Stress and Alzheimer’s Disease: A Senescence Link?

Abstract

Chronic stress has been shown to promote numerous aging-related diseases, and to accelerate the aging process itself. Of particular interest is the impact of stress on Alzheimer’s Disease (AD), the most prevalent form of dementia. The vast majority of AD cases have no known genetic cause, making it vital to identify the environmental factors involved in the onset and progression of the disease. Age is the greatest risk factor for AD, and measures of biological aging such as shorter telomere length, significantly increase likelihood for developing AD. Stress is also considered a crucial contributor to AD, as indicated by a formidable body of research although the mechanisms underlying this association remain unclear. Here we review human and animal literature on the impact of stress on AD and discuss the mechanisms implicated in the interaction. In particular we will focus on the burgeoning body of research demonstrating that senescent cells, which accumulate with age and actively drive a number of aging-related diseases, may be a key mechanism through which stress drives AD.

Introduction

Chronic stress is a risk factor for a number of aging-related diseases, including Alzheimer’s Disease (AD), the most prevalent form of dementia. The vast majority of AD cases have no known genetic cause, indicating an important role for environmental factors. In humans, high stress levels increase risk for developing AD (Moceri et al., 2001; Wilson et al., 2003). High stress levels both accelerate the age of onset for familial early-onset AD, and increase likelihood of progressing to advanced stages of the disease in late-onset AD (Moceri et al., 2001; Peavy et al., 2012). In addition to chronological age, measures of biological aging, such as shorter telomere length, significantly increase likelihood for developing AD (Honig et al., 2012; Zhan et al., 2015). Here, we will survey the current literature on the impact of stress on AD and discuss the mechanisms implicated in the interaction, particularly a role for cellular senescence as mediator.

The stress response

Since its first descriptions, the phenomenon known as stress has been intimately linked to disease and aging (Selye, 1950, 1936). While its definition has been, and continues to be debated and updated overtime (Kagan, 2016; Koolhaas et al., 2011; McEwen, 1998; Nederhof and Schmidt, 2012; Pacák and Palkovits, 2001; Romero et al., 2009; Sapolsky et al., 2000; Selye, 1936), this association has endured. The stress nomenclature has been revised to restrict it to conditions characterized by uncontrollability and unpredictability that exceed the regulatory range and adaptive capacity of an individual (Koolhaas et al., 2011). Challenges to homeostasis and the body’s associated physiological responses are a constant feature of living organisms. Stressors are distinguished from these challenges by having physiological demands outside the natural regulatory capacity for an organism. Activation of the acute stress response is a normal part of homeostatic regulation of the body, but under some conditions this response becomes sustained. This persistent activation is associated with a number of maladies, can lead to sudden death in vulnerable individuals, and can significantly shorten individual lifespan (Razzoli et al., 2020, 2018; Snyder-Mackler N, et al. 2020; Stringhini et al., 2017; Tung et al., 2016; Zipple et al., 2019).

The neuroendocrine stress response is classically defined as having two major components: the sympathetic-adrenal medullary (SAM) system and the hypothalamic pituitary adrenal (HPA) axis (for review, see Charmandari et al., 2005 and Ulrich-Lai and Herman, 2009). In response to a stressor, the SAM system, commonly referred to as the “fight or flight response”, (Cannon, 1915) is rapidly activated. Preganglionic neurons in the sympathetic nervous system activate chromaffin cells in the adrenal medulla resulting the release of epinephrine (and to a lesser extent, norepinephrine) into general circulation. These chemicals act on via adrenergic receptors in a wide range of tissue types, resulting in increased heart rate, respiratory rate, glycolysis, glycogenolysis and lipolysis.

In the second component, the HPA axis, activated neurons in the paraventricular nucleus of the hypothalamus release corticotropin releasing hormone (CRH) into the hypophysial portal system. CRH reaches the anterior pituitary gland, causing it to release adrenocorticotropin hormone (ACTH) into general circulation. ACTH stimulates the adrenal cortex to produce glucocorticoids (GCs) which act on a number of targets. GCs also exert negative feedback on the hypothalamus, halting the release of CRH, and on the pituitary, halting the release of ACTH. However, in cases of sustained stress or aging, GC hypersecretion can persist (Sapolsky et al., 1986; Spiga et al., 2014). Additionally, this negative feedback system is impaired with age, largely through the downregulation of glucocorticoid receptors (GR) in the hippocampus (Sapolsky et al., 1986). HPA axis hyperactivity is associated with numerous negative outcomes, including major depression (Pariante and Lightman, 2008), immune suppression (Oppong and Cato, 2015) and memory impairment (Tatomir et al., 2014).

In addition to the canonical “stress axes” described above, several additional neural and neuroendocrine system have been shown to regulate the acute stress response (Sapolsky 2000 and Ulrich-Lai and Herman, 2009 for review) including a newly identified role for osteocalcin (Berger et al., 2019).

Stress and Alzheimer’s Disease

In humans, stress has been associated with a number of neurodegenerative diseases, including Alzheimer’s Disease (AD) (Yuede et al., 2018), other dementias (Bonanni et al., 2018; Johnston, 2000), and Parkinson’s Disease (Hemmerle et al., 2012). AD is the most common form of dementia and results in a debilitating loss of memory and cognitive function. Early on, the disease primarily manifests in the form of impaired memory, reflecting the location of initial neural degeneration in the entorhinal cortex and hippocampus. Subsequently, neural atrophy spreads to other limbic and neocortical regions and symptoms progress to severely alter cognitive and behavioral domains (Braak et al., 2006; Braak and Braak, 1991). The pathophysiological process begins years, and even decades before the onset of clinical symptoms (Morris, 2005). Since its identification by Alois Alzheimer in 1910, AD neuropathology has been characterized by profound neuron loss as well as two hallmarks: neurofibrillary tangles (NFTs) and amyloid β (Aβ) plaques. NFTs appear intracellularly and are comprised of aggregated hyperphosphorylated tau protein (Alonso et al., 1996). Neurotoxic Aβ plaques occur extracellularly and are composed of clumped Aβ peptide (Selkoe, 2001). While these are considered markers of the disease and their presence in postmortem analysis is required for a definitive AD diagnosis, neuron and synapse loss are believed to underlie the declines in memory and cognition. Indeed, studies have shown that one of the best pathological correlate to early disease progression is synapse loss (Davies et al., 1987; Masliah et al., 2001; Scheff et al., 2007).

With the exception of rare early onset forms, symptoms typically first appear in patients over age 65, making AD primarily a disease of aging. Though the cause(s) of AD remain poorly understood (over 90% of cases have no known genetic cause (Bertram and Tanzi, 2004), it is clear that biological aging plays a role in its etiology. Multiple studies show a correlation between shorter telomere length in circulating immune cells and increased risk for AD (Blasco, 2005; Honig et al., 2012). A more sophisticated analysis established a causal link between shorter telomeres and AD risk, concluding that for each standard deviation above average telomere attrition rate, risk for AD increases by 36% (Zhan et al., 2015). Additionally, it is widely believed that environmental factors such as air pollution (Killin et al., 2016), nutrition (Morris, 2009) and stress (Yuede et al., 2018) play a role in disease onset and progression, along with biological factors such as aging (Prather et al., 2015). Interactions between such environmental elements with aging processes like telomere length, oxidative stress and cellular senescence may play a critical role in their effect on AD risk.

In humans, there is strong epidemiological evidence linking chronic stress and dementia. For example, in the Religious Orders study, individuals who scored in the 90th percentile for “proneness to psychological distress” were twice as likely to develop AD as those with scores in the 10th percentile. In particular, distress proneness was associated with a decline in episodic memory, though not with other cognitive domains (Wilson et al., 2003). Thus, individuals who experienced a higher degree of perceived stress were at much higher risk for developing AD. A similar result was found in a broader population as well (Crowe et al., 2007). In this study, higher reactivity to stress was associated with a dementia diagnosis 30 years later, even within twins. Furthermore, veterans with posttraumatic stress disorder have a 2-fold higher risk for developing dementia than veterans without the condition (Yaffe et al., 2010).

Several measures of socioeconomic status predict risk for AD. Low socioeconomic status can confer an immense burden of stress involving economic insecurity, uncertainty about future stability, and feelings of exclusion or marginalization due to occupational or economic status (Marmot, 2020; Marmot et al., 1991; Stringhini et al., 2017). It has been observed that low education and occupational attainment increase risk of AD (Räihä et al., 1998; Stern et al., 1994). In particular, low socioeconomic status in childhood increases an individual’s risk of AD by 1.8 times. This increase in risk is even greater if the individual also has the apolipoprotein (APOE) E4 allele, a variant associated with late onset AD. People with this gene variant who grew up in a low socioeconomic household were 2.35 times more likely to develop AD than carriers who grew up in higher socioeconomic status households, and 7.44 times more likely than non-carriers who grew up in higher socioeconomic status households (Moceri et al., 2001). Since this effect is greater than additive, it suggests an interaction between genes and environment (Moceri et al., 2001). Some reports suggest that APOE function may influence circulating glucocorticoid levels, as carriers of the APOE E4 allele tend to have higher cortisol concentrations in the cerebrospinal fluid, irrespective of AD diagnosis (Peskind et al., 2001). This report, along with several others, further observes that stress can accelerate the age of onset in familial AD (Mejfa et al., 2003). High job strain (high demands combined with low control) increases risk for developing AD later in life, but is not affected by APOE genotype (Wang et al., 2012).

Additionally, stress later in life can also precipitate transition to late onset AD. Mild cognitive impairment (MCI) is sometimes classified as preclinical or early stage dementia (Morris et al., 2001). Occurrence of highly stressful events is associated with accelerated cognitive decline in MCI patients resulting in full dementia over a 2-3-year period (Peavy et al., 2012). In these patients, the level of salivary cortisol, one of the classical endocrine markers of chronic stress, was not correlated with disease progression (Peavy et al., 2012). However, AD patients frequently exhibit elevated cortisol levels (salivary, plasma, cerebrospinal fluid) compared with cognitively normal individuals (Vyas et al., 2016) and higher cortisol levels are associated with accelerated progression and increased severity of dementia (Csernansky et al., 2006; Davis et al., 1986). This suggests a differential effect of cortisol depending on the extent of neuropathology and underlines the complexity of the relationship between neuroendocrine pathways implicated in stress and AD.

Though it will not be the focus of this review, it is notable that stress has also been linked to a number of other neurodegenerative disorders, including vascular dementia (Gerritsen et al., 2017) Parkinson’s Disease (Hemmerle et al., 2012) and Huntington’s Disease (Mo et al., 2014). This indicates that stress can more broadly promote degeneration in multiple neural systems.

Insights into stress and neurodegenerative pathology from rodent research

Studies in animals mirror the stress-induced association and exacerbation of AD-related pathology observed in humans (Carroll et al., 2011). They furthermore contribute insights into the molecular mechanisms underlying this association. Evidence exists for interactions between stress and neuropathology in animal models that carry human AD-associated mutations, as well as molecular and behavioral changes in wildtype animals. Although age-related cognitive decline is common across species, no animals aside from humans (and arguably non-human primates (Kalinin et al., 2013; Latimer et al., 2019; Uchihara et al., 2016)) spontaneously develop AD (Gallagher and Nicolle, 1993; Herndon et al., 1997). Though wildtype rodents do not develop AD, studying neural alterations associated with cognitive decline in wildtype rodents can help to identify the preclinical molecular changes that facilitate or appear prior to plaques and tangles. These earlier stages of pathology are key in understanding how and why AD develops and may be the point at which intervention is possible. A key turning point in AD research was the development of transgenic rodent models of AD that carry human copies of genes mutations known to cause familial AD in humans (Games et al., 1995; Hsiao et al., 1996). These AD mouse models have allowed profound insights into the pathophysiology of tau and Aβ inclusions and serve central roles in preclinical testing. Thus, much has been learned from studies in rodents, both wildtype and transgenic, about the interaction between stress and AD-related outcomes.

Stress effects on neurofibrillary tangles and tau

NFTs, one of the hallmarks of AD, are comprised of aggregates of the microtubule binding protein tau. The incidence of NFTs is correlated with cognitive deficits and neuronal loss in AD (Arriagada et al., 1992). Tau is abundant in neurons, where it binds and stabilizes microtubules in the cytoskeleton, but is less common in other cell types. Tau structure, encoded by the MAPT gene, is largely conserved among mammals, with differences largely confined to the N-terminal domain (Hernández et al., 2019). Mouse tau specifically differs from human tau in its lack of 11 amino acid residues in the N-terminal region. Mouse tau also exists primarily in only 3 isoforms, while in humans it occurs in 6. All of these isoforms are generated by alternative splicing and have a number of potential phosphorylation sites, which when occupied disrupt the protein’s capacity for microtubule binding and stabilization (Mandelkow and Mandelkow, 2012). Additionally, hyperphosphorylated tau can self-aggregate to form insoluble paired helical filaments which can then aggregate further to form NFTs (Alonso et al., 1996; Gustke et al., 1992). Although NFTs appear later in disease progression, and correlate with disease stage, many researchers now believe that soluble, non-aggregated tau can also drive cognitive decline (Kopeikina et al., 2012). Thus, the abundance of phosphorylated tau, both aggregated and not, appears to play an important role in AD pathogenesis.

In wildtype rodents, it is well established that stress can increase phosphorylation of tau (Table 1). Over 20 years ago, it was first shown that acute cold water stress exposure increases tau phosphorylation in rat and mouse brain (Korneyev et al., 1995). This result has since been replicated in numerous experiments (Feng et al., 2005; Korneyev, 1998; Okawa et al., 2003) and acute immobilization stress and restraint stress have also been demonstrated to increase tau phosphorylation (Filipcik et al., 2012; Kvetnansky et al., 2016). Acute stress-induced tau phosphorylation extends to many brain areas, including not only subregions of the hippocampus, but also frontal cortex, nucleus basalis of Meynert and the locus coeruleus (Filipcik et al., 2012).

Table 1:

Rodent studies on impact of stress on AD-related markers tau and Aβ.

Stress paradigm	Duration	Animal model	Effect on tau pathology	Effect on Aβ pathology	Reference(s)
Acute cold water	(4-10 min)	Wildtype rat	↑ phosphorylation	---	Korneyev et al., 1995 Korneyev, 1998 Okawa et al., 2003 Feng et al., 2005
Acute multimodal stress	5h	3xTg-AD mouse	No change	↑ Aβ, ↑ACE	Baglietto-Vargas et al., 2015
	30 min	Wildtype mouse, CRHR2 KO mouse	↑ phosphorylation (soluble and insoluble) up until 90min after acute stress	---	Rissman et al., 2007
Restraint Stress	3h	Wildtype rat	---	↑ APP, ↑ Aβ	Ray et al., 2011
	chronic 30min/day; 14 days	Wildtype mouse, CRHR2 KO mice	↑ phosphorylation (soluble and insoluble) 24h after final stressor	---	Rissman et al., 2007
	1h/day; 14 days	Wildtype rat	↑ phosphorylation (soluble and insoluble); changes distribution	---	Yan et al., 2010
	2h/day; 16 days	Tg2576 mouse	↑ phosphorylation	↑ Aβ, ↑ plaque deposition	Lee et al., 2009
	2h/day; 7 days or 6h/day; 21 days	Wildtype rat	---	↑ APP	Rosa et al., 2005
	(6h/day; 5 days)	5XFAD mouse	---	↑ Aβ, ↑ APP, ↑ BACE1, ↑ plaque deposition in females but not males	Devi et al., 2010
	(6h/day; 6 days/week; 4 weeks)	PS19 mouse	↑ phosphorylation, ↑ insoluble forms, ↑ aggregation	---	Carroll et al., 2011
	(6h/day; 6 days/week; 4 weeks)	Tg2576 mouse	---	↑ ACE	Carroll et al., 2011
Chronic unpredictable mild stress	(3 weeks)	Wildtype rat	↑ phosphorylation	---	Yang et al., 2014
	(4 weeks)	Aged wildtype rat (14-15 months old)	↑ phosphorylation	⇑↑ ACE	AbdAlla et al., 2015 Sotiropoulos et al., 2011
	(6 weeks)	Wildtype rat	↑ phosphorylation	↑ Aβ40, ↑ BACE1	Briones et al., 2012
Elevated open platform	10-20 days	Wildtype rat	---	↑ APP	Sayer et al., 2008
Chronic isolation	3 months	Tg2576 mouse	---	↑ Aβ, ↑ plaque deposition, earlier age of onset for plaques	Kang et al., 2007
	5-8 months	Tg2576 mouse	---	↑ Aβ, ↑ plaque deposition	Dong et al., 2004
Chronic mild social stress (random cage composition)	6h/day; 2-3 days / week; 6 weeks	3xTg-AD mouse	No change	↑ Aβ	Rothman et al., 2012
Maternal separation	3h/day; PND 2-21	Wildtype rat	↑ phosphorylation	↑ Aβ,↑ Aβ40, ↑ Aβ42, ↑ BACE1	Solas et al., 2010 Martisova et al., 2013
Postnatal reduced bedding material	PND2-9	APP/PS1 mouse	---	↑ Aβ, ↑ plaque deposition	Hoeijimakers et al., 2017
	PND2-P9	APP-V717I x Tau-P301L mouse	No change	↑ Aβ	Lesuis et al., 2016
Immobilization Stress	10, 30, 60, 90 or 120 minutes	Wildtype mouse	↑ phosphorylation at 30 min; no change after other durations	—	FIlipicik et al., 2012
	10, 30, 60, 90 or 120 minutes	CRH KO mouse	No change	—	FIlipicik et al., 2012
	2h/day; 7 days	Wildtype mouse; CRH KO mouse	↑ phosphorylation	—	Kvetnansky et al., 2016

Additionally, several different rodent models of chronic stress have been shown to increase tau phosphorylation. Chronic restraint stress has been reported to increase levels of both soluble hyperphosphorylated tau and insoluble tau (Rissman et al., 2007; Yan et al., 2010). It also decreases levels of the microtubule assembly-promoting protein MAP2 (Yan et al., 2010). This decrease in MAP2 may reflect additional neuronal destabilization. Chronic restraint stress additionally changes the distribution of tau, with more tau in areas of the hippocampus that also exhibit decreased MAP2 expression (Yan et al., 2010). Three weeks of chronic unpredictable mild stress (CUMS) in young adult rats is sufficient to increase tau phosphorylation (Yang et al., 2014). Furthermore, vulnerable rats that develop anhedonia in the sucrose preference test, indicative of a depression-like state, after 6 weeks of CUMS, have increased tau hyperphosphorylation as well as increased levels of Aβ₄₀ (Briones et al., 2012). 4 weeks of CUMS in aged rats (15 months), in addition to increasing tau hyperphosphorylation, also upregulates angiotensin converting enzyme (ACE) and downregulates structural proteins, such as MAP2 and synuclein-gamma (AbdAlla et al., 2015).

Stress-induced tau hyperphosphorylation appears to occur very rapidly, with just 4 minutes of cold water exposure sufficient to induce an effect (Korneyev et al., 1995). The reversibility of this effect seems to scale with the chronicity of stressor exposure. With a single stressor exposure, phosphorylated tau levels return to amounts indistinguishable from those of unstressed controls 2 hours after cessation of immobilization stress (Filipcik et al., 2012; Rissman et al., 2007), Reversal of tau hyperphosphorylation remains possible 24-hours post final stressor exposure following 6 (Kvetnansky et al., 2016), but not 14 days (Rissman et al., 2007) of daily stressor exposure. In keeping with this pattern, after 6 weeks of CUMS, increased tau phosphorylation is still detectable 7 days after the final stress exposure (Briones et al., 2012). Thus, it appears that recovery from stress-induced tau hyperphosphorylation is possible but becomes increasingly impaired the longer stress exposure continues.

Although the evidence is compelling that chronic stress can cause changes in the brain and behavior in wildtype rodents associated with neurodegeneration, the limitation remains that these animals have not been shown under any condition to develop AD. However, by employing the use of transgenic humanized mouse models of AD, it is possible to study the interaction between stress and the proteins most closely tied to AD.

Unlike work in transgenic lines that model Aβ pathophysiology, few stress studies have been performed in mouse lines that model the tauopathy component of AD. However, it has been demonstrated that in mice that carry mutant copies of the tau gene that mimic those found in human disease, chronic stress further exacerbates tau-related pathologies. In the PS19 mouse model, which is characterized by hyperphosphorylated tau that forms neurofibrillary tangles, 4 weeks of restraint stress promotes tau hyperphosphorylation, accumulation of insoluble tau, and worsened fear memory (Carroll et al., 2011). Additionally, chronic restraint stress in the Tg2576 mouse line, which models Aβ pathology, also increases phosphorylation of tau (Lee et al., 2009). Another study treated 3xTg-AD mice, which harbor knock-in mutations of presenilin 1, APP and tau, with the synthetic glucocorticoid dexamethasone for 7 days. This treatment increased mislocalization of tau to the somatodendritic region of neurons in the hippocampus, cortex and amygdala, but did not increase levels of phosphorylation (Green et al., 2006). However, this same treatment regimen in rTg4510 mice, which express the same tau and presenilin 1 mutations as the 3xTg-AD model, but not the APP mutation, did not alter levels of tau, suggesting that elevated glucocorticoid signaling may not play a direct role in this tau mislocalization; rather it occurs downstream of an interaction between the glucocorticoids and APP (Green et al., 2006).

Overall, the evidence is compelling that stress can both acutely and chronically alter tau expression and hyperphosphorylation. These changes are highly relevant to AD, as decreasing levels of tau, even without disrupting NFTs can improve cognitive function (Oddo et al., 2006).

Stress effects on Aβ plaques

The other main marker of AD is the accumulation of Aβ plaques. Much research focus has been put into their ontology, based on the “amyloid hypothesis”, which postulates that the buildup of Aβ, particularly the longer, more hydrophobic Aβ₄₂ isoform, is the root cause of AD (Selkoe and Hardy, 2016). The Aβ peptide is derived from the longer amyloid precursor protein (APP) through two cleavage events, the first executed by BACE1 (β site APP cleavage enzyme 1), and the second by a γ secretase complex. APP can alternatively be processed by sequential cleavage by α secretase and γ secretase, which does not produce Aβ (O’brien and Wong, 2011). Familial early onset forms of AD are caused by mutations in the precursor protein APP, or in the γ-secretase components, presenilin 1 and presenilin 2 (PSEN1, PSEN2), which result in an increased ratio of Aβ₄₂ to Aβ₄₀(Scheuner et al., 1996). Given this causal relationship between Aβ-related genetic mutations and AD, the amyloid hypothesis has been compelling and widely accepted. The theory posits that an increase in Aβ₄₂ results in oligomerization and the formation of Aβ plaques. The plaques accumulate, impairing synaptic activity and causing reactive inflammation in surrounding cells and setting off a cascade of reactions, including tau tangle formation, synapse loss and neuron death, which together lead to the cognitive deficits that constitute dementia (for review, see Selkoe and Hardy, 2016). However, recent findings have called into question the precise role of plaques in AD. In particular, the failure of clinical trials targeting Aβ plaques (Egan et al., 2018; Mullane and Williams, 2013) and the observation that there is a very weak correlation between the abundance of Aβ plaques and degree of cognitive impairment (Nelson et al., 2012) have put the original amyloid hypothesis into question, underlining the need for further research. Regardless of their precise role in the genesis of AD, it is clear that Aβ cleavage and buildup are core processes in the disease. Additionally, there is a compelling body of research on the effect of stress on Aβ (Table 1) which will be discussed below.

Similar to its effect on tau phosphorylation, stress in wildtype rats and mice increases levels of APP mRNA and protein and Aβ (Ray et al., 2011; Rosa et al., 2005; Sayer et al., 2008; Solas et al., 2010). Daily maternal separation during the first 3 weeks of life increases levels of Aβ₄₀ and Aβ₄₂ in adulthood and increases expression of the BACE1 (Martisova et al., 2013). It also results in decreased hippocampal cell number, lower brain-derived neurotrophic factor (BDNF) levels and diminished synaptophysin and postsynaptic density-95 (PSD-95) (Martisova et al., 2013). 4 weeks of CUMS in aged (15 month old) rats increases levels of ACE and 6 weeks of CUMS in young adult rats increases levels of Aβ₄₀(AbdAlla et al., 2015; Briones et al., 2012).

Chronic stress also aggravates amyloid pathology in numerous transgenic mouse models of AD. In mice that express APP with a mutation known to cause familial early onset AD (Tg2576; Hsiao et al., 1996), several stress procedures intensify pathology. 3 months of social isolation stress not only elevates levels of Aβ, but also increases plaque deposition and worsens cognitive impairment (Kang et al., 2007; Lee et al., 2009). The extent of the impact of stress on AD-related pathology depends on, among other factors, the duration of stress exposure. In Tg2576 mice, elevated Aβ is measurable after just 1 hour of restraint stress, and peaks 10 hours after cessation of the restraint (Kang et al., 2007). Long-term social isolation stress (from weaning onward) anticipated age of onset for Aβ plaque deposition from 9 months to 6 months of age. It also reduced neurogenesis in hippocampus, which was associated with impaired contextual memory (Dong et al., 2004).

In the 3xTg-AD mouse model, which carries 3 genes associated with familial AD in humans (to the APP, MAPT and PSEN1 genes), a single instance of 5 hour intense “modern life-like” stress (restraint stress on rotating platform accompanied by loud noise) increased Aß levels and reduced the number of synapse-bearing dendritic spines (Baglietto-Vargas et al., 2015). 6 weeks of mild social stress (randomized cage composition for 6 hours, 2-3 times weekly) in the same mouse line increased both Aβ and decreases BDNF (Rothman et al., 2012). 7 day treatment with the glucocorticoid receptor (GR) agonist dexamethasone (5mg/kg) in this model increased levels of APP, Aβ₄₀ and Aβ₄₂ and increased activity of BACE (Green et al., 2006).

Similarly, 5 days of restraint stress increased levels of Aβ₄₂ and plaque deposition in the hippocampus of female but not male 5XFAD mice, which carry 5 familial AD mutations in APP and PSEN1 genes (Devi et al., 2010; Oakley et al., 2006). This is an interesting finding given that women are much more likely to develop AD than men (Alzheimer’s Association, 2014). However after 4 weeks of restraint stress, both male and female 5XFAD mice had increased levels of Aβ along with impaired spatial and fear memory, compared with unstressed littermates (Carroll et al., 2011). Together these studies indicate that females harboring these genetic mutations may be more vulnerable to stress than males, but that long term chronic stress will ultimately exacerbate pathology in both sexes.

In most of the experiments discussed above, stress protocols were applied during the young adult period. Early-life stress can also result in worsened Aβ pathology later in life. Postnatal stress increases Aβ plaque burden and inflammatory markers and compromises survival later in life in APP/PS1 mice (Hoeijmakers et al., 2017; Lesuis et al., 2016). These results mirror the epidemiological findings in humans discussed earlier, which demonstrate that early life stress can particularly increase likelihood of developing AD.

Stress-induced changes to synapses and neuronal survival

Although NFTs and Aβ plaques are regarded as the primary markers of AD pathology, synapse and neuron loss are thought to be the net result of their presence and directly cause the cognitive and memory impairments associated with AD. Indeed, studies show that neuronal degeneration is a better predictor of dementia than tau or Aβ inclusions (Price et al., 2001). Synaptic loss has the strongest correlation with the staging of dementia and with early stage AD (Nelson et al., 2012; Scheff et al., 2007). It has been proposed that synapse loss underlies the cognitive impairments in early stage AD and that this disruption of plasticity impairs neuronal viability, resulting in the neuron loss that characterizes later stages of the disease (Shankar and Walsh, 2009). Numerous lines of research indicate that chronic stress can directly lead to synapse loss and impaired neurogenesis. In this way, stress may promote the advent and accelerate the progression of AD.

The hippocampus, the origin point of AD pathology, is particularly stress-sensitive possibly because of its robust expression of GR and CRH receptor 1 (CRHR1) receptors (Braak et al., 2006; Braak and Braak, 1991). It is a major target of stress-associated hormones (McEwen et al., 1968), and stress-induced degeneration in the hippocampus is well documented (McEwen, 1998; McEwen et al., 1968). In particular, chronic stress shrinks apical dendrites in area CA3 of the hippocampus. Numerous studies have demonstrated that chronic restraint stress induces atrophy of apical dendrites (shorter length, fewer branch points) in hippocampal CA3 pyramidal neurons (Magariños and McEwen, 1995; Watanabe et al., 1992). While much of this work has been done in rats, it has been shown that this effect is preserved across species (Bartolomucci et al., 2002; Czeh et al., 2001; Magariños et al., 1996). Acute, as well as chronic stress can alter CA3 dendrites: a single exposure to a 5-hour stressor reduces CA3 dendritic spine density and long-term potentiation. Moreover, these effects correlate with stress-induced deficits in selected learning and memory processes (A. Bartolomucci et al., 2002; Chen et al., 2010).

The vast majority of dendritic spines have synapses, and thus their loss has profound functional consequences (Arellano et al., 2007; Harris, 1999). Studies that examine specific synaptic markers find similar results to those that examine dendritic structure. Both chronic stress and GC exposure reduce the number of synapses between mossy fibers from dentate gyrus granule cells onto CA3 pyramidal cells (Sousa et al., 2000). Additionally, multimodal stress – restraint combined with loud music, has a greater effect than either manipulation alone as measured by memory performance and reduction in synapse number in CA3 (Maras et al., 2014).

Prolonged exposure to either chronic psychosocial stress or GCs does not impact cell number but does reduce CA3 and hippocampal volume, which may be accounted for through a loss of synapses (Czeh et al., 2001; Tata et al., 2006). In humans, hippocampal volume begins to decrease concomitantly with the onset of memory decline (Reiman et al., 1998), suggesting that this is a highly relevant parameter.

While abundant evidence supports stress-induced deterioration in area CA3, there are fewer reports of changes in other areas of the hippocampus. Some studies have found that stress or GC induced dendritic decline in CA1 pyramidal neurons (Sousa et al., 2000), but others have not (Alfarez et al., 2008; Woolley et al., 1990). Interestingly, one study did find that while 3 weeks of unpredictable stress did not affect CA1 dendritic length, if brain slices from these animals were subsequently exposed to elevated GC levels, there was a significant reduction in apical dendritic length, compared with slices from control animals (Alfarez et al., 2008). This suggests that some sensitization may occur in CA1 neurons in response to chronic stress that renders them more sensitive to future stress. Another study found that multimodal stress but not a single stressor (either restraint or loud noise) can reduce synapse number in CA1. It also induced a reduction in connectivity (as measured by c-fos) with the septum and thalamus and an increase in connectivity with the amygdala and the bed nucleus of the stria terminalis (Maras et al., 2014). These results suggest that while CA1 may be less vulnerable to a single bout of chronic stress, repeated stressor exposure and/or facing multiple stressors simultaneously may lead to CA1 deterioration. This is particularly pertinent to the role of stress in neurodegeneration, as AD pathology particularly affects CA1 (Masurkar, 2018; Padurariu et al., 2012).

Chronic stress has profound influence on other brain regions, in addition to the hippocampus. The prefrontal cortex (PFC) is involved in the processing of emotional stimuli as well as complex cognitive tasks (Cerqueira et al., 2007). Prenatal stress impairs connectivity between the hippocampus and PFC and disrupts synaptic plasticity within the PFC (Mychasiuk et al., 2012), which is similar to the reduced connectivity observed in humans after stressful experiences (Admon et al., 2013). Furthermore, chronic subordination stress decreases serotonin turnover in the PFC in susceptible mice (Bartolomucci et al., 2010). As in the hippocampus, chronic stress (both chronic social defeat stress or chronic variable stress) reduces dendritic spine density in the prefrontal cortex in a CRHR1-dependent manner (Chen et al., 2008; Radley et al., 2013; Shu and Xu, 2017). In wildtype rodents, chronic stress promotes Aβ accumulation and tau phosphorylation in the prefrontal cortex (PFC), as well as the hippocampus (Ray et al., 2011; Sotiropoulos et al., 2011; Yang et al., 2014).

Stress also inhibits adult neurogenesis and impairs survival of newly born neurons in the dentate gyrus (Czeh et al., 2001; Krugers et al., 2010) A decline in neurogenesis occurs with aging and is thought to contribute to age-related cognitive impairment by reducing capacity for plasticity (Apple et al., 2017). Thus, by impairing the formation and survival of new neurons, stress may further promote the decline in cognitive performance associated with AD. Acute stressor exposure reduces proliferation and increases the number of apoptotic granule cells in the dentate gyrus (Gould et al., 1998; Heine et al., 2005) while chronic stressor exposure additionally increases the expression of the cell cycle inhibitor p27Kip1 in the dentate gyrus (Heine et al., 2004). GCs alone can impair neurogenesis by inhibiting cell proliferation and survival (Wong and Herbert, 2006). This is another mechanism by which chronic stress can aggravate age-related decline in brain having significant implication for neurodegenerative disorders.

Stress-induced promotion of neuroinflammation

Neuroinflammation is a common pathological feature in AD and other neurodegenerative diseases (Cribbs et al., 2012; Crotti and Glass, 2015; Griffin et al., 1989; Herrero et al., 2015; Sudduth et al., 2013). In response to insults, microglia and astrocytes secrete inflammatory factors which facilitate immune cell recruitment and clearance of infection and tissue damage. In AD, it is believed that neuroinflammation results as a reaction to accumulation of Aβ, and is one of the earliest features of the disease. Microglia respond to Aβ by phagocytosing the abnormal protein and secreting proinflammatory cytokines to recruit additional microglia to the plaques (Bhaskar et al., 2014; Smith et al., 2012). Astrocytes to a lesser degree can degrade Aβ and also secrete inflammatory mediators in response to Aβ (Carrero et al., 2012).

This immune response appears to be initially beneficial, increasing clearance of Aβ (Chakrabarty et al., 2010; Shaftel et al., 2007). However, prolonged inflammation can be detrimental to neurons and synapses (Leszek et al., 2016) and can ultimately impair microglial phagocytosis of Aβ (Hickman et al., 2008), while simultaneously promoting Aβ production (Chong, 1997; Liaoi et al., 2004) and tau hyperphosphorylation (Quintanilla et al., 2004). Overall, neuroinflammation has a complex and multifaceted role in AD. While it can have beneficial effects by promoting Aβ clearance, it can also directly promote other aspects of AD neuropathology, increasing both Aβ burden and tau hyperphosphorylation and impairing synapse maintenance and neuronal survival.

Increased inflammation is a common phenotype of normal brain aging across species (Frank et al., 2006), and may represent one mechanism by which age increases risk of AD. Additionally, it is well established that chronic or traumatic stressors increase neuroinflammation. Thus, by promoting inflammation, stress may facilitate the development of AD in a similar way to natural aging. A number of 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, which is particularly impaired in AD (Bian et al., 2012; Kojo et al., 2010). This increase in microglial activation has been demonstrated to lead to the secretion of cytokines and chemokines, which facilitate development of anxiety-like behavior (Tynan et al., 2010; Wohleb et al., 2013, 2012, 2011). 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 (Heidt et al., 2014; Powell et al., 2013; Wohleb et al., 2011).

Additionally, in recent studies it was demonstrated that activation of the C3aR1 receptor by TLQP-21 or C3a (Sahu et al., 2019), which was previously linked with to both the stress response (Razzoli et al., 2012) and microglial function (Doolen et al., 2017), increases microglial phagocytosis of Aβ (Cho et al., 2020; El Gaamouch et al., 2020) and improves amyloid pathology in the 5XFAD mouse model of AD (El Gaamouch et al., 2020). However, activation of C3aR1 has also been shown to promote early synapse loss (Hong et al., 2016), cognitive decline (Shi et al., 2017) and tau pathology (Litvinchuk et al., 2018). This biphasic contribution is consistent with the dual role that microglial activation plays in AD pathology, and the C3aR1-mediated signaling pathway may represent an important mediator of stress-induced microglial activation and inflammation in AD.

Furthermore, prior stressor exposure can potentiate the CNS inflammatory response to subsequent immune challenges (Frank et al., 2007; Giovanoli et al., 2013; Wohleb et al., 2012; Yoo et al., 2011). This mechanism could be particularly relevant to the contribution of early life and middle life stress to AD risk in old age.

Potential Mechanisms of stress-induced exacerbation of neurodegeneration and AD

There are multiple potential mediators of stress-induced effects on neurodenegeration and AD. At this point, the best researched mediators are two of the primary HPA axis hormones: CRH and GCs.

Glucocorticoids

GCs are the main output of the HPA axis. Under normal circumstances, GCs exert negative feedback on the hypothalamus and pituitary, returning GC levels to baseline, but in some instances of repeated stress exposure, GC hypersecretion can persist. The seminal Glucocorticoid Cascade Hypothesis suggested that age-related GR loss results in GC hypersecretion, and that subsequent stress-induced GCs both directly damage hippocampal neurons and impair their ability to withstand any additional damage or insult, resulting in accelerated neuron degeneration and death (Sapolsky et al., 1986). It furthermore proposes that this is the major mechanism by which hippocampal degeneration occurs with age and promotes AD.

An extensive body of research indicates that aging increases basal HPA axis function and stress responsiveness (Sapolsky, 1992). Aged rats have fewer GR receptors in the hippocampus and to a smaller extent, in the amygdala (Sapolsky et al., 1983). Given that much of the GC negative feedback on the paraventricular nucleus is mediated by neurons in the hippocampus, this may be responsible for the impaired termination of the stress response observed in aged individuals. In rats, 3 months of corticosterone injections that replicate the high physiological range resulted in depleted GR in the hippocampus, which did not change after 4 months of recovery. This also resulted in a reduced number of cells in area CA3 and the appearance of “dark microglia” (Sapolsky et al., 1985). Exogenous glucocorticoids also potentiate the capacity for Aβ infusion to promote tau hyperphosphorylation and intracellular accumulation (Sotiropoulos et al., 2011). In humans, several case studies indicate that high dose corticosteroids can induce long-lasting impairments in learning and memory, similar to those observed in patients with dementia (Arndt et al., 2004; Sacks and Shulman, 2005). Similarly, patients with Cushing’s Syndrome frequently manifest with psychiatric and cognitive dysfunction termed “steroid dementia” (Bernini and Tricò, 2016). However, while some aspects of the glucocorticoid cascade hypothesis, have been substantiated, it has become clear that GC action alone cannot account for aging and AD-related hippocampal degeneration.

Further experiments have demonstrated that the stress/GC-induced decrease in hippocampal volume is now primarily attributed to dendritic loss rather than neural death (Czeh et al., 2001; Tata et al., 2006) and several stress-induced effects on AD pathology have been shown to be glucocorticoid independent (Chen et al., 2010, 2008; Radley et al., 2013; Shu and Xu, 2017).

Corticotropin Releasing Hormone.

CRH has also been shown to be a primary mediator of stress effects on several neurodegenerative phenomena. Direct injection of CRH into the brain results in increased Aβ in the interstitial fluid (Kang et al., 2007). Repeated administration of a CRHR1 agonist into the basolateral amygdala is sufficient to increase APP and Aβ in prefrontal cortex (Ray et al., 2011).

Additionally, pharmacological blockage of CRHR1 receptors prevents an increase in Aβ induced by acute restraint stress. 2 weeks of restraint stress in adrenalectomized wildtype mice increases levels of phosphorylated tau but can be prevented by administering a CRHR1 antagonist (Rissman et al., 2007). Phosphorylated tau is also blocked in CRHR1 knock-out mice but not in CRHR2 knock-out mice, suggesting a specific role for CRHR1 mediated signaling (Rissman et al., 2007). In this case, it is possible that GCs and CRH have redundant effects on tau phosphorylation. Exacerbations in pathology produced by restraint stress in a mouse model that expresses mutant human tau can similarly be prevented by a CRHR1 antagonist (Carroll et al., 2011).

It is clear that both GCs and CRH are important mediators of both the stress response, and stress-effects on neurodegeneration. Less clear however is the precise roles they play. A single exposure to a 5-hour stressor reduces CA3 dendritic spine density and long-term potentiation and these effects correlate with stress-induced memory deficits (Chen et al., 2010). Both GR and CRHR1 are implicated in this effect. Chronic exposure to GCs similarly reduces CA3 apical dendritic complexity (Woolley et al., 1990). However, direct infusion of a CRHR1 antagonist into the brain prevented the spine loss and memory impairments caused by a 5-hour stress exposure (Chen et al., 2010). Thus, both mechanisms may be at play but with different timescales, with CRHR1 mediating acute spine loss and GR mediating chronic stress induced dendritic shortening.

Catecholamines.

Though it has been less well studied in the context of neurodegeneration, the actions of the SAM axis and sympathetic nervous system in AD pathogenesis must be considered. Epinephrine and norepinephrine, the major effectors of these systems respectively, do not cross the blood brain barrier, and thus circulating levels released from the adrenal glands and sympathetic nerves do not directly act on the CNS (Weil-Malherbe et al., 1959). However, the catecholamines-mediated actions, such as increasing blood pressure may indirectly contribute to the development of AD. Hypertension, particularly in midlife, is a risk factor for AD (Luchsinger et al., 2005) and antihypertensive medications have been associated lower incidence of AD diagnosis and better cognitive function (Hajjar et al., 2005; Hoffman et al., 2009; Khachaturian, 2006). Additionally, norepinephrine released by sympathetic nerves promotes the production and release of myeloid cells, which are trafficked throughout the body and release pro-inflammatory cytokines (Dhabhar et al., 2012; Hanke et al., 2012; Heidt et al., 2014). This can lead directly to increased neuroinflammation, which is associated with AD, and also to the promotion of other diseases such as atherosclerosis, which increases risk for AD (Kalback et al., 2004).

Norepinephrine is also secreted by catecholaminergic neurons in the brain, whose activity is particularly increased in response to acute stress and facilitate a number of anxiety-like behavioral responses (for review, see Morilak et al., 2005). The locus coeruleus (LC) is the main source of norepinephrine in the brain, and has widespread projections to the cortex, hippocampus and amygdala, among many other brain regions. LC neurons degenerate early in AD, and evidence indicates that this promotes further disease progression (Grudzien et al., 2007; Kelly et al., 2017; Zarow et al., 2003). However, the role of noradrenergic signaling in AD pathogenesis is complicated and seemingly contradictory. Selective ablation of noradrenergic neurons increases Aβ deposition in the APP23 model (Heneka et al., 2006), the APP717 model (Kalinin et al., 2007) and the APP/PS1 model (Jardanhazi-Kurutz et al., 2010) and increases levels of phosphorylated tau in female APP/SL mice (Oikawa et al., 2010).The β2 adrenergic receptor (AR) agonist terbutaline prevents LTP inhibition induced by exogenous application of Aβ to rat hippocampal sections (Wang et al., 2009) and an antagonism of β2AR with ICI 118,551 in the 3xTg model was found to increase Aβ levels and further impaired cognitive function (Branca et al., 2014). The above evidence suggests that noradrenergic signaling may play a protective role in AD, but is contradicted by several other studies, which indicate that it may exacerbate AD pathogenesis. Stimulating α₂ adrenergic receptor disrupts APP interaction with Golgi sorting receptors, altering the outcome of its proteolytic processing to promote Aβ generation and subsequent neuropathology (Chen et al., 2014). In APP/PS1 mice, administration of a selective α2 receptor antagonist ameliorates AD-related cognitive deficits (Chen et al., 2014; Scullion et al., 2011). Activation of β₂AR enhances γ secretase activity, and thus Aβ production in both transgenic AD mouse models and wildtype mice (Ni et al., 2006; Yu et al., 2010). Antagonism of either or both β₁ and β₂ARs can reduce Aβ production in amyloidosis mouse models of AD (Ni et al., 2006; Wang et al., 2013). Blockade of β₂AR receptors also reduces stress-induced Aβ generation in nontransgenic mice (Yu et al., 2010). Inhibition of both β₁ and β₂ARs with propanalol improves cognitive function both in the amyloidosis Tg2576 mouse model and in the accelerated aging SAMP8 model (Dobarro et al., 2013a, 2013b). Thus, it is clear that norepinephrine and the neurons in the LC that produce it, have an intimate association with AD pathogenesis. While ablation of neurons in the LC clearly is detrimental to health and cognition, the outcome of noradrenergic signaling may be highly dependent on contextual factors including stage of disease progression, the receptors it interacts with and the outcome being measured. The specificity of noradrenergic signaling may be key, with LC neurons participating in circuits that cannot be recapitulated by CNS-wide α or β adrenergic receptor stimulation. Glutamate. Acute restraint stress elevates glutamate levels in the hippocampus in a glucocorticoid-dependent manner (Lowy et al., 1993). GC induced glutamate secretion likely plays a role in stress-induced dendritic remodeling, as blocking NMDA receptors blocks this effect (Magariños and McEwen, 1995). Excess glutamatergic activity is also associated with neuron loss following trauma and is implicated in neurodegeneration and aging. Aged rats treated with riluzole, which increases glutamate uptake through glial transporters, were protected against age-related cognitive decline and dendritic spine loss (Pereira et al., 2014). Riluzole also prevented many age-related hippocampal gene expression changes. In particular, it prevented gene expression changes associated with AD in humans (Pereira et al., 2017). This suggests that excess glutamate signaling is a key modulator of stress or GC induced dendritic retraction and cognitive decline. Cellular senescence. A core tenant of the stress concept originally described by Hans Selye was that stress, like aging, was the “results of life’s wear and tear” (Selye, 1959). The field’s understanding of stress has been refined since then, but the link between stress and aging has endured. In addition to AD, stress is observed to increase risk for a number of conditions and diseases associated with aging, including hypertension, atherosclerosis, diabetes and dementia (Dimsdale, 2008; Murdock et al., 2016; Razzoli et al., 2018; Yuede et al., 2018; Zhang et al., 2020). Given its contribution to a number of aging-related diseases as well as to biological aging itself, is it possible that stress promotes these diseases at least in part by impacting a fundamental aging mechanism?

One potential mediator of this interaction has gained momentum recently: the process known as cellular senescence. Cellular senescence, the phenomenon by which cells cease to divide and undergo other distinctive phenotypic changes, reflects a deterioration of function with age on a cellular level, and is believed to contribute to aging at an organismal level (Hayflick and Moorhead, 1961; McHugh and Gil, 2018). After entering senescence, cells do not necessarily die; on the contrary they become resistant to apoptosis (Campisi and d’Adda di Fagagna, 2007; Childs et al., 2014; Gorgoulis et al., 2019; Kirkland et al., 2017). Though viable, senescent cells undergo major gene expression changes, secreting a number of pro-inflammatory factors collectively termed the senescence-associated secretory phenotype (SASP) (Coppé et al., 2010). In this way, senescent cells can alter their tissue microenvironment and detrimentally affect the function of neighboring cells. Thus, just a few senescent cells could alter the activity of a far greater number of other cells either increasing the number of senescent cells or affecting the functions of neighboring cells. Increasing evidence suggests that senescent cells, which accumulate throughout the lifetime, may be major drivers of functional decline associated with aging. Thus far, senescent cells have been shown to actively drive a number of diseases, including atherosclerosis (Childs et al., 2016), cancer (Campisi and d’Adda di Fagagna, 2007) and AD (Bussian et al., 2018; Musi et al., 2018; Zhang et al., 2019). Specifically, NFT bearing neurons extracted from post-mortem human tissue and the rTg4510 tauopathy mouse model tissues have transcriptomic profiles consistent with that of senescent cells (Musi et al., 2018). Oligodendrocyte precursor cells associated with Aβ plaques in brain tissue from AD patients or APP/PS1 mice also express senescence markers (Zhang et al., 2019). Additionally, pharmacological or pharmacogenetic elimination of senescent cells in two different tauopathy mouse models (rTg4510 and PS19) reduced NFT burden, attenuated brain volume loss and improved cognition and memory (Bussian et al., 2018; Musi et al., 2018). Pharmacological removal of senescent cells in a mouse model characterized by Aβ deposits also lessened Aβ plaque load and improved cognition (Zhang et al., 2019). These studies provide strong evidence that senescent cells play an important, but as yet understudied role in AD.

Evidence in humans suggests that chronic stress might accelerate aging by promoting cellular senescence. Women who report higher levels of stress have shorter telomeres in circulating immune cells and less telomerase activity than those with low reported levels of stress (Epel et al., 2004). Shortened telomeres are the classical trigger for cellular senescence (Hayflick and Moorhead, 1961), and higher stress levels are also associated with greater oxidative stress – an independent senescence promoter (Aschbacher et al., 2013). Low socioeconomic status and life stress have been associated with lower telomere length in some (Cherkas et al., 2006) but not all studies (Adams et al., 2007; Rentscher et al., 2019), and increased p16^INK4 expression (Rentscher et al., 2019). Finally, a recent study demonstrated that lifelong psychosocial stress in mice shortened lifespan and increased expression of senescence markers such as p16 and p53 in several organs in subordinate individuals (Razzoli et al., 2018). These findings are particularly relevant in the context of human literature demonstrating that shorter telomeres increases risk for AD (Zhan et al., 2015). Together, these findings indicate that senescent cells may be key mediators in stress-induced promotion of AD, possibly through their secretion of proinflammatory factors and dysregulation of local tissue homeostasis.

Conclusion

Strong evidence indicates that stress contributes to the onset and progression of AD, one of the most common and devastating diseases associated with aging. Human literature indicates that high levels of stress increase risk for developing AD, and that stress can accelerate the age of onset in individuals genetically predisposed to develop the disease. We have gained important insights into the mechanisms that might underlie this association through rodent research, in which stress also consistently increases AD-related measures. In wildtype rodents, stress promotes expression of molecular markers implicated in the disease, including hyperphosphorylated tau, and APP and Aβ. It also reduces synapse number in key areas related to early stage AD. Stress in rodent models of AD results in exacerbated molecular and behavioral pathologies and can advance the age at which animals begin to show signs of decline. The molecular underpinnings of these effects have yet to be fully understood. Both glucocorticoids and CRH have been shown to play important, and connected roles in producing these AD-related changes (Figure 1).

Figure 1.

Diagram of stress-induced and aging-associated changes in Alzheimer`s disease (AD) neuropathological changes.

An emerging mechanism to consider is the impact of stress on the aging and senescence process per se. Senescent cells can secrete pro-inflammatory factors, which have been shown to exacerbate pathological features of AD. They may be an important mediator of stress-induced neuroinflammation, which is appears to be an active driver of AD pathology. Age is the greatest risk factor for AD, and a long history of research supports the theory that stress impacts and accelerates the aging process. However, the aging aspect of AD is particularly challenging to study mechanistically, given the vast difference in lifespan between humans and rodent models as well as the resource burden of waiting for animals to age naturally before performing experiments. Most transgenic AD rodent models have shorter lifespans than their wildtype counterparts which is perhaps a double-edged sword for understanding the connection between aging and AD. On the one hand, the accelerated timeline towards frailty and death in these transgenic models of AD is an opportunity to study the relationship between AD pathophysiology and aging mechanisms. On the other, these interactions may not reflect mechanisms of normal aging and the dramatic phenotype of these models may obfuscate the roles of other physiological mechanisms. An important task for the future will be to identify if and how stress-induced cellular senescence in the brain and stress-induced AD neuropathology, and to develop interventions to ameliorate the current burden of AD on our society.

Highlights.

• High stress levels increase risk of late-onset AD and accelerates age of onset for familial AD

• Stress causes synaptic loss and neuroinflammation in brain regions affected by AD

• In wildtype rodents, chronic stress promotes accumulation of tau and Aβ

• In mouse models of AD, chronic stress intensifies molecular and cognitive pathology

• Chronic stress increases markers of cellular senescence

• Cellular senescence has been linked to AD

Abbreviations:

ACE

angiotensin converting enzyme

ACTH

Adrenocorticotropin hormone

AD

Alzheimer’s Disease

APOE

Apolipoprotein E

APP

amyloid precursor protein

Aβ

Amyloid β

BACE

β site APP cleavage enzyme

BDNF

brain-derived neurotrophic factor

CRH

Corticotropin releasing hormone

CRHR1

corticotropin releasing hormone receptor 1

CUMS

chronic unpredictable mild stress

GC

Glucocorticoid

GR

Glucocorticoid receptor

HPA

hypothalamic pituitary adrenal axis

MAP2

microtubule assembling protein 2

MAPT

Microtubule associated protein tau

MCI

Mild Cognitive Impairment

NFT

Neurofibrillary tangle

PFC

prefrontal cortex

PSD-95

postsynaptic density protein 95

PSEN

presenilin

SAM

Sympathetic-adrenal medullary