The impact of life stress on hallmarks of aging and accelerated senescence: connections in sickness and in health

Abstract

Chronic stress is a risk factor for numerous aging-related diseases and has been shown to shorten lifespan in humans and other social mammals. Yet, how life stress causes such a vast range of diseases is still largely unclear. In recent years, the impact of stress on health and aging has been increasingly associated with the dysregulation of the so-called hallmarks of aging. These are basic biological mechanisms that influence intrinsic cellular functions and whose alteration can lead to accelerated aging. Here, we review correlational and experimental literature (primarily focusing on evidence from humans and murine models) on the contribution of life stress – particularly stress derived from adverse social environments – to trigger hallmarks of aging including cellular senescence, sterile inflammation, telomere length, production of reactive oxygen species, DNA damage, and epigenetic changes. We also evaluate the validity of stress-induced senescence and accelerated aging as an etiopathological proposition. Finally, we highlight current gaps of knowledge and future directions for the field, and discuss perspectives for translational geroscience.

Graphical abstract

INTRODUCTION

Life stress – defined as the occurrence of unpredictable, uncontrollable and potentially/actually life-threatening conditions (Koolhaas et al., 2011) – is a major risk factor for a multitude of aging-related diseases, as shown by epidemiological, clinical, and preclinical studies. In particular, chronic stress confers increased risk and often earlier onset of aging-related diseases, which severely degrade quality of life and are the primary causes of death in older adults. For example, life stress increases risk for age-related diseases such as obesity, type 2 diabetes, cardiovascular disease, Alzheimer’s disease, vascular dementia, Parkinson’s disease, osteoarthritis and many others (Bonanni et al., 2018; Gerritsen et al., 2017; Hemmerle et al., 2012; Kivimäki et al., 2022; Kivimäki and Steptoe, 2018; McEwen, 1998; Mejfa et al., 2003; Snyder-Mackler et al., 2020).

In spite of the decades old proposition that chronic stress accelerates aging via global dysregulation of homeostatic mechanisms (Sapolsky et al., 1986; Selye, 1970), how life stress causes such a vast range of diseases is still largely unclear. The advent of geroscience provides a conceptual framework to interpret the impact of life stress on health and aging by addressing mechanisms that regulate the fundamental biological processes affecting aging at the cellular, organ and systemic level (Duque et al., 2023; Epel, 2020; Kennedy et al., 2014; Kivimäki et al., 2022; López-Otín et al., 2023; Lyons and Bartolomucci, 2020; Polsky et al., 2022; Sierra and Kohanski, 2023; Snyder-Mackler et al., 2020). For example, graying of the hair, a classical visual biomarker of aging and premature senescence, has recently been shown to be induced by stress (Rosenberg et al., 2021), with a potential role played by stress-induced hyperactivation of the sympathetic nervous system (SNS) (Zhang et al., 2020). Similarly, bone loss and osteoporosis increase with aging, and chronic social stress has been shown to affect cartilage-to-bone transition during bone growth and repair upon fracture in mice, via a mechanism requiring SNS activation (Tschaffon-Müller et al., 2023).

Here, we review the available correlational and experimental literature – primarily focusing our analysis on human data and preclinical murine models – on the contribution of stress to trigger select hallmarks of aging (Kennedy et al., 2014; López-Otín et al., 2023, 2013) including: i) cellular senescence and sterile inflammation; ii) telomere length; iii) production of reactive oxygen species; iv) DNA damage; v) epigenetic changes. Several excellent reviews discussed the role of some of the hallmarks of aging as potential biomarkers, or causal factors, of the link between life stress, aging-associated diseases, and accelerated biological aging (e.g. (Blackburn et al., 2015; Kivimäki et al., 2022; Palma-Gudiel et al., 2020; Polsky et al., 2022; Raffington and Belsky, 2022), but to the best of our knowledge, no previous review integrated animal models and human data in a translational geroscience perspective, as presented in this paper. Additionally, several important research papers have been published in the ~12 months preceding this publication, which have not been reviewed extensively elsewhere (e.g (Chiou et al., 2022; Dion et al., 2022; Kong et al., 2023; Obeng-Gyasi et al., 2023; Poganik et al., 2022; Razzoli et al., 2023; Rentscher et al., 2022; Tschaffon-Müller et al., 2023). The aims of this review are, i) to briefly summarize previous knowledge on the hallmarks of aging and their involvement in life stress (pointing readers to previous reviews, whenever possible), ii) to highlight recent findings in human studies and animal models of social determinants of health and aging (SDoHA), that enable the transition from correlational studies to causal inference, and finally iii) integrate those concepts within the framework of geroscience highlighting gaps of knowledge and future perspectives.

LIFE STRESS AND ACCELLERATION OF BIOLOGICAL AGING

A certain degree of physical, emotional, and social challenge is a constant feature of life, and all organisms must be capable of managing and adapting to perturbations to homeostasis. However, sometimes life challenges impose physiological demands outside an individual’s adaptive capacity. These challenges can be defined as stressors (Koolhaas et al., 2011). Over the years, the field has revised its nomenclature multiple times and there is not currently a universally accepted definition of stress (Aschbacher et al., 2013; Koolhaas et al., 2011; McEwen, 1998; Nederhof and Schmidt, 2012; Sapolsky et al., 2000). Here we will refer to the “stress revised” terminology, defining life stress as circumstances characterized by perception of/exposure to an uncontrollable and unpredictable hazard that can pose at least a perceived potential life threat and requires adaptations outside of an individual’s regulatory range (Koolhaas et al., 2011).

Aside from its definition, it is accepted that the physiological stress-response system includes activation of several neural and neuroendocrine axes such as the SNS, the sympatho-adrenal medullary (SAM) system and the hypothalamic pituitary adrenal (HPA) axis (Charmandari et al., 2005; Sapolsky et al., 2000; Ulrich-Lai and Herman, 2009). In response to a stressor, the SNS/SAM are rapidly activated mediating the classical “fight or flight response” characterized by increased heart rate, respiratory rate, glycolysis, glycogenolysis and lipolysis to sustain an acute metabolic demand to the body. Unlike the fast responding and rapidly deactivating responses mediated by the SNS, the HPA axis operates on a slower timescale. It’s activation originates from neurons in the paraventricular nucleus of the hypothalamus, that in response to perceived threat, release corticotropin releasing hormone (CRH) into the hypophysial portal system. This leads to release of pituitary adrenocorticotropic hormone (ACTH) into general circulation. In turn, this results in glucocorticoid (GC) secretion from the adrenal cortex. GCs activate several physiological responses while also exerting a negative feedback on the HPA axis. However, in cases of sustained stress, or during aging, GC hypersecretion can persist (Sapolsky et al., 1984; Spiga et al., 2014). The negative feedback system becomes impaired with age due in large part to downregulation of glucocorticoid receptors (GR) in the hippocampus and other critical nodes of the HPA axis (Polsky et al., 2022; Sapolsky et al., 1986). While this defines the main stress-related neuro-endocrine axes, additional endocrine mediators such as thyroid and growth hormone axes (Charmandari et al., 2005; Sapolsky et al., 2000; Tsigos and Chrousos, 2002), as well as adipose-derived IL-6 (Qing et al., 2020) and osteoclast-derived osteocalcin (Berger et al., 2019) are activated acutely and/or chronically during stress (Figure 1), which are also implicated in aging (Bonewald, 2019; Cappola et al., 2023; van den Beld et al., 2018).

Figure 1. Summary of the major stress-related axes, and stress-induced changes to select hallmarks of aging.

The major stress-related pathways include the hormones of the hypothalamus-pituitary-adrenocortical (HPA) axis: corticotrophin releasing factor (CRH), adrenocorticotropic hormone (ACTH), and cortisol (or corticosterone in rodents). The sympatho-adreno-medullary (SAM) axis secretes catecholamines, granins, and other hormones and neuropeptides. The neurotransmitters secreted by the sympathetic nervous system (SNS), include noradrenaline, ATP, neuropeptide Y (NPY), and other factors. The parasympathetic nervous system (PNS) releases neurotransmitters such as acetylcholine and other factors. Other neuroendocrine axes include the growth hormone (GH) axis [growth hormone releasing hormone (GHRH), GH, and insulin-like growth factor 1 (IGF1)], the thyroid axis [thyrotropin releasing hormone (TRH), thyroid stimulating hormone (TSH), triiodothyronine (T3), and thyroxine (T4)]. The adipose tissue (brown or white adipocytes) secretes factors such as interleukin-6 (IL-6), leptin, and others. Osteoclasts have been shown to secrete osteocalcin in response to stress. Evidence discussed in this review demonstrate that life stress promotes hallmarks of aging in several organs and systems. These include blood and other body fluids, circulating immune cells [peripheral blood mononuclear cells (PBMC)], central nervous system (CNS), thymus, lung, liver, gastric, muscle, and bone tissues. Stress-induced changes to hallmarks of aging have been primarily studied in animal models; studies in humans have been limited to PBMC and body fluids and are presented in italics.

Among the life stressors affecting body function and aging, stressors of social nature, aka “social stressors”, have singularly potent effects (Bartolomucci, 2007; Koolhaas et al., 2011; Sapolsky, 2005; Snyder-Mackler et al., 2020). Indeed, SDoHA - including adverse childhood experiences (ACE), low socioeconomic status (SES), loneliness, and caregiver stress - are increasingly regarded as modulators of the stress response, accelerators of biological age, and risk factors for aging-associated diseases (Kivimäki et al., 2022; Noren Hooten et al., 2022; Snyder-Mackler et al., 2020; Wang et al., 2023). For example, ACE cause particularly severe harm to both healthspan – the number of years lived without functional impairment – and lifespan (Kivimäki et al., 2022; Snyder-Mackler et al., 2020; Tung et al., 2016). Exposure to an increasing number of ACE elevates risk for chronic diseases in adulthood, including ischemic heart disease, cancer, and chronic lung disease (Kivimäki et al., 2022; Kivimäki and Steptoe, 2018; Lupien et al., 2009). Like ACE, low SES confers an immense burden of social stress and is associated with elevated mortality risk from nearly all causes (Elo, 2009; Marmot et al., 1991). The factors involved can include financial insecurity, uncertainty about future stability, and feelings of exclusion or marginalization due to occupational or economic status, in addition to limited or absent appropriate nutritional, educational and medical resources (Adler et al., 1994; Kivimäki et al., 2022; Marmot et al., 1991; Snyder-Mackler et al., 2020; Stringhini et al., 2017). Interestingly, the adverse effect of ACE on health and aging was recapitulated in some [baboons (Tung et al., 2016)] but not all [gorilla (Morrison et al., 2023)], non-human primates, suggesting that species-specific behavioral and socioecological factors can inform on biological resilience to ACE in humans.

While a number of factors contribute to the myriad negative health outcomes associated with low SES, material deprivation does not entirely explain this phenomenon (Richterman et al., 2023). Rather, it is widely hypothesized that chronic social stress and its physiological consequences at least partially underlie the shorter lifespan and numerous poor disease outcomes associated with low SES. Indeed, low SES is associated with higher basal levels of catecholamines and cortisol, as well as higher self-reported levels of stress (Cohen et al., 2006; Szanton et al., 2005). Supporting psychosocial stress as a principal intermediary of SES and poor health outcomes is a study showing that low SES individuals who still experience a high sense of control have health metrics comparable with individuals in higher income groups (Lachman and Weaver, 1998).

One cohort study followed >3,000 participants aged 6–18 years at baseline for over 30 years via repeated biomedical examinations (Kivimäki et al., 2018). The findings suggested that experiencing lower SES and its associated stressors accelerated age-related metabolic changes. At baseline, groups with high vs low SES stress did not differ in any metabolic traits, but by early adulthood the low SES group had notably increased blood lipid levels, decreased insulin sensitivity, and increased compensatory insulin secretion that was still insufficient to normalize circulating glucose concentrations. By middle age, low SES individuals were more likely to be obese, hypertensive, have a fatty liver, and have diabetes compared with those who consistently lived in high SES neighbourhoods (Kivimäki et al., 2018). Another study found that SES experiences from each life course period (childhood, early adulthood, and late adulthood) additively combine to predict the number of years an individual can expect to live with or without dementia (Cha et al., 2021). Childhood adversities alone have a profound effect on the likelihood of developing dementia. The risk of developing dementia for individuals experiencing two disadvantages and three disadvantages were found to respectively be 56%, and 87% higher than for those with no socioeconomic disadvantages during childhood. Low SES status during all three life periods shortened life expectancy for both men and women by 8.66 and 6.89 years, respectively. 18% of those years for men, or 17% for women were predicted to be lived with dementia, compared with 3% for men or 4% for women who experienced high SES throughout all life stages (Cha et al., 2021).

Although the relationship between low SES and poorer health/lifespan has long been documented, the biological mechanisms underlying this effect have been difficult to study (Kivimäki et al., 2022). Some important recent advancements have been made, e.g. the Environmental Risk Longitudinal Twin Study found that children raised in more socioeconomically disadvantaged neighbourhoods develop epigenetic DNA methylation patterns distinct from their more advantaged peers (Reuben et al., 2020). A Mendelian randomization study investigated causal associations of 19 modifiable socioeconomic, lifestyle, and cardiometabolic factors with DNA-based measures of biological age (Kong et al., 2023). In addition to known risk factors such as smoking, high education, and household income, operational measures of high SES, were found to be among the strongest protective factor for the DNA age of an individual (Kong et al., 2023).

Although Mendelian randomization studies are increasingly used to attempt to answer cause-effect relationships in human studies (Kivimäki et al., 2022), a more precise understanding of causal physiological underpinnings of the health risks conferred by low social status requires the use of animal models where randomization experiments are possible. This approach is supported by the existence of social gradients in health in virtually all social mammals (Snyder-Mackler et al., 2020). Consistent with human studies, in experimental animal models of social determinants of health, social stressors (e.g. chronic subordination stress) are particularly potent activators of stress responses compared with other types, such as physical (e.g. cold water stress, foot shock) or psychological (e.g. restraint stress) (Bartolomucci, 2007; Bartolomucci et al., 2009). In baboons, social rank as well as the strength/stability of an individual’s social bonds predict transcriptional profiling related to biological aging and survival (Chiou et al., 2022; Tung et al., 2016; Zipple et al., 2019) and experimental manipulation of social rank causes a lasting effect on immune functions (Snyder-Mackler et al., 2016). Similarly, subordinate status within a group of mice has been studied as a model for low SES in humans. Subordinate mice show persistent immunoendocrine impairments, reduced lifespan, elevation of cellular senescence markers and development of early-stage atherosclerotic lesions (W. Lee et al., 2022; Razzoli et al., 2018). Furthermore, new data demonstrate changes in methylation patterns linked to mouse genetic background and social rank, adding further credibility to social subordination as a relevant model of SDoHA (Razzoli et al., 2023).

EFFECT OF LIFE STRESS ON SELECTED HALLMARKS OF AGING

Accelerated biological aging is considered to reflect the cumulative progressive dysfunction and cellular damage compounded during lifespan. A set of cellular/physiological processes has been identified as age-promoting factors, and are collectively referred to as pillars/hallmarks of aging, including cellular senescence, telomere length, ROS and oxidative stress, genetic/epigenetic changes (Kennedy et al., 2014; López-Otín et al., 2023, 2013). While for clarity’s sake, this review describes some of these hallmarks individually in relation to stress exposure, it is crucial to note that a central feature of these pillars/hallmarks is their connectedness and interplay (Epel, 2020; Kennedy et al., 2014; López-Otín et al., 2023). This interconnection posits a conceptual challenge in that it is difficult to pinpoint individual physiological mechanisms responsible for age-associated damage. Many homeostatic and pathological processes occur concomitantly and/or consequentially during the life course, differentially affecting the pace of aging. Here, we will briefly review selected hallmarks/pillars of aging such as: i) telomere length, ii) cellular senescence and systemic inflammation, iii) production of reactive oxygen species, iv) DNA damage and v) epigenetic changes, and discuss evidence of their link to life stress (Figure 1). Because most/all of the hallmarks of aging are interconnected and multiple markers are often measured in a single research paper, the narrative will often create connection and overlaps while attempting to keep the sections separate. This review will primarily focus on in vivo studies where life stress can be investigated. However, cellular models exposed acutely/chronically to stress-related hormones such as glucocorticoids, offer promising mechanistic insights and will also briefly reviewed here. For examples, a recent study showed that chronic glucocorticoid exposure increases cellular energy expenditure (caused by a metabolic shift from glycolysis to mitochondrial oxidative phosphorylation), induced mitochondrial (mt)DNA instability, modulated senescence associated secretory phenotype (SASP), and accelerates cellular aging assessed using DNA methylation and telomere shortening (Bobba-Alves et al., 2023).

EFFECTS OF LIFE STRESS ON TELOMERES

Telomeres are repeated TTAGGG DNA sequences that cap the end of chromosomes, are subject to progressive reduction during replication and aging, and whose reduction is counterbalanced by the activity of telomerase (Blackburn et al., 2015; de Lange, 2009). Before telomeres are completely lost, they reach a critical length, termed the Hayflick limit, which triggers the DNA damage response (DDR) and activation of the tumor suppressor p53 resulting in either apoptosis, or exit from the cell cycle (Saretzki et al., 1999). In a given cell, the different chromosomes may have a range of telomere lengths; however, the stability and viability of the cell is defined by the shortest telomere present (Hemann et al., 2001). Numerous environmental factors, including life stress as discussed below, can affect telomere length maintenance (See also (Blackburn et al., 2015; Polsky et al., 2022) for additional review) (Table 1).

Table 1.

Effect of chronic life stress or neuroendcirne stress mediators on selected hallmarks of aging: effect on telomeres

Finding	Species	Population	Stress model/proxy	Tissue	Reference
Shorter telomeres, lower telomerase activity, greater oxidative damage	Human	Pre-menopausal mothers	Reported stress levels	PBMCs	Epel et al., 2004
Shorter telomeres, higher cytokine production by PBMCs	Human	Men and women, avg age 65+/− 5 years	Caregivers of AD patients	PBMCs	Damjanovic et al., 2007
Shorter telomeres associated with stress and urinary catecholamine level	Human	Women with a sister with breast cancer, but not cancer themselves	Reported stress levels	Leukocytes	Parks et al., 2009
Shorter telomeres	Human	Female twins, avg age 46	Low SES	White blood cells	Cherkas et al., 2006
No difference in telomere length	Human	Men and women, avg age 50+/− 5 years	Low SES	PBMCs	Adams et al., 2007
Shorter telomeres	Human	Men and women over 60; avg age 70	Experience of childhood adverse events	Leukocytes	Schaakxs et al., 2016
Lower baseline telomerase activity; equivalent telomerase activity following exposure to acute stressor	Human	Women age 51–75; avg age 62	Dementia caregivers reporting high stress levels; Trier Social Stress Test	PBMCs	Epel et al., 2010
decreased telomere	Mice	wild type young adult	chronic mild stress	liver	Xie et al. 2019
No change in telomere length	Mice	wild type 17 month old	Lifelong chronic psychosocial stress	liver, spleen, lung	Razzoli et al., 2018
Lower telomere length	Human	Sedentary postmenopausal women	Reported stress levels	Whole blood	Puterman et al., 2010

The first study to directly link life stress with one of the markers of biological aging was published in 2004 by Epel and coworkers (Epel et al., 2004). Participants were premenopausal mothers to either a healthy child or a chronically ill child. Telomere length and telomerase activity were measured in peripheral blood mononuclear cells (PBMCs) . Within the group of mothers with chronically ill children, there was a negative correlation between duration of caregiving and telomere length, independent of age. However, rather than the health status of the child, the degree of perceived stress reported by the mothers was a better predictor of telomere length and telomerase activity. Although the average age of the two groups was similar, telomeres in the higher stress group were shorter by an average of 550-bp, suggesting a difference of 9–17-years difference in biological age. The high stress group also had lower telomerase activity and higher levels of higher oxidative stress. This negative correlation between life stress level and telomere length has been replicated in many (Damjanovic et al., 2007; Parks et al., 2009) but not all studies (Rentscher et al., 2019). It has also been shown that urinary catecholamine content, indicative of chronic activation of one of the major stress neuroendocrine axes, is associated with shorter telomeres suggesting a possible causative link (Parks et al., 2009). Similar to life stress, low SES has been associated with lower telomere length in some (Cherkas et al., 2006) but not all studies (Adams et al., 2007). These differing findings may be influenced by the observation that in older adults, stress-associated differences in telomere length are more difficult to parse. In older age, other factors such as disease burden and disease survivorship account for greater variability (Schaakxs et al., 2016). Despite this, the negative effect of adverse childhood events can still be discerned in the telomere length of older adults (Schaakxs et al., 2016).

In a mouse model of lifelong chronic subordination stress, absolute telomere length in liver, lung and spleen of 17 month-old mice was not altered in spite of increased organ damage, increased markers of cellular senescence and shorter lifespan (Razzoli et al., 2018). Similarly, early life stress had no consequences on telomere length in hippocampi when mice were tested at 20 months of age (Kotah et al., 2021). Unlike early life stress and subordination, chronic mild stress in young adult mice led to a shortening of telomeres in the liver, in presence of increased oxidative damage (Xie et al., 2019). The tissue specificity of telomere length shortening due to chronic stress was also highlighted in a study of mice exposed to chronic immobilization stress, showing tissue-specific acceleration (Baek et al., 2019).

A dynamic relationship between psychological stress and telomerase activity has also emerged. High stress caregivers to a family member with dementia were found to have lower baseline telomerase activity; however, when exposed to an acute stressor they displayed increased telomerase activity equivalent to that of age-matched low stress controls (Epel et al., 2010). The increases in telomerase activity were associated with the magnitude of cortisol response to the stressor. However, another study found that caregivers to a dementia patient had higher basal telomerase activity, despite shorter telomeres (Damjanovic et al., 2007). Follow-up studies have looked for possible interventions for stress-induced telomere erosion. It was found that regular physical exercise can buffer the effect of stress-induced telomere shortening in post-menopausal women (Puterman et al., 2010).

EFFECTS OF LIFE STRESS ON CELLULAR SENESCENCE, SASP FACTORS AND INFLAMMATION

Cellular senescence is a complex phenomenon with yet undefined molecular definition (Gorgoulis et al., 2019; P. J. Lee et al., 2022), and implications for prevention of uncontrolled proliferation in cancer development, normal wound repair and tissue regeneration (Campisi and d’Adda di Fagagna, 2007; Childs et al., 2017; Gorgoulis et al., 2019). Senescent cells (SNC) accumulate with age, and in recent years have been demonstrated to actively drive numerous aging related diseases including atherosclerosis, cancer and Alzheimer’s disease (Bussian et al., 2018; Childs et al., 2017; Kirkland et al., 2017; Xu et al., 2018). Highlighting the complex and yet unresolved link between SNCs and diseases, clearance of senescent alveolar macrophages was recently shown to protect from early-stage lung tumorigenesis in mice (Prieto et al., 2023).

Cellular senescence is considered one of the fundamental aging process in which cells cease division and undergo a number of other distinctive phenotypic changes (Hayflick and Moorhead, 1961; McHugh and Gil, 2018). Throughout its lifespan, a cell is subjected to internal and external challenges leading to gradual accrual of damage. Simultaneously, as cells divide, they incrementally lose telomeric DNA, destabilizing the entire chromosomal structure. Both of these forms of damage make a cell susceptible to developing deleterious DNA mutations. Senescence programming prevents damaged cells from passing on damaged DNA to the next generation, making it a potent tumor suppressive mechanism. This process can be is advantageous in ensuring that an individual survives long enough to reproduce; however, it appears to be less advantageous to an aging organism. SNC become resistant to many apoptotic signals, contributing to their tendency to accumulate with age (Campisi and d’Adda di Fagagna, 2007). As these cells remain metabolically active and develop numerous phenotypic changes, their accrual can result in severe tissue dysfunction (Campisi and d’Adda di Fagagna, 2007; Childs et al., 2017). Apart from telomere erosion, multiple forms of cellular damage can induce senescence. Non-telomeric DNA damage, mitochondrial dysfunction, oxidative stress, chromatin disruption, and oncogene signaling are all capable of activating senescence programming. These cellular stresses can result from internal or external sources. Regardless of the initial stimulus, signaling converges on one of two pathways to halt progression through the cell cycle progression. There are two main tumor-suppressive pathways responsible for cell cycle arrest and senescence induction. These pathways are controlled by p16^INK4a (p16)/ pRB (retinoblastoma protein), and p53 respectively. While both are capable of halting replication, the two pathways are not entirely equivalent. In addition to being activated by different stimuli, the p53 pathway, via the effector p21, induces escapable cell cycle arrest, while p16/pRB induction results in permanent cell cycle arrest (Childs et al., 2017). Shortened telomeres or other forms of DNA injury prompt a DDR, which activates the master gene regulator, p53. Depending on the nature and severity of the DNA damage, p53 activation can result in either cell death via apoptosis, or cellular senescence (Chen et al., 2000; Debacq-Chainiaux et al., 2005; Probin et al., 2006). The interplay between these alternative cell fates remains unclear. In conditions favoring senescence, p53 activates transcription of the cyclin-dependent kinase inhibitor p21 (Zhang et al., 1999). p21 prevents progression through the cell cycle by blocking the cyclin-dependent kinases responsible for moving the cell from G1 to S phase. If the damage is repaired and DDR signaling ceases, p21-induced arrest also ceases, and the now healthy cell is able to re-enter the cell cycle. If the DNA damage is not repaired, sustained arrest eventually causes induction of p16, resulting in permanent exit from the cell cycle (Dulić et al., 2000). p16 can also be activated independently in response to oncogenic Ras or BRAF signaling, although this process is incompletely understood (Campisi and d’Adda di Fagagna, 2007; Di Micco et al., 2006). Bias towards p16 vs p21 activation varies between cell types and species (Sturmlechner et al., 2021).

In addition to exit from the cell cycle, SNC are also characterized by: resistance to apoptosis, increased lysosomal volume and lysosomal β-galactosidase production (senescence-associated β-galactosidase; SA βGal), chromatin remodeling, enlarged and flattened morphology, and secretion of a set of factors termed the senescence associated secretory phenotype – SASP (Campisi and d’Adda di Fagagna, 2007; Childs et al., 2017; Gorgoulis et al., 2019). SASP composition varies between cell types, but in general includes a host of proinflammatory cytokines, chemokines, extracellular matrix molecules and even reactive oxygen species (Coppé et al., 2011). These molecules act as paracrine signals in a form of sterile inflammation; in this way, SNC can alter their tissue microenvironment and negatively affect the function of neighboring cells. While overexpression of p16 alone is sufficient to induce replicative arrest, DDR signaling is required for the SASP (Coppé et al., 2011; Rodier et al., 2009). It is believed that secretion of these proinflammatory molecules underlies most of the detrimental effects of SNC. A systemic rise in inflammatory mediators is a common feature of aging. This phenomenon, termed “inflammaging” is associated with elevated risk for cardiovascular disease as well as other chronic aging-related diseases (Ferrucci and Fabbri, 2018; Franceschi et al., 2018). The root causes of inflammaging remains poorly understood, though immune dysregulation and possibly SNC accumulation are likely critical factors (Ferrucci et al., 2010; Fulop et al., 2017; Yousefzadeh et al., 2021). Overall, increasing evidence suggests that accumulation of SNC may be a major driver of the functional decline associated with premature aging.

A growing body of literature suggests that in both humans and mouse models chronic stress might promote accumulation of SNC, SASP and sterile inflammation (See also (Polsky et al., 2022) for review) (Table 2). Life stress has been associated with increased p16^INK4a in PBMC (Rentscher et al., 2019). Consistently, lifelong psychosocial stress in mice shortened lifespan and increased expression of senescence markers such as p16^INK4a and p53 in several peripheral organs in subordinate individuals (Razzoli et al., 2018). Additionally, chronic exposure to a non-social psychological stressor (restraint stress) increased p16^INK4a, p21, and SASP, and lowered DNA damage repair in severe combined immunodeficient (SCID) male mice (Rentscher et al., 2022). As discussed above, the vast majority of SASP factors contribute to a state of sterile local or systemic inflammation known as “inflammaging” (Ferrucci and Fabbri, 2018; Franceschi et al., 2018). Thus, studies measuring inflammation have the potential to capture signals coming from SNC, even if the SNC per se were not evaluated as a main source of those mediators. Likewise, aging immune cells can drive SNC and ageing of solid organs (Yousefzadeh et al., 2021) creating a circular loop resulting in increased sterile inflammation and organ damage. Several studies demonstrate that psychological stress promotes inflammation both in the brain, and in the periphery (Calcia et al., 2016; Frank et al., 2006; Rohleder, 2014). This finding has been replicated in both humans and laboratory rodents. For example, caregiver stress, low SES, work-related stress and childhood adversity are all associated with elevations in inflammatory markers (Rohleder, 2014). During an extended period of chronic stress, there are fluctuations in multiple cytokines. This was demonstrated in a study that subjected female mice to 6-hours of daily restraint stress for 28 days, monitoring serum cytokine levels throughout, and during 2 weeks of recovery (Voorhees et al., 2013). Circulating plasma levels of IL-6 were increased throughout the 4 weeks of restraint stress but returned to levels indistinguishable from controls by day 7 of recovery. Serum IL-10 was lower overall during the both the restraint and recovery phases. In the brain — specifically the cortex and hippocampus — IL-10 was reduced throughout the period of restraint and recovery as well. Chronic unpredictable mild stress increases circulating plasma levels of IL-6, IL-1β, TNF-α and IL-10 in rats (López-López et al., 2016). Changes in circulating cytokine levels are apparent after 20 days of stress and continued up until the conclusion of the experiment at 60 days. This preceded elevations in markers of oxidative damage to proteins and lipids in the liver.

Table 2.

Effect of chronic life stress or neuroendcirne stress mediators on selected hallmarks of aging: SNCs, SASP and sterile inflammation

Finding	Species	Population	Stress model/proxy	Tissue	Reference
Elevated proinflammatory cytokines	Rats	wildtype	Chronic unpredictable mild stress; 20–60 days	Blood plasma	Lopez-Lopez et al., 2016
Oxidative damage to proteins and lipids	Rats	wildtype	Chronic unpredictable mild stress; 20–60 days	Liver	Lopez-Lopez et al., 2016
Elevated Il-6, decreased IL-10	Mice	Female wildtype	Restraint; 6h/day for 28 days	Blood plasma	Voorhees et al., 2013
Decreased IL-10	Mice	Female wildtype	Restraint; 6h/day for 28 days	hippocampus, cortex	Voorhees et al., 2013
Microglial activation	Mice	Male wildtype	Restraint; 2 30 min sessions per day; 14 days	Brain	Tynan et al.,2010
Microglial activation and production of proinflammatory cytokines	Mice	Male wildtype	Repeated social defeat; 2h per day for 6 days	Brain	Wohleb et al., 2011
Elevated proinflammatory cytokines, iNOS Elevated	Mice	Male wildtype	Repeated social defeat; 2h per day for 6 days	Brain	Wohleb et al., 2012
proinflammatory cytokines, invasion by peripheral immune cells	Mice	Male wildtype	Repeated social defeat; 2h per day for 6 days	Brain	Wohleb et al., 2013
Elevated proinflammatory immune cell activity	Human	Young adults	Low SES	PBMCs	Powell et al., 2013
Elevated proinflammatory immune cell activity	Mice	Male wildtype	Repeated social defeat; 2h per day for 6 days	PBMCs	Powell et al., 2013
Elevated proinflammatory immune cell activity	Human	Medical residents	Reported stress levels	PBMCs	Heidt et al., 2014
Increased p16 or p53 in various organs;	Mice	wild type	chronic psychosocial stress	liver, lung, spleen	Razzoli et al., 2018
Increased p16	Human	adults	chronic stress	PBMC	Rentscher et al., 2019

Neuroinflammation is another well-established consequence of chronic stress and increased inflammation is a common phenotype of normal brain aging across species (Frank et al., 2006). Several stressors promote microglial activation, as measured by Iba1 activity (Calcia et al., 2016). In particular, a very strong effect has been found in area CA1 of the hippocampus (Bian et al., 2012). This increase in microglial activation leads to the secretion of cytokines, chemokines and other SASP factors facilitating the development of anxiety-like behavior (Tynan et al., 2010; Wohleb et al., 2013). Stress exposure also promotes production of proinflammatory monocytes in the bone marrow, which are released into general circulation and can migrate and localize in the brain in addition to other organs (Heidt et al., 2014; Wohleb et al., 2013) thereby secreting SASP factors. Finally, a less explored area in this field is the neuro-immune modulation of bone structure during homeostasis and upon trauma. Mice lacking tyrosine hydroxylase, the rate limiting enzyme for the synthesis of norepinephrine in myeloid cells are protected from chronic psychosocial stress-induced disturbance of bone growth and healing (Tschaffon-Müller et al., 2023), a finding matched by knockout of β2-adrenoceptor in chondrocyte.

EFFECT OF LIFE STRESS ON ROS AND DNA DAMAGE

The free radical theory of aging, first proposed in the 1950s, posits that aging is the result of accumulated damage created by reactive oxygen species (ROS) (Harman, 2009). Though it has now been shown that oxidative damage does not alone drive aging, it is certainly one of the mechanisms contributing to the aging process. ROS, and the damage that they wreak are a major cause of cellular senescence. ROS and oxidative damage increase with age, and increased production of ROS is associated with shortened lifespan (Kirkwood and Kowald, 2012; Stadtman and Levine, 2000). ROS are byproducts of normal cellular metabolism, are present at low levels during homeostasis (Boonstra and Post, 2004) and serve as signaling molecules in optimally functioning cells. ROS are also capable of modifications to DNA. The most frequent form of oxidative damage to DNA is 8-hydroxylation of guanine (Poulsen et al., 2014). In some cases, ROS-induced damage causes rupture to the DNA sugar-phosphate backbone resulting in single or double strand breaks (DSB), which are the most dangerous form of DNA damage. Attempts to repair DSBs can cause loss or translocation of chromosomal segments causing tumorigenesis or replicative senescence. Specialized protein kinases sense the DSB, causing local phosphorylation of the histone H2AX (called γ-H2AX in its phosphorylated form), a classical marker of SNC and a signaling cascade culminating in activation of the master regulator p53. p53 induces cell cycle arrest by activating transcription of p21, as described above in the cellular senescence section.

Studies over the last 30 years have demonstrated that psychological stress, both acute and chronic, can result in the formation of ROS (See also (Salim, 2016; Sharifi-Rad et al., 2020) for additional review) (Table 3). Measures of oxidative damage were obtained in lymphocytes from undergraduate students immediately prior to an exam, and on a “non-stress” occasion for comparison (Sivonová et al., 2004) and found increased DNA damage, greater lipid susceptibility to oxidation, and decreased antioxidant levels in plasma on the exam day. In another study on undergraduate students, exam stress increased saliva levels of the antioxidant catalase and sialic acid, and decreased levels of oxidized proteins (Tsuber et al., 2014). The increase in catalase was greater in female participants than in male, indicative of greater intrinsic antioxidant defense. This could represent a compensatory mechanism to combat elevated ROS. Exposure to the Trier Social Stress Test (Kirschbaum et al., 1993) – a classical test in which participants perform a public speaking task and then attempt mental arithmetic – elevates saliva levels of several miRNAs (Wiegand et al., 2018) including miR-21, directly inhibits superoxide dismutase (SOD), resulting in impaired ROS degradation.

Table 3.

Effect of chronic life stress or neuroendcirne stress mediators on selected hallmarks of aging: ROS and DNA damage

Finding	Species	Population	Stress model/proxy	Tissue	Reference
Greater DNA damage (assessed by comet assay), higher lipid peroxidation, decreased antioxidant levels	Human	University students	Taking an exam	Leukocytes, plasma	Sivonova et al., 2004
Greater protein and lipid oxidative damage; lower antioxidant level	Human	Female university students	Taking an exam	Blood serum	Nakhaee et al., 2013
Lower antioxidant levels; wack sperm	Human	Male university students	Taking an exam	Semen	Eskiocak et al., 2005
Higher inhibitors of antioxidants	Human	University students	Trier Social Stress Test	Saliva	Wiegand et al., 2018
Increased antioxidant levels	Human	University students	Taking an exam	Saliva	Tsuber et al., 2014
Greater oxidative stress	Human	Female adolescents	Adverse childhood events	Urine	Horn et al., 2019
Greater lipid peroxidation, IL-6	Human	Children	Low SES	Blood serum	Mansur et al., 2016
Greater oxidative stress	Human	Hypertensive patients	High trait anxiety	PBMCs, blood plasma	Yasunari et al., 2006
Greater lipid peroxidation	Rat	wildtype	Acute air jet stress	Plasma	D’Angelo et al., 2010
			Single 5-hour communication box stress		Adachi et al., 1993
No effect	Rat	wildtype		Liver
Increased oxidative damage; returns to control levels within 1 hour	Rat	wildtype	3–4 exposures to 5-hour communication box stress	Liver	Adachi et al., 1993
Increased oxidative damage, HSP70 decreased antioxidant activity,	Rat	wildtype	Communication box stress exposure; 1h/day for 3 or 5 weeks	Masseter muscle	Li, Zhang et al., 2011
Increased lipid peroxidation, DNA fragmentation, decreased cytochrome C	Rat	wildtype	Water immersion stress; acute; 90min	Gastric/Intestinal mucosa	Bagchi et al., 1999
Increased lipid peroxidation, DNA fragmentation, decreased cytochrome C	Rat	wildtype	Water immersion stress; 15min/day for 15 days	Gastric/Intestinal mucosa	Bagchi et al., 1999
Elevated ROS, elevated NO, increased lipid peroxidation, protein carbonylation; reduced antioxidants	Mice	wildtype	Restraint; 5h/day, 14 days	hippocampus, cortex, cerebellum	Son et al., 2016
Increased oxidative DNA damage, reactive nitrogen species	Rat	wildtype	Social isolation; 14 days	Nucleus accumbens, hypothalamus	Colaianna et al., 2013
Increased oxidative DNA damage, reactive nitrogen species Increased	Rat	wildtype	Social isolation; 28 days	prefrontal cortex	Colaianna et al., 2013
oxidative DNA damage, reactive nitrogen species	Rat	wildtype	Social isolation; 49 days	adrenal glands	Colaianna et al., 2013
Increased oxidative DNA damage	Mice	SAMP-10 Human	confrontational housing	cerebral cortex	Unno et al., 2011
Increased ROS, RNS, DNA damage, impaired DNA repair	Cell culture	Breast cancer lines MDA-MB-231, MCF-7 Human	Cortisol (15, 30, or 90min)	cell culture	Flaherty et al., 2017
Increased ROS, RNS, DNA damage, impaired DNA repair	Cell culture	Breast cancer lines MDA-MB-231, MCF-7	Cortisol (15, 30, or 90min)	cell culture	Flaherty et al., 2017
Increased iNOS	Mice	females injected in mammary fat pad with human breast cancer cells	Restraint; 5h/day, 14 days	Mammary tumors	Flaherty et al., 2017
Elevated DNA damage, impaired DNA repair, upregulated cell cycle inhibitor CDC25A	Cell culture	Murine 3T3 cells	Cortisol (30 min)	cell culture	Flint et al., 2007
Elevated DNA damage, impaired DNA repair, upregulated cell cycle inhibitor CDC25A	Cell culture	Murine 3T3 cells	Norepinephrine (30 min)	cell culture	Flint et al., 2007
Elevated DNA damage, upregulated cell cycle inhibitor CDC25A	Cell culture	Murine 3T3 cells	Epinephrine (30 min)	cell culture	Flint et al., 2007
Poorer DNA repair	Human	Nonpsychotic psychiatric patients avg age 32	Reported distress (Minnesota Multiphasic Personality Index)	Lymphocytes	Kiecolt-Glaser et al., 1985
Increased DNA damage	Mice	wildtype males	Isoproterenol	Thymus	Hara et al., 2011
Increased DNA damage	Cell culture	Human U2OS, Murine MEF	Isoproterenol	cell culture	Hara et al., 2011

The stress increase in ROS suggests that compensatory mechanisms to combat acute stress-induced ROS can be insufficient in the long term under certain conditions. Adolescents who experience multiple ACE showed increased urinary levels of the oxidative stress markers F2-isoprostanes (Horn et al., 2019). Socioeconomic disadvantage is associated with elevated plasma serum thiobarbituric acid-reactive substance as well as IL-6 (Mansur et al., 2016). Finally, hypertensive patients with high trait anxiety have higher plasma levels of norepinephrine, which correlates with increased ROS in mononuclear immune cells (Yasunari et al., 2006).

Rodent studies have recapitulated the stress-induced ROS phenomenon observed in humans and have shed greater light on the mechanisms underlying this effect. A 5-hour exposure to the communication box stress paradigm did not affect rat liver levels of OH⁸dG – a product of hydroxyl radical induced DNA damage (Adachi et al., 1993). The induced increase returned to levels equivalent to that of controls 1-hour after the stress exposure. Thus, it appears that neutralization of ROS is possible after several repeated stress exposures. However, after a second, third or fourth exposure, stress-exposed rats had significantly higher levels of OH⁸dG in liver nuclear DNA. Another experiment further elucidated the time course in which stress-induced ROS recovery can be compromised. This experiment employed the same communication box stress paradigm on rats in a chronic study design. The rats were exposed for 1-hour per day for 1, 3 or 5 weeks. Researchers subsequently found an increase in lipid peroxidation marker MDA content in the masseter muscles after 3 and 5 weeks but not 1 week of stress (Li et al., 2011). Concurrently, following 3 or 5 weeks of stress, they found a decrease in the activity of antioxidant enzymes SOD, glutathione peroxidase, and catalase in the masseter muscles. In rats exposed for 5 weeks there was also an elevation in the cellular stress-sensitive protein HSP70. Together, these studies suggest that psychological stress can induce an imbalance favoring greater ROS and resulting in oxidative damage, despite upregulation of the protective HSP70. Psychological stress is capable of increasing ROS in numerous tissue types, though it seems to have a particularly profound effect on the brain. 14 days of non-social psychological restraint stress resulted in higher levels of ROS in the hippocampus and cerebral cortex, and higher levels of nitric oxide in the hippocampus and cerebellum in mice (Son et al., 2016). It also induced greater levels of lipid peroxidation marker MDA in all three brain regions and protein peroxidation marker protein carbonyl in the hippocampus and cortex. In contrast it reduced levels of the antioxidant enzymes SOD and glutathione reductase-transferase.

In rats, 2 weeks of social isolation are sufficient to increase mRNA expression of NADPH oxidase (NOX2), an enzyme that catalyzes formation of superoxide anion O_2, in the nucleus accumbens and hypothalamus (Colaianna et al., 2013). This is associated with an increase in 8-OHdG in both regions, and an increase in nitrotyrosine – a marker of reactive RNS— in the hypothalamus. After 4 weeks, elevations of NOX2 and 8-OHdG in the prefrontal cortex can be detected as well. After 7 weeks, these markers are also elevated in the adrenal glands. The majority of 8-OHdG and nitrotyrosine positive cells in the hypothalamus were neurons. In the prefrontal cortex, the NOX2 subunit p47^phox was increased specifically in pyramidal neurons, not in GABAergic neurons, astrocytes, or microglia (Schiavone et al., 2012). Alterations in behavior, increase in glutamate levels and loss of parvalbumin were detectable after 4 weeks of social isolation.

In the accelerated senescence SAMP-10 mouse model, confrontational housing — a model for psychosocial stress — increased levels of the DNA damage marker 8-oxo-dG can be found in the cerebral cortex (Unno et al., 2011). This was also associated with accelerated cognitive decline and shorter lifespan.

The mechanisms underlying stress-induced ROS overproduction have been further illuminated using molecular mediators of the stress response using cell culture models. Acute exposure to cortisol or norepinephrine was sufficient to increase levels of ROS and RNS in breast cancer cells, which was paralleled by increase in γ-H2AX foci, indicative of DNA damage and senescence, an effect mediated by their respective receptor antagonists (Flaherty et al., 2017). In vitro and in vivo studies using endocrine mediators of the stress response indicate that stress may impair the repair of damaged DNA. Short term (<30 minute) in vitro exposure of murine 3T3-L1 cells to cortisol or norepinephrine not only increase DNA damage on its own, but also interfered with DNA repair in cells exposed to UV radiation (Flint et al., 2007). Epinephrine exposure produced a similar degree of DNA damage but did not impair DNA repair. All three stress-related hormones induced upregulation of DNA damage sensors Chk1 and Chk2, and the proto-oncogene CDC25A, which is involved in cell cycle delay following DNA damage. Consistently, 24-hour exposure to epinephrine or norepinephrine increases DNA damage, in the murine 3T3-L1 cells, an effect that was blocked by pre-treatment with the β-adrenergic receptor antagonist propranolol (Flint et al., 2013). This damage is functionally relevant, as cells treated with either catecholamine undergo greater malignant transformation in-vitro and become tumors more rapidly in-vivo when implanted in mice (Flint et al., 2013). These findings in murine systems can be extended to other cell types and species. In human breast cancer cells, DNA repair following exposure to H₂O₂ was impaired in the presence of cortisol (Flaherty et al., 2017). Furthermore, lymphocytes sampled from participants who scored highly for distress on the Minnesota Multiphasic Personality Inventory exhibited poorer DNA repair 5 hours after X-irradiation (Damjanovic et al., 2007; Kiecolt-Glaser et al., 1985). Consistent with in vitro work, mice treated with the β-adrenergic receptor agonist isoproterenol accumulated DNA damage in the thymus, as quantified by γ-H2AX (Hara et al., 2011). Isoproterenol treatment also decreased levels of the DNA damage responder, and senescence regulator p53. These findings were replicated in vitro in several cell lines and was further demonstrated to be mediated specifically by β2-adrenergic receptor (β2-AR) signaling. The authors suggest that stress-induced sympathetic nervous system activity, promotes DNA damage both directly, and indirectly via suppression of p53-mediated DNA repair (Hara et al., 2011). This may consequently result in worse DNA damage, and a lower likelihood of successful repair ultimately leading to permanent senescence.

This substantial body of work provides convincing evidence that life stress or stress-related hormones causes an increase in ROS production, and that this results in oxidative damage to DNA in numerous cell types/organs. Stress-induced ROS overproduction directly results in DNA damage — one of the primary triggers for cellular senescence. Multiple stress hormones appear to induce DNA damage via increased ROS production, as well as impair DNA repair. However, stress relevant studies using murine models are limited in this area and conclusions obtained using in vitro system remain to be confirmed in vivo.

EFFECT OF LIFE STRESS ON EPIGENETIC REGULATION

The hallmarks of aging discussed in the previous sections are implicated in shaping the progression of biological aging, the gradual process that leads to a functional decline over the lifetime and that represents a major risk factor for age-related diseases. Biological age is not necessarily reflective of an individual’s chronological age because it encodes exposures to disease, lifestyle changes, environmental factors, stress, etc., that result in acceleration or deceleration of the pace of aging and of the linked propensity to develop pathologies (Zhang and Gladyshev, 2020). Biological age can be estimated using various approaches including epigenetic biomarkers – specifically DNA methylation (DNAm) changes. DNAm consists of the addition of a methyl group to a CpG dinulceotide in most cases associated with transcriptional repression via its effect on chromatin accessibility. DNAm occurs mostly following predictable patterns, initially established for organismal development and later modulated for maintenance, resulting in gain of DNAm at gene promoters and loss of general DNAm during aging (Singer, 2019). Similarly, epigenetic alterations to DNA and chromatin play a crucial role in mediating the induction and maintenance of cellular senescence, though processes such as heterochromatin loss and instability, histone acetylation and methylation, the action of Polycomb-group proteins (Crouch et al., 2022). Like in aging, DNAm patterns in senescent cells consist of a general state of hypomethylation (Cruickshanks et al., 2013). While sharing considerable overlap with age-associated methylation patterns, the ones associated with senescence also present differences at specific CpG sites, most commonly at HOX genes and genes mediating cellular differentiation (Bork et al., 2010). Altogether, alterations in DNAm and associated enzymes during aging and senescence significantly contribute to cellular dysfunction, facilitate a chronic inflammation state via SASP activation and may mediate certain age-related pathologies.

These changes in methylation patterns occupy a major field in geroscience. The predictability of these changes enable the application of machine learning techniques to develop age predictors – termed epigenetic clocks (Horvath and Raj, 2018). Since their advent (Horvath clock, (Horvath, 2013); Hannum clock, (Hannum et al., 2013)), a number of additional epigenetic clocks have been developed incorporating DNAm levels at sites correlating with morbidity and mortality and therefore better associated with measures of physiological dysregulation PhenoAge, (Levine et al., 2018); GrimAge, (Lu et al., 2019); DunedinPoAm, (Belsky et al., 2020)).

Increasing evidence links psychosocial stress with accelerated epigenetic aging (See also (Palma-Gudiel et al., 2020) for additional review) (Table 4). Social stress is believed to become biologically encoded through epigenetic modifications of genes such as the glucocorticoid receptor, CRH and FKBP5 genes (Cunliffe, 2016; Zannas et al., 2019). Stress related alterations in neuronal activity have also been linked to changes in DNAm (Fitzgerald et al., 2021), possibly due to numerous CpG sites they include are in the glucocorticoid response elements (Horvath et al., 2013). The administration of exogenous glucocorticoids has been shown to affect DNAm and transcription in many of the genes neighboring CpG of the Horvath’s clock (Zannas et al., 2020). The DunedinPACE biomarker of aging represents the proof of concept that the pace of aging can be measured from the DNAm levels of the most reliable CpG sites assessed at one timepoint jointly with biomarkers of physiological integrity (Belsky et al., 2022). The landmark Dunedin studies reflect at a cohort of over 1000 individuals followed longitudinally, determined biological age through indicators of organ system integrity recorded in multiple occasions and a single time point DNAm and estimated a faster aging pace in young adults with a history of ACE (Belsky et al., 2022, 2020).

Table 4.

Effect of chronic life stress or neuroendcirne stress mediators on selected hallmarks of aging: Epigenetic changes

Finding	Species	Population	Stress model/proxy	Tissue	Reference
Resilience dependent Grim Age acceleration	Human	adults	Perceived stress score	Whole blood	Bergquist et al., 2022
No association with DNA methylaton age acceleration	Human	older adults	Perceived stress score	Whole blood	Vetter et al., 2022
Increased epigenetic aging via DNA methylation	Human	adolescents	economic trajectories during the Great Recession	blood leukocytes	Chen et al., 2016
Increased epigenetic aging via DNA methylation	Human	adults	Low SES	Whole blood	Fiorito et al., 2017
Increased epigenetic aging via DNA methylation	Human	adults	Low SES	Whole blood	Simons et al., 2017
Older epigenetic age in high ranking individuals	baboons	adults	high rank	Whole blood	Anderson et al, 2022
Increased DNAm in low rank individuals	mice	adult/old	low rank	liver	Razzoli et al., 2023

Childhood trauma has been linked by several studies to epigenetic age acceleration similarly to long term shift work, sleep deprivation and obesity (Lim et al., 2022). Interestingly, the effect of stress on epigenetic aging depends on the individual coping style and resilience, with multiple studies showing an inverse correlation between acceleration of epigenetic age due to stress exposure and coping ability in individuals exposed to earlier trauma (Bergquist et al., 2022; Galow and Peleg, 2022; Vetter et al., 2022).

Human epigenetic clocks are based on DNAm levels obtained mostly with micro-array assay, while the development of array-based aging clock for other species has been somewhat trailing behind, due to the lack of availability of BeadChips platforms for non-human species. Nevertheless, new studies now show that in social species individuals experiencing high levels of glucocorticoids and upregulated inflammation related pathways also show acceleration of biological aging based on telomere length as well as epigenetic clocks (Anderson et al., 2021). Wild baboons have been shown to pay the elevated energetic cost of maintaining high ranking social position by accruing an older epigenetic age than lower ranking individuals (Anderson et al., 2021), thus attributing to social determinants as the source of the variance in biological aging observed in natural populations. Murine epigenetic clocks have been developed based on sequencing-based methods as tissue-specific (i.e., liver (Wang et al., 2017), blood (Petkovich et al., 2017), brain (Liu et al., 2021); or multi-tissue (Kerepesi et al., 2021; Stubbs et al., 2017). Among the manipulations that have been found to affect epigenetic age in mice so far are, caloric restriction (Cole et al., 2017), dietary rapamycin treatment (Wang et al., 2017) or high fat diet feeding (Stubbs et al., 2017). The association between stress and epigenetic changes has been recently reviewed, particularly regarding their impact on rodent behavior (Dion et al., 2022). However, only a recent study to date has specifically examined the relationship between life stress, DNAm, epigenetic clocks and lifespan in a mouse model of chronic psychosocial stress (Razzoli et al., 2023). This study identified epigenetic changes (DNAm) in the liver of male mice, which are in line with increased risk conferred by genetic background (Sv129Ev > C57BL/6J > CD1) and social rank (low rank > high rank) on healthspan and lifespan. In this study, mouse epigenetic age was found to be strain dependent, while no link was found to the mouse social standing, likely due to the inadequacy of the available epigenetic clocks based on mouse DNAm array data. This limited knowledge on how life stress impacts epigenetic clock, as well as the unclear interaction of stress, epigenetic clocks, biological age and cellular senescence is likely to be rectified as research and technology advances. Learning from the evolution undergone by the human epigenetic clock, it’s desirable that future mouse studies will be designed taking into consideration measures stress experiences, DNAm and physiological phenotypes in the same population. Acquisition of this knowledge will be facilitated by accessibility of reliable experimental and computational tools and by conscious experimental design.

KNOWLEDGE GAP AND FUTURE DIRECTIONS

Given the degree of novelty the field of geroscience is infused of in terms of conceptual, computational, biological and technological innovations (Duque et al., 2023; Epel, 2020; Kennedy et al., 2014; Sierra and Kohanski, 2023), it is not surprising that knowledge gaps of are still numerous. Amongst the more urgent needs, studies are required in which social experiences, biological age measures and stress phenotypes are measured in the same population to elucidate the mechanistic link between social experience and system function. In line with this, more animal studies should examine the effect of varying type/timing of social experience on biomarkers of aging and stress phenotype, both in the laboratory as well as in wild settings.

While DNAm is one of the most accurate and reliable biomarkers of aging and disease, our understanding of it is still in its infancy. Particularly whether changes in DNAm is causal to aging remains to be shown (Simpson and Chandra, 2021). Similarly, despite the role of SNC in aging and disease risk, there is sparse information about their cellular identity and features as well as about the implications of their heterogeneity in human tissues. None of the current SNC biomarkers are specific or unique to SNC (P. J. Lee et al., 2022), calling for the need of developing multiple targeted and unbiased approaches to identify these cells. Overall, these efforts should culminate in the identification of sensitive and specific gerodiagnostics – biomarkers to quantify the various hallmarks of aging. – and in the design of interventions; this would allow the elevation of biological age by stress to become a truly quantifiable and actionable therapeutic target (Chaib et al., 2022; Kirkland et al., 2017).

Lastly, a significant gap of knowledge concerns the mechanism through which life stress affects multiple hallmarks of aging, impairs multi-organ physiology, causes comorbid diseases, and overall affects life expectancy. Figure 1 describes the major stress-related neuro-endocrine systems as well as the consequences of life stress on selected hallmarks of aging at the organ level in humans and animal models. However, the specific link between one or more of those stress-related axes, and a given change in these hallmarks is still unclear, and largely unexplored. Cellular models have been used to tackle this question by testing the effects of “stress hormones” such as cortisol on induction of hallmarks of aging (Bobba-Alves et al., 2023; Laberge et al., 2012; Zannas et al., 2020). Limited studies have also tested the role of such mediators (catecholamines) on grey hair formation and stem cell exhaustion (Zhang et al., 2020). Yet contrasting findings have been obtained in those studies ranging from induction to inhibition of SNC, SASP and telomere shortening, suggesting that the in vivo “neuro-symphony of stress” mediators (Joëls and Baram, 2009) can interact in complex manners during the life course to impact fundamental biological mechanisms affecting organ function and aging. In essence, it appears that the use of conventional reductionist approaches of decomposing organisms into unitary tracts manipulated one at a time by using loss/gain of function of single genes/factors can be ineffective to tackle the fundamental question of how life stress accelerates aging. To use Jacob’s words “Clearly, an understanding of the simple is necessary to understand the more complex, but whether it is sufficient is questionable” (Jacob, 1977). An alternative holistic approach focused on the organism as a unit, regulated by multiple – sometime opposing – physiological control mechanisms should be preferred (Stambler, 2017). These physiological control mechanisms should be ideally investigated simultaneously to determine how the organism adapts or succumbs to the strain posed by multiple environmental challenges (including social interactions). However, such a construct remains to be defined in terms of experimental design and feasibility. Artificial intelligence algorithms and innovative bioengineering approaches can ultimately benefit the use of holistic approaches by sharing the potential to address multiple functions in the same organism over its lifespan as well as in large enough populations to contrast background noise and intrinsic inter-individual variability.

The field of geroscience attempts to identify fundamental processes affecting aging in multiple organs and in multiple dimensions, aiming at developing therapies to extend the healthy lifespan. Classical examples of interventions improving healthy-lifespan, and/or extending lifespan in some animal studies, include caloric restriction and exercise (Balasubramanian et al., 2017; Huffman et al., 2016; Waziry et al., 2023). One of the major innovations concerns the promise offered by drug treatments with anti-aging effect such as rapamycin, metformin and others (Partridge et al., 2020) as well as systemic, or targeted clearance of SNC (senolytics) or inhibition of the effects exerted by SASP (senomorphics) to offset multi-morbidities of aging and compress pathologies overall benefiting healthy aging (Childs et al., 2017; Kirkland et al., 2017; P. J. Lee et al., 2022). To the best of our knowledge, no study has been conducted in which human subjects or animal models exposed to life stress – characterized by impaired healthspan, increased biological age, and reduced lifespan - were randomized to receive therapeutic interventions proven to exert a beneficial effect in stress-free populations.

Intriguingly, in parallel to the extensive literature discussed above showing that life stress increase biological aging in humans and mice, a recent study showed that surgery, pregnancy or viral infection increase epigenetic age in humans, albeit reversibly upon recovery (Poganik et al., 2022). Although pregnancy, surgery, and viral infections cannot be defined as stressors per se [based on the definition used in this paper (Koolhaas et al., 2011) and by others (Epel et al., 2018)], they share some similarity with life stressors, including elevated HPA axis, and often catecholamines, suggesting that life stress-induced acceleration of biological aging could be sensitive to interventions increasing resilience and/or inducing reversal.

Additionally, genetic reprogramming of aging-induced DNA methylation patterns has recently been proposed to “rejuvenate” organ functions and hallmarks of aging in mice models (Lu et al., 2020; Poganik et al., 2022; Yang et al., 2023). Can clearance of stress-induced SNC or targeted/global reprogramming of stress-induced DNA hypo/hyper methylation be sufficient to oppose the multi-morbidities imposed by life stress? No empiric answer is available yet.

CONCLUSION

Chronic stress poses a significant public health threat and is a likely major source of health disparities. It is associated with elevated risk for chronic aging related diseases, which are the primary causes of disability and premature death in older adults. Psychosocial stress is a particularly potent form of stress and is a key contributor to the numerous adverse health outcomes associated with ACE and low SES. Chronic stress has long been theorized to accelerate the aging process. In this Review, we gave an overview of the scientific literature that implicates a link between psychological stress and several pillars/hallmarks of aging including cellular senescence and DNAm. Stress has been demonstrated to promote numerous senescence triggers, including telomere erosion, oxidative damage, and DNA damage. Moreover, stress is associated with systemic inflammation, which can both promote senescence-inducing cellular damage, and be a consequence of SNC accumulation. Overall, available studies suggest that life stress can lead to a vast array of aging-related diseases at least in part by activating cellular senescence and other biological processes associated with premature aging. Therefore, we postulate that these senescence/aging pathways should be considered as an intrinsic biological manifestation of chronic stress-induced perturbations. More attention should be focused on how endocrine/neural mediators of the stress response affect aging-associated processes and how buffering those negative effects could be used therapeutically to reduce the pathological burden of stress on life expectancy.

Highlights.

• Chronic stress is a major risk factor for a multitude of age-related diseases.

• Hallmarks of aging are basic biological mechanisms that can lead to accelerated aging.

• Stress has been increasingly associated with the dysregulation of the hallmarks of aging.

• We review evidence on the contribution of life stress on cellular senescence, sterile inflammation, telomere length, production of reactive oxygen species, DNA damage, and epigenetic changes.

• We evaluate stress-induced senescence and accelerated aging as valid etiopathological proposition.