wide transcript profiles in aging and calorically restricted Drosophila melanogaster. Current Biology, 12(9), 712-723.
Provencal, N., et al. (2012). The signature of maternal rearing in the methylome in rhesus macaque prefrontal cortex and T cells. The Journal of Neuroscience, 32(44), 15626-15642.
Rea, S.L., et al. (2007). Relationship between mitochondrial electron transport chain dysfunction, development, and life extension in Caenorhabditis elegans. PLOS Biology, 5(10), e259.
Reddy, T.E., et al. (2009). Genomic determination of the glucocorticoid response reveals unexpected mechanisms of gene regulation. Genome Research, 19(12), 2163-2171.
Robinson, G.E., et al. (2008). Genes and social behavior. Science, 322(5903), 896-900.
Robinson, G.E., et al. (2005). Sociogenomics: Social life in molecular terms. Nature Reviews Genetics, 6(4), 257-270.
Rohleder, N., et al. (2001). Sex differences in glucocorticoid sensitivity of proinflammatory cytokine production after psychosocial stress. Psychosomatic Medicine, 63(6), 966-972.
Roth, T.L., et al. (2009). Lasting epigenetic influence of early-life adversity on the BDNF gene. Biological Psychiatry, 65(9), 760-769.
Sapolsky, R.M. (2004). Social status and health in humans and other animals. Annual Review of Anthropology, 33, 393-418.
Sapolsky, R.M. (2005). The influence of social hierarchy on primate health. Science, 308, 648-652.
Sarup, P., et al. (2011). Flies selected for longevity retain a young gene expression profile. Age, 33(1), 69-80.
Slavich, G.M., and S.W. Cole (2013). The emerging field of human social genomics. Clinical Psychological Science, 1(3), 331-348.
Snyder-Mackler, N., et al. (2014). Shared signatures of social stress and aging in peripheral blood mononuclear cell gene expression profiles. Aging Cell. doi: 10.1111/acel.12239. Available: www.ncbi.nlm.nih.gov/pubmed/24956926 [July 2014].
Somel, M., et al. (2010). MicroRNA, mRNA, and protein expression link development and aging in human and macaque brain. Genome Research, 20(9), 1207-1218.
Subramanian, A., et al. (2005). Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proceedings of the National Academy of Sciences of the United States of America, 102(43), 15545-15550.
Taylor, S.E., et al. (1997). Health psychology: What is an unhealthy environment and how does it get under the skin? Annual Review of Psychology, 48, 411-447.
Tilstra, J.S., et al. (2012). NF-kB inhibition delays DNA damage-induced senescence and aging in mice. The Journal of Clinical Investigation, 122(7), 2601.
Tsankova, N.M., et al. (2006). Sustained hippocampal chromatin regulation in a mouse model of depression and antidepressant action. Nature Neuroscience, 9(4), 519-525.
Tung, J., and Gilad, Y. (2013). Social environmental effects on gene regulation. Cellular and Molecular Life Sciences, 70, 4323-4339.
Tung, J., et al. (2012). Social environment is associated with gene regulatory variation in the rhesus macaque immune system. Proceedings of the National Academy of Sciences of the United States of America, 109, 6490-6495.
Vijg, J., and Suh, Y. (2003). Functional genomics of ageing. Mechanisms of Ageing and Development, 124(1), 3-8.
Wang, J., et al. (2012). Sequence features and chromatin structure around the genomic regions bound by 119 human transcription factors. Genome Research, 22(9), 1798-1812.
Wang, Z., et al. (2009). RNA-Seq: A revolutionary tool for transcriptomics. Nature Reviews Genetics, 10(1), 57-63.
Weaver, I.C.G., et al. (2004). Epigenetic programming by maternal behavior. Nature Neuroscience, 7, 847-854.
Wu, D., et al. (2009). Multiple mild heat-shocks decrease the Gompertz component of mortality in Caenorhabditis elegans. Experimental Gerontology, 44(9), 607-612.
Yen, K., and Mobbs, C.V. (2010). Evidence for only two independent pathways for decreasing senescence in Caenorhabditis elegans. Age, 32(1), 39-49.
Yuan, Y., et al. (2012). Accelerated aging-related transcriptome changes in the female prefrontal cortex. Aging Cell, 11(5), 894-901.