The Interactive Effects of Psychosocial Stress and Diet Composition, on Health in Primates

Abstract

Social disadvantage and diet composition independently impact myriad dimensions of health. They are closely entwined, as social disadvantage often yields poor diet quality, and may interact to fuel differential health outcomes. This paper reviews effects of psychosocial stress and diet composition on health in nonhuman primates and their implications for aging and human health. We examined the effects of social subordination stress and Mediterranean versus Western diet on multiple systems. We report that psychosocial stress and Western diet have independent and additive adverse effects on hypothalamic-pituitary-adrenal and autonomic nervous system reactivity to psychological stressors, brain structure, and ovarian function. Compared to the Mediterranean diet, the Western diet resulted in accelerated aging, nonalcoholic fatty liver disease, insulin resistance, gut microbial changes associated with increased disease risk, neuroinflammation, neuroanatomical perturbations, anxiety, and social isolation. This comprehensive, multisystem investigation lays the foundation for future investigations of the mechanistic underpinnings of psychosocial stress and diet effects on health, and advances the promise of the Mediterranean diet as a therapeutic intervention on psychosocial stress.

1. Introduction

1.1. Psychosocial Stress and Human Health

For the purposes of this review, psychosocial stress is defined as the activation of behavioral and physiological systems by stressors stemming from social interaction. When chronically experienced, psychosocial stress increases the risk of anxiety, depression (Southwick et al., 2005), insomnia (Birch and Vanderheyden, 2022), gastrointestinal disorders (Black et al., 2022), chronic pain (Meade and Garvey, 2022), susceptibility to viral infection (Cohen, 2005), overeating (Bahi and Dreyer, 2023), hypertension, diabetes, obesity (Bergmann et al., 2014; Chandola et al., 2006; Clougherty et al., 2021; Hackett and Steptoe, 2017; Spruill, 2010), some cancers (Hong et al., 2022; Mohan et al., 2022), cardiovascular disease (CVD) (Dar et al., 2019; Osborne et al., 2020; Santosa et al., 2021; Steptoe and Kivimäki, 2012; Vancheri et al., 2022), dementia (Franks et al., 2021; Luo et al., 2022; Wu-Chung et al., 2022), and all-cause mortality (Rutters et al., 2014; Santosa et al., 2021). By far, the strongest and largest connection between stress and health is for CVD, where psychological stress more than doubles the risk of myocardial infarction (Kivimäki and Steptoe, 2018). The risk of several of these disorders is related e.g., psychosocial stress, depression, and anxiety increase the risk of CVD (Cohen et al., 2015), and the gut microbiome can modulate risk of anxiety and depression (Foster and McVey Neufeld, 2013). Psychosocial stress is thought to accelerate biological aging as it is a major cause of aging-related diseases and decreased life expectancy (Razzoli et al., 2023). Many of these effects are mediated through common pathways such as sustained sympathetic nervous system (SNS) arousal, hypothalamic-pituitary–adrenal (HPA) activation, and inflammation (Henein et al., 2022; Peirce and Alviña, 2019; Steptoe et al., 2007) which involve basic cellular processes like mitochondrial function (Picard and McEwen, 2018b). Low socioeconomic status is stressful as evidenced by higher self-reported perceived stress, and high cortisol and catecholamine levels indicating HPA and SNS activation (Cohen et al., 2006b) (Kim and Fredriksen-Goldsen, 2017). The stressful nature of low SES is attributed to relative high rates of crowding, crime, noise pollution, discrimination and poor sleep quality, fewer opportunities for exercise, and poor diet quality (Clougherty et al., 2021; Darmon and Drewnowski, 2015; Nadybal et al., 2020; Yusuf et al., 2020; Zaman et al., 2022). While the profound impact of psychosocial stress on health is widely recognized, there is no population level treatment to reduce stress and associated disease.

1.2. Diet Composition and Human Health: Western versus Mediterranean-like Diets

Diet composition has far-reaching effects on health. We focus on the Western and Mediterranean diets because of the preponderance of data suggesting that western diet consumption has deleterious effects on health, and that Mediterranean diet consumption is associated with beneficial effects on health. As we reviewed this literature, we adopted the diet names/definitions used in the articles reviewed i.e., the diet is termed “Western” if it was described as a Western in the original report; likewise, the diet is termed “Mediterranean” if it was described as a Mediterranean in the original report.

Western diets derive much of their fat and protein content from animal sources, and include relatively high levels of salt and simple sugars. A large body of data from human studies suggests that Western diet consumption has deleterious effects on health. These data derive primarily from long-term population studies, and short-term feeding trials. Western diet consumption increases the risk of CVD, type 2 diabetes (Piernas et al., 2022), nonalcoholic fatty liver disease (Vachliotis et al., 2022), some cancers (Seo et al., 2022), depression (Matison et al., 2021), and all-cause mortality (Piernas et al., 2022). In contrast, diets that derive much of their fat and protein content from plant sources have comparatively beneficial effects on health and longevity (Herpich et al., 2022). For example, the Mediterranean diet, which derives most fat and protein from plant sources, is associated with decreased risk of CVD (Richardson et al., 2022), depression, some cancers, obesity, and related metabolic disorders (Muscogiuri et al., 2022). Much of this evidence comes from population studies (and some from intervention studies) that rely on self-reported food intake.

1.3. Diet Composition, Psychosocial Stress, , and Health

The adverse health effects of psychological stress may be amplified by other environmental stressors, including poor diet quality. As reviewed above, low SES is stressful and accompanied by poor diet quality. Observational studies suggest lower perceived stress is associated with high fruit, vegetable, and protein intake (Kye and Park, 2012; Laugero et al., 2011; Nguyen et al., 2017) and adherence to a Mediterranean diet pattern (Bonaccio et al., 2018; Crichton et al., 2013; Hodge et al., 2013), whereas higher perceived stress is associated with a Western dietary pattern (Ng and Jeffery, 2003; Tseng and Fang, 2011). Likewise, higher cortisol levels are associated with high simple sugar and saturated fat intake (Laugero et al., 2011; Michels et al., 2013).

Small, short-term, crossover studies in humans suggest that high fat diets, particularly those rich in saturated animal fats, increase cardiovascular responses to a standardized stressor (Jakulj et al., 2007; Straznicky et al., 1993; West et al., 2010). Thus, the limited clinical data available are consistent with the hypothesis that diet composition may affect physiological stress reactivity. Whether these responses are sustained chronically is unknown. Controlled clinical trials of long-lasting changes in stress responses due to diet are difficult and costly. Human population-based studies are not definitive since dietary data and stress levels are self-reported, often collected retrospectively, and nutrient intakes are drawn from food composition tables which estimate nutrient content. Thus, actual nutrient intake is unknown. Furthermore, observational studies compare individuals who choose to consume either a prudent or Western diet, and these two groups may differ in other characteristics that influence health such as socioeconomic status (SES), alcohol consumption, smoking behavior, and exercise levels (Nanri et al., 2010; Sanchez-Villegas et al., 2009). Other factors that affect stress reactivity (e.g., physical activity, medications) may be inaccurately reported, and in cross-sectional studies, there is the possibility of reverse causality (i.e., stressors may influence food choices). Finally, long-term, comprehensive, multi-system assessments of behavioral, anatomical, and physiologic responses to psychosocial stress and dietary interventions, under controlled conditions, are too demanding or unethical to carry out in human populations.

2. Nonhuman Primate Models of Human Health

2.1. Animal Models

Animal models can be a useful adjunct to clinical studies as many more potential confounding factors can be controlled (e.g., day/night cycle, food composition, assessment and feeding schedule, physical and social aspects of housing, access to healthcare, etc.), manipulations such as controlled dietary intake can be maintained long-term, and measurements may be made that are not possible in clinical studies. Rodent models have been useful to study the effects of psychological stress on health, and diet effects on health, but little has been done on the diet composition-stress relationship and its impact on health. Furthermore, few animal model studies have focused on diets that are similar to those consumed by humans. For example, often diets described as Western-like are actually high fat diets that focus on only one attribute of the diet in this case fat content (e.g., mouse diet TD 06414, Harlan, 60% of calories from fat, half from lard). Likewise, those described as Mediterranean-like are often standard rodent chow in which soybean oil has been replaced with olive oil (e.g., AIN-93G). In contrast, much nutritional research points to diet patterns, not single nutrients, as responsible for health impacts in human populations (Hu, 2002). Here we describe a nonhuman primate model of human health, long-tailed (a.k.a. cynomolgus) macaques (Macaca fascicularis). Long-tailed macaques are omnivores and opportunistic foragers that live throughout southeast Asia and Mauritius. They inhabit a wide range of habitats including riverine, secondary and primary forest, forest periphery, mangrove and nipa swamp, coastal forest, and urban and agricultural settings that differ in food source types and seasonal availability. As opportunistic foragers, wild macaques augment their natural food supply with food from human agriculture and garbage. We focused on this model for several reasons: 1) Their phylogenetic proximity to humans increases the likelihood of similar underlying biologic mechanisms; 2) their reliance on social relationships results in psychosocial stress, discussed in detail below; and 3) their physiologic and adverse health responses to Western-like diets are similar to those observed in humans. For example, early studies showed that Western-like diets (described in the next section) cause hypercholesterolemia, inflammation, and coronary artery atherosclerosis (Clarkson et al., 1985; Register, 2009). We also focused on diet pattern, rather than single-nutrient effects on health.

2.2. Long-tailed Macaque Model of Psychosocial Stress Effects on Health

Social subordination appears stressful in female macaques (Macaca spp.) because they receive more aggression, less affiliative attention (e.g., time being groomed), spend more time alone, and are more vigilant than their dominant counterparts (Shively, 1997, 1998; Shively and Day, 2015). In general, dominants control access to resources, whereas it is up to subordinates to continually gauge the social environment to avoid agonistic interactions. From this perspective, subordinates appear to fit the model of high demand and low control (Karasek et al., 1988).

These behavioral characteristics are accompanied by impaired ovarian function (Adams et al., 1985; Kaplan et al., 2010), impaired glucose handling and insulin resistance (Silverstein-Metzler et al., 2022), heightened heart rate (HR) responses to acute stressors (Shively, 1998), higher morning circulating cortisol concentrations (Shively, 1997), greater cortisol responses to adrenocorticotropin (ACTH) challenge (Kaplan et al., 1986; Shively, 1998), and insensitivity to glucocorticoid negative feedback in dexamethasone suppression tests of the hypothalamic-pituitary –adrenal axis (Kaplan et al., 2010; Shively, 1997), relative to their dominant counterparts (Figure 1). Subordinates also have lower bone density (Kaplan et al., 2010), greater visceral obesity (Shively et al., 2009b), more inflammation (Walker et al., 2008), increased biomarkers of breast and endometrial cancer risk (Shively et al., 2004), a greater likelihood of exhibiting depression-like phenotypes (Shively, 1997), and more diet-induced coronary and carotid artery atherosclerosis than their dominant counterparts (Shively et al., 2009a; Shively et al., 2009b; Shively et al., 2004). Male macaques are also deleteriously affected by psychosocial stress, however they are affected by different social stressors than females (Doyle et al., 2008; Fontenot et al., 1995). Notably, all of these observations were made in macaques consuming a Western-like diet.

Figure 1. Social subordination appears stressful in female macaques and results in multisystems perturbations that increase risk of disease.

Subordinate monkeys receive more aggression, less affiliative attention (e.g., being groomed), spend more time alone, and are more vigilant than their dominant counterparts (Shively, 1997, 1998; Shively and Day, 2015). Such chronic psychosocial stressors impact multiple systems (inner circle), including: increased inflammation (Walker et al., 2008), impaired glucose metabolism and insulin resistance (Silverstein-Metzler et al., 2022), neurobiological perturbations (Willard et al., 2009), heightened heart rate responses to acute stressors (Shively, 1998), higher morning circulating cortisol concentrations (Shively, 1997), greater cortisol responses to adrenocorticotropin (ACTH) challenge (Kaplan et al., 1986; Shively, 1997), and insensitivity to glucocorticoid negative feedback in dexamethasone suppression tests of the hypothalamic-pituitary–adrenal axis (Kaplan et al., 2010; Shively, 1997). These body conditions can promote the development of disorders and disease (outer circle). Compared to their dominant counterparts, subordinates have greater visceral obesity (Shively et al., 2009b), impaired ovarian function (Adams et al., 1985; Frye et al., 2023; Kaplan et al., 2010), are more likely to exhibit depression-like phenotypes (Shively, 1997), develop more diet-induced coronary and carotid artery atherosclerosis (Shively et al., 2009a; Shively et al., 2009b; Shively et al., 2004), have increased biomarkers of breast and endometrial cancer risk (Shively et al., 2004), and exhibit lower bone density (Kaplan et al., 2010). Notably, all of these observations were made in macaques consuming a Western-like diet. Figure created with Biorender.

Limited data from nonhuman primate (NHP) studies are also consistent with the hypothesis that diet may exacerbate responses to stressors, particularly the psychosocial stress of subordination. Hypothalamic-pituitary-adrenal (HPA) and autonomic nervous system (ANS) function have been characterized in female macaques consuming standard lab chow in some studies and Western-like diets in other studies. Intriguingly, post hoc comparisons suggest that the cortisol response to ACTH was higher in subordinate than dominant female macaques in studies in which animals were fed a Western-like diet (Kaplan et al., 1986; Shively, 1998), whereas the cortisol response to ACTH was lower in subordinates than dominants in studies in which monkeys consumed a chow diet (Michopoulos et al., 2012). In addition, we found that long-term Western diet consumption increased 24-hour heart rates in subordinates, but not dominants, relative to a baseline period consuming chow (Shively and Day, 2015). Thus, Western diet appeared to exacerbate the adverse effects of social subordination. However, these studies were not definitive as the cortisol response to ACTH was a post hoc comparison between the outcomes of two different studies, and the heart rate study did not include a diet control or comparison group – all animals ate the Western diet, thus diet and aging, were confounded.

The Western diet in these studies was modeled on the typical North American diet and contained 40% of calories from fat (mostly saturated), 0.25-0.40 mg cholesterol/kcal (450-700 mg cholesterol /day human equivalent), with protein and fat primarily derived from animal sources. In contrast, monkey chow is very low in fat (13% of calories) and cholesterol (trace amounts), with protein and fat almost entirely from vegetable sources (Diet #5037/5038, LabDiet, St. Louis, MO). The translational relevance to human health of physiological responses when monkey chow is fed is unknown since monkey chow has no human diet parallel. (See Table 1 for chemical composition comparison of human and nonhuman primate diets.) Thus, a more human-relevant diet comparison is needed.

Table 1.

Experimental diet compositions and the human diet patterns on which they were modeled

	Human		Nonhuman Primate
	Western	Mediterranean	Western*	Mediterranean*	Chow#
% of Calories
Protein	15a	17b	16	16	18
Carbohydrate†	51a	51b	54	54	69
Fat	33a	32b	31	31	13
% of Total Fats
Saturated	33a	21b	36	21	26
Monounsaturated	36a	56b	36	57	28
Polyunsaturated	24a	15b	26	20	32
Other Nutrients
ω6:ω3 Fatty Acids	15:1c	2.1-3:1d	14.8:1	2.9:1	12:01
Cholesterol mg/Cal	0.13a	0.16b	0.16	0.15	trace
Fiber g/Cal	0.01a	0.03e	0.02	0.04	0.01
Sodium mg/Cal	1.7a,f	1.3b,e	1.7	1.0	0.25

^*Developed and formulated at Wake Forest School of Medicine (Shively et al., 2019)

^#LabDiet Chemical Composition Diet 5037/8. Type of fat known in 86% of total fat. Omega-6 from corn and pork fat.

^†Human carbohydrate calories include alcohol.

^a(U.S. Department of Agriculture, 2010)

^b(Bedard et al., 2012)

^c(Simopoulos, 2006)

^d(Cordain et al., 2005)

^e(Kafatos et al., 2000)

^f(Powles et al., 2013)

reprinted with permission (Shively et al., 2019)

3.0. Diet Composition Modulation of Psychosocial Stress Effects on Health: A Longitudinal, Randomized, Nonhuman Primate Trial

3.1. Trial Design (Figure 2)

Figure 2.

Experimental design of the nonhuman primate randomized trial of Western versus Mediterranean diet consumption. HPA=hypothalamic-pituitary-adrenal; ANS=autonomic nervous system; MRI=magnetic resonance imaging

To test the hypothesis that long-term Western, compared to Mediterranean, diet consumption would result in greater psychosocial stress reactivity and adverse health outcomes, we conducted a longitudinal, randomized preclinical trial (Figure 2) in 38 socially housed middle-aged, female long-tailed macaques (Macaca fascicularis) (Shively et al., 2020). They lived with each other in groups of about 4 females each, and social status was determined monthly based on the outcomes of agonistic interactions (Shively et al., 2020). We chose to study diet patterns rather than single dietary constituents because human diets include multiple characteristics that may affect health (Hu, 2002). The experimental diets were formulated to emulate dietary patterns consumed by humans (see Table 1 for diet compositions). We focused on females because of mammalian sex differences in stress responses (Bale and Epperson, 2015). Additionally, much of the literature detailing social status differences in stress responses and their relationship to health outcomes is based on data from female NHPs. As the lifespan of these NHPs is approximately 3-4 times shorter than humans, the 31-month treatment period is approximately equivalent to about a decade of human adult life (Shively et al., 2021). Thus, changes over time may reflect normal aging that occurs in middle-age. Important to this investigation, stress resiliency is thought to slow aging and improve overall health (Faye et al., 2018; Shively et al., 2020). This study began in 2013, and is ongoing at this writing in 2023. The primary data analysis was conducted using 2 (Mediterranean, Western) X 2 (Dominant, Subordinate) analysis of variance or covariance where indicated, with repeated measures for dependent variables measured at multiple time points. Significant effects are summarized below and include main effects of diet, social status, and time, and diet by status, and time by diet interactions.

3.2. The Effects of the Western and Mediterranean Diets on Eating Behavior, Body Composition, and Metabolism

Compared to baseline measures taken while monkeys consumed standard lab chow, Western diet consumption increased caloric intake, body weight, body fat (determined by computed tomography), activity, energy expenditure (calculated from activity levels and body weight), insulin resistance, and hepatosteatosis (nonalcoholic fatty liver disease). Those that consumed the Mediterranean-like diet maintained stable and “healthy” levels of all of these metrics, and also reduced circulating triglyceride levels over the 3-year experimental period (Shively et al., 2019). In addition, circulating monocytes in the Western group exhibited a markedly different gene expression profile and pro-inflammatory signature relative to those that consumed the Mediterranean die t(Johnson et al., 2021).

3.3. Mediterranean Diet Confers Resilience to Stress and Aging-Related Declines in Middle-Aged Long-Tailed Macaques (Figure 3)

Figure 3. Theoretical Framework: Diet effects on stress resilience, susceptibility and the aging of the autonomic nervous system (ANS) and hypothalamic-pituitary-adrenal (HPA) axis. The Mediterranean and Western diets are described in detail in Table 1.

NHPs that consumed the Mediterranean diet exhibited enhanced stress resilience as indicated by lower sympathetic activity, brisker and more overt heart rate responses to acute stress, more rapid recovery, and lower cortisol responses to acute psychological stress and adrenocorticotropin (ACTH) challenge. Furthermore, age-related increases in sympathetic activity and cortisol responses to stress were delayed by the Mediterranean diet (Shively et al., 2020). Figure created with Biorender.

We examined the effects of diet on the ANS and the HPA axis because they have bidirectional effects on each other as well as independent and synergistic effects on molecular and cellular processes that promote downstream pathologies when chronically activated (Agorastos et al., 2019; Ginty et al., 2017; Timmermans et al., 2013; Wirtz and von Känel, 2017). The study focused on responses to the chronic stress of social subordination, and to the acute stress of brief social separation accomplished by removing the NHP from their home pen for 30 minutes. Both of these stressors are commonly experienced by this species. HPA axis function was assessed by measuring cortisol responses to ACTH challenge and to acute social separation (30 minutes). Heart rate variability was assessed with power spectral analysis to determine the relative inputs of the SNS (reflected in very low frequency (VLF) bands) and parasympathetic nervous system (PNS) (reflected in the high frequency (HF) band, and somewhat in low frequency LF bands) (Shively et al., 2020).

Compared to animals fed a Western diet, those fed the Mediterranean diet exhibited enhanced stress resilience as indicated by lower sympathetic activity, brisker and more overt heart rate responses to acute stress, more rapid recovery, and lower cortisol responses to acute psychological stress and ACTH challenge (Shively et al., 2020). Thus, diet composition had direct effects on stress resilience.

The HPA axis and ANS are sensitive to aging, respond to dietary patterns, and play interrelated roles in the pathophysiology of stress responses in humans (Barthelemy et al., 2022; Gaffey et al., 2016; Wichi et al., 2009). Thus, we examined longitudinal effects of Western versus Mediterranean diets and psychosocial stress on the aging of the HPA axis and ANS (Shively et al., 2020). In general, sympathetic activity increased whereas parasympathetic activity decreased over the course of the experiment, consistent with known changes with aging. However, these transitions occurred more slowly in the Mediterranean than the Western group. At one year, the Mediterranean group had lower sympathetic and higher parasympathetic activity than the Western group; however, these group differences disappeared by 2.5 years of experimental diet consumption. Likewise, cortisol reactivity to the acute social stressor increased over time for both diet groups consistent with known effects of aging in the HPA axis. However, while the Western group had significant increases in cortisol response (from baseline) by 12 months, the Mediterranean group only exhibited significant increases (from baseline) after 29 months of experiment. Taken together, these findings suggest that Mediterranean diet delayed the onset of aging-related shifts in the ANS and HPA axis (Shively et al., 2020).

4. Psychosocial Stress, Diet Composition and the Gut Microbiome

4.1. Diet Composition, the Gut-Brain Axis, and Health

The microbial communities occupying the mammalian gastrointestinal tract, the gut microbiome, contribute widely to host physiology and health (Cénit et al., 2014; Shreiner et al., 2015). Rodent studies have shown that gut bacteria influence metabolic (Visconti et al., 2019), immune (Zheng et al., 2020), and central nervous system (Cryan et al., 2019) function of their hosts through dynamic bidirectional communication along the gut-brain axis (Morais et al., 2021). Characteristics of the gut microbiota, such as relatively low diversity, low Lactobacillus abundance, and a high ratio of phylum Firmicutes to Bacterioides, are associated with chronic diseases of aging including obesity (Muscogiuri et al., 2019), rheumatoid arthritis (Gupta et al., 2021), CVD (Trøseid et al., 2020), anxiety, cognitive decline, and Alzheimer's disease (AD) (Cryan et al., 2019; Herselman et al., 2022; Saji et al., 2020). Diet composition significantly impacts the gut microbiota. In humans, Western diet consumption is associated with lower gut microbial diversity, a greater abundance of Firmicutes, and lower abundance of Bacterioidetes (Malinowska et al., 2022), whereas the Mediterranean diet is associated with higher microbial diversity (Garcia-Mantrana et al., 2018). In mice, Western diet promotes microbiome dysbiosis including reduced diversity and shifts in relative abundances between taxa (Agus et al., 2016; Martinez-Medina et al., 2014) which are associated with chronic diseases of aging (Gupta et al., 2021; Muscogiuri et al., 2019; Saji et al., 2020; Trøseid et al., 2020). Thus, diet impacts on gut microbial diversity are thought to be one pathway through which diet composition affects disease risk.

In rodents, psychological stress decreases gut microbial diversity and Lactobacillus abundance (Herselman et al., 2022), and in one rodent study, stress-induced despair behavior was reduced by restoration of gut Lactobacillus (Karl et al., 2018). Similar observations have been made in nonhuman primates. For example, among free-ranging rhesus macaques, less sociable individuals exhibited lower abundances of beneficial bacteria and higher abundances of genus Streptococcus, some species of which are actively pathogenic (Johnson et al., 2022b), and a study of captive cynomolgus macaques found associations between gut microbiome perturbations and depressive behavior (Zheng et al., 2021). However, evidence of a simple causal relationship between psychological stress and gut dysbiosis in humans is scarce. For example, the extreme stress of arctic combat training altered microbiota in military personnel, but the stressors were largely physical rather than psychosocial (Karl et al., 2017). One study reported that as university student stress increased over the semester, health-promoting bacteria decreased (Knowles et al., 2008), whereas others report little association with perceived stress (Almand et al., 2022), or unpredictable patterns of dysbiosis (Zaneveld et al., 2017). The clearest data linking the gut and brain come from rodent studies in which the complexity of human brain structure and function is only partially recapitulated; thus, more experimental data from large animal models and clinical cohorts are needed (Berding et al., 2021; Morais et al., 2021).

4.2. Effects of Diet Composition on the Gut Microbiome of Long-Tailed Macaques

In the randomized NHP trial described above, the macaque gut microbiome, sampled at the end of the treatment phase, had human-like responses to Mediterranean and Western diets (Nagpal et al., 2018a). The Mediterranean group had higher microbial diversity, as measured by the Shannon diversity index, whereas the Western diet group had a higher Firmicutes:Bacteroides ratio (Nagpal et al., 2018a), a profile thought to reflect gut dysbiosis, and compromised health (Pascale et al., 2019; Yang et al., 2015). The Mediterranean microbiome was enriched in the families Clostridiaceae and Lactobacillaceae, and the genera Lactobacillus, Clostridium, Faecalibacterium, and Oscillospira, and had lower levels of Ruminococcus and Coprococcus. (Nagpal et al., 2018a). Interestingly, Western and Mediterranean group microbiomes differed in ways consistent with patterns also associated with health impacts of psychological stress i.e., the Western diet decreased microbial diversity, whereas the Mediterranean diet promoted microbial diversity and increased bacterial abundances associated with reduced psychological stress. These observations suggest that psychosocial stress and Western diet may have similar and thus perhaps additive adverse effects on gut microbial diversity. More study is needed of the relationships between psychosocial stress, diet, gut microbiome, and cognitive and emotional behavior in primates to clarify these relationships.

5. Effects of Diet and Psychosocial Stress on Mitochondrial Function

5.1. Diet Composition, Mitochondrial Dysregulation, and Health

Mitochondria, subcellular organelles responsible for energy production and cellular respiration, power basic biological processes critical to homeostasis. Mitochondrial function is thus critical for all physiologic functions. Skeletal muscle mitochondria provide chemical energy for physical movement (Hood et al., 2019). In the central nervous system, mitochondria support neuronal viability and functionality and shape all aspects of brain function including cognition (Picard and McEwen, 2014). Mitochondrial dysfunction has been extensively associated with chronic diseases of aging including sarcopenia, frailty (Bellanti et al., 2021; Marzetti et al., 2013), type 2 diabetes (Lowell and Shulman, 2005), CVD (Chistiakov et al., 2018), and AD (Wang et al., 2020).

In rodents and in vitro, diet-nutrient composition impacts mitochondrial bioenergetic function (Putti et al., 2015). High saturated fat, as found in Western diets, increases mitochondrial intake and oxidation of surplus dietary fatty acids resulting in increased production of reactive oxidative species (ROS). Thus, mitochondria modulate oxidative stress which causes cell damage and promotes inflammation (Putti et al., 2015). In contrast, antioxidants including dietary omega-3 polyunsaturated fatty acids found in abundance in Mediterranean diets improve mitochondrial function and reduce ROS production. Some rodent studies show diet effects on mitochondrial function in brain (Cavaliere et al., 2019), while others report effects on muscle but not brain mitochondria (Jørgensen et al., 2015), implying that diet effects on mitochondrial function could be tissue specific (Didier et al., 2007). In humans, obesity and nonalcoholic hepatosteatosis are associated with altered mitochondrial characteristics in adipose tissue which increase ROS production (Khalil et al., 2022).

Mitochondrial function is a fundamental component of the stress response. Psychosocial stress and mitochondrial function are associated in two ways. First, in rodents, chronic stress decreases mitochondrial energy production capacity, alters morphology, and increases ROS production. Antioxidants may protect against stress-induced mitochondrial dysfunction. Additionally, in human studies, positive associations between self-reported psychosocial stress and some aspects of mitochondrial biology e.g., increased function and size have been observed (Picard and McEwen, 2018a). Second, mitochondrial function mediates physiological stress responses. Rodent studies show that once mitochondria sense and respond to psychosocial stresses, they initiate multi-system stress responses such as increased HPA production of glucocorticoids and increased SNS, decreased PNS and reduced HRV (Picard and McEwen, 2018b). There is no experimental evidence of direct psychosocial stress effects on mitochondrial structure or function in primates.

5.2. Diet Effects on Mitochondrial Function in Skeletal Muscle and Brain of Long-Tailed Macaques

In our NHP diet trial, diet composition affected mitochondrial respiration in vastus lateralis skeletal muscle (Gonzalez-Armenta et al., 2019). Consistent with observations in rats fed high-fat diets (Hancock et al., 2008; Turner et al., 2007), overall mitochondrial respiration was higher in permeabilized muscle fibers from the Western diet group than the Mediterranean group, as indicated by fatty acid β-oxidation rates as well as maximal electron transport system capacity. Respiration measures were positively correlated with insulin resistance as measured by the Homeostatic Model Assessment of Insulin Resistance (HOMA IR) and fasting plasma insulin concentrations, suggesting causal links between insulin resistance and mitochondrial dysfunction in skeletal muscle (Gonzalez-Armenta et al., 2019).

Diet effects on mitochondrial function were also assessed in three brain regions: the prefrontal cortex, entorhinal cortex, and cerebellum (Amick et al., 2021). Mean respiratory capacity within individual regions was similar between diet groups. Distinct brain regions have different metabolic requirements related to their function and disease susceptibility. By comparing mitochondrial bioenergetic capacity between regions, we observed that animals fed a Mediterranean diet maintained distinct patterns of mitochondrial bioenergetic capacity between brain regions, with prefrontal cortex mitochondria exhibiting the highest rates of respiration compared to the entorhinal cortex and cerebellum. In contrast, in the Western group, these patterns were lost. Furthermore, bioenergetic function in the prefrontal and entorhinal cortices of the Western group, but not the Mediterranean group, exhibited associations with blood glucose and insulin levels. This is notable given the susceptibility of these two brain regions to neuropathologic changes in Alzheimer’s disease (Salat et al., 2001; van Hoesen et al., 1991), and type II diabetes (Chen et al., 2015; Thielen et al., 2019). These observations suggest that the Western diet disrupts expected patterns of mitochondrial function, and that mitochondrial function in prefrontal and entorhinal cortex, both of which are strongly linked to neurocognitive disease, may be shaped by peripheral glucose and insulin levels which are deleteriously affected by Western diet consumption (Amick et al., 2021; Shively et al., 2019). While these are among the first experimental evidence of diet effects on mitochondrial function in nonhuman primates, more study is needed to determine pathways by which diet, and particularly psychosocial stress, influence brain mitochondrial function in primates.

6. Diet Composition and Psychosocial Stress Effects on Brain Structure and Function

6.1. Associations Between Diet Composition, Psychosocial Stress, and Brain Structure in Humans

In human beings, associations between psychosocial stress and brain structure have been observed. In a longitudinal study, the number of stressful life events was associated with a decrease in gray matter volume in the anterior cingulate, hippocampus, and parahippocampal gyrus (Papagni et al., 2011). Early life stress was associated with smaller anterior cingulate cortex and caudate nuclei (Cohen et al., 2006a). These observations are consistent with experimental observations of atrophy of hippocampus and other brain areas in response to repeated stress in animal models (McEwen, 2000).

Diet exerts profound effects on brain structure and function (Medawar and Witte, 2022). The Western diet, in particular, is thought to have negative consequences on brain structures, including the hippocampus and cortex, that likely affect cognitive function (López-Taboada et al., 2020). In contrast, Mediterranean diets may provide neuroprotective effects, including decreasing the likelihood of developing AD (Lourida et al., 2013). However, with few exceptions, data from humans are drawn from observational studies that rely on self-reported data of dietary and other lifestyle habits, which may be inaccurate (Ravelli and Schoeller, 2020) or confounded by behavioral patterns correlated both with diet and neurocognitive outcomes (Akbaraly et al., 2019; Marcone et al., 2020). Experiments testing relationships between diet and neurobiological outcomes are therefore essential to understand the direct role of diet in modulating behavior and risk for affective and neurocognitive disorders.

6.2. Diet Composition, Psychosocial Stress, and Brain Structure in Long-Tailed Macaques

In our NHP diet trial, structural MRI was conducted at baseline and the end of the study to determine the effects of diet composition and psychosocial stress on brain anatomy (Frye et al., 2021). Structural phenotypes included intracranial volumes of cerebrospinal fluid, white matter, gray matter, total brain volume, intracranial volume (total brain volume + cerebrospinal fluid), and the volumes of specific regions of interest (ROIs), including hippocampus (whole, anterior, and posterior), amygdala, temporal (auditory and visual), prefrontal, and parietal cortex, and cerebellum. We also measured thicknesses comprised of gray matter, and volumes of several temporo-parietal cortical ROIs relevant to AD i.e., angular gyrus, the inferior, middle, and superior temporal gyrus, entorhinal cortex, fusiform cortex, supramarginal gyrus, precuneus, and parahippocampus. From these, we generated composite temporo-parietal thickness- and volumetric metaROIs, as analogous AD metaROIs more reliably predict AD neuropathology in humans than their individual components (Schwarz et al., 2016).

In the Mediterranean group, global brain volumes were stable over the course of the study, whereas global brain volumes changed in the Western group. Relative to the Mediterranean group, the Western group had greater total brain volumes, cortical thicknesses and gray matter volumes, and reduced white matter and cerebrospinal fluid volumes. The observed effects of diet on brain volumes are consistent with results from the Alzheimer’s Disease Neuroimaging (ADNI) study which showed in middle-older aged humans, that increases in gray matter volumes in mid-life predict the occurrence of Alzheimer’s in later life. These investigators hypothesized that these increases in gray matter in mid-life reflect neuroinflammation (Eskildsen et al., 2013).

Social status was also associated with brain structure: subordinates had smaller total brain volumes but larger volumes of temporoparietal ROIs relevant to AD neuropathology. These observations add to a body of literature associating chronic psychosocial stress and social status with structural differences in the brain (De Looze et al., 2023; Noonan et al., 2014; Tan and Tan, 2023; Willard et al., 2009).

6.3. Transcriptomic Evidence of Diet-Induced Neuroinflammation in Long-Tailed Macaques

We also assessed the effects of diet on transcriptional profiles of lateral temporal cortex (Negrey et al., in review). At an FDR=0.05, we identified seven differentially expressed genes (DEGs) that distinguished the diet groups. Cyclin dependent kinase 14 (CDK14) was the only DEG that was downregulated in the Mediterranean group relative to the Western group. The remaining six genes were upregulated in the Mediterranean group: butyrophilin subfamily 2 member A1 (BTN2A1), katanin regulatory subunit B1 (KATNB1), “lunatic fringe” (LFNG), mannose receptor C type 2 (MRC2), solute carrier family 3 member 2 (SLCA32), and transmembrane protein 268 (TMEM268). Four of the seven genes—CDK14, LFNG, MRC2, and SLCA32—have established links with inflammatory pathways (Derada Troletti et al., 2018; Jridi et al., 2021; López-Guisa et al., 2012; Yang et al., 2018). These results were further supported by correlations with global brain volumes and cortical thicknesses. CDK14, which has known proinflammatory properties, was positively correlated with changes in total brain, total gray matter, and cortical gray matter volumes as well as the AD-associated temporoparietal meta-ROI thicknesses, and was negatively correlated with changes in total white matter and cerebrospinal fluid volumes. Several of the remaining DEGs, including LFNG, MRC2, and SLCA32, exhibited the opposite correlations with brain volumes e.g., negatively correlated with cahnges in total gray matter volume, positively correlated with changes in CSF volume, consistent with anti-inflammatory or neuroprotective effects. These transcriptomic data demonstrate differential effects of diet on temporal cortex gene expression and support the hypothesis that Western diet promotes neuroinflammation in mid-life resulting in increased gray matter volumes while the Mediterranean diet appears to be neuroprotective.

7.0. Diet Effects on Social Behavior

7.1. Social isolation, Anxiety, and Disease Risk in Humans

Social isolation and anxiety are prevalent, often comorbid (Holt-Lunstad, 2018a; Holt-Lunstad, 2018b), increase cardiometabolic and cardiovascular disease risk (Horsten et al., 1999; Kahl et al., 2015; Lee et al., 2022; Olafiranye et al., 2011; Valtorta et al., 2016), and may be influenced by diet. Consumption of Western diets is associated with increased prevalence of anxiety whereas adherence to Mediterranean diets is associated with decreased anxiety (Bakhtiyari et al., 2013; Jacka et al., 2017; Jacka et al., 2010). However, much of these data are associations derived from cross-sectional studies, rely on self-report of diet consumption, and reverse causation, i.e. anxiety driving eating patterns, cannot be ruled out (Keck et al., 2020).

7.2. Diet Impacts on Anxiety and Social Isolation in Female Long-Tailed Macaques

In our NHP diet trial, the behavior of the two diet groups was distinctly different (Johnson et al., 2021; Johnson et al., 2022a). Principal components analyses showed that the Western group was more socially isolated, and less socially integrated than the Mediterranean group as evidenced by less time in body contact with group mates and more time alone, and exhibited more anxiety-like (e.g., self-directed) behaviors. In contrast, the Mediterranean group spent more time resting, attentive to the environment, and in body contact with groupmates, suggesting they were more relaxed, engaged, and socially integrated. These diet-altered behaviors mediated the effects of Western and Mediterranean diet on inflammatory gene expression in circulating monocytes. This suggests that diet composition may have neurologic effects that impact monocyte gene expression (Johnson et al., 2021). Analysis of longitudinal effects of diet on behavior revealed that these effects were rapid, occurring in the first 3 months, and persisted throughout the study. When compared to the baseline period during which the animals were fed laboratory chow, we found that these group differences were largely driven by behavioral changes in the Mediterranean group (Johnson et al., 2022a). These findings suggest that Mediterranean diets may be used as therapeutic interventions on social isolation and anxiety.

8.0. Diet, Psychosocial Stress, and Ovarian Function

8.1. Associations of Diet and Psychosocial Stress with Ovarian Function in Humans

Ovarian dysfunction is associated with increased risk of a number of adverse health effects, including coronary heart disease, osteopenia, and neurological disorders (Podfigurna-Stopa et al., 2016; Solomon et al., 2002). Self-reported adherence to Western-style diets is also associated with ovarian dysfunction, including irregular cycles, hormonal insufficiency, and anovulation; whereas self-reported adherence to a diet characterized by monounsaturated fats, plant proteins, low glycemic carbohydrates, iron supplements, and multivitamins is associated with reduced risk of ovarian disorders (Chavarro et al., 2007; Fontana and Torre, 2016; Jurczewska and Szostak-Węierek, 2022). Psychosocial stress is also associated with ovarian dysfunction in women (Berga, 2019), and experimental outcomes in long-tailed macaques support the clinical observations that psychosocial stressors can impair ovarian function (Kaplan et al., 2010; Kaplan, 2008; Kaplan and Manuck, 2004). These findings suggest the combination of Western diet and stress may have an additive negative effect on ovarian function, and that Mediterranean-like diets may have a beneficial effect. However, experimental data describing diet composition effects on ovarian function are scarce likely due to the difficulty of long-term study of integrated effects of diet and stress, and the challenge of collecting detailed hormonal data over long time periods.

8.2. Diet and Stress Effects on Ovarian Physiology in Long-Tailed Macaques

True menstrual cycles are characteristic of humans and some other primates (Emera et al., 2012). Little is known regarding the effects of diet on the menstrual cycles of primates, however, in young adult macaque ovaries, Western diet consumption altered the structure and function of antral follicles, and also suppressed survival and function of secondary follicles in vitro (Bishop et al., 2016; Xu et al., 2015).

To address this issue, we examined three metrics of ovarian function – luteal phase peak progesterone for each cycle, menstrual cycle length, and menstrual cycle regularity – in the cohort of 38 socially housed, middle-aged, female long-tailed macaques described above (Frye et al., 2023). Since social subordination is stressful in this species we used social status as marker of psychosocial stress (Figure 1) (Shively and Day, 2015). Ovarian function was assessed during the last 12 months of the dietary intervention with repeated vaginal swabbing for menses (6/week) and blood sampling (3/week) for hormone assessments. To identify potential mechanistic pathways through which diet and stress affect ovarian function, we used mediation analysis (as described by VanderWeele (VanderWeele, 2013)) to determine the extent to which obesity, cortisol reactivity, or social isolation are responsible for the effects of diet and/or social status on ovarian physiology.

We found that both Mediterranean diet and high social status resulted in higher circulating progesterone reflecting healthy ovarian functioning. Consumption of the Mediterranean diet was also associated with greater menstrual cycle regularity. Mediation analyses indicated that diet-driven differences in ovarian function were partially explained by variation in adrenal reactivity and also social isolation. These findings provide causal evidence that long-term (equivalent to about a decade in humans) consumption of a Mediterranean diet results in better ovarian health compared to consumption of Western diets. Thus, Mediterranean diets may represent intervention techniques to promote ovarian health. Such interventions, in turn, may have lasting influences on women’s health by protecting against the pathologic sequelae of ovarian dysfunction.

9. Synthesis and the Way Forward

9.1. Effects of Diet Composition and Psychosocial Stress on the HPA axis and ANS

While the significant impact of psychosocial stress on health is widely recognized, there is no population level treatment to reduce stress and associated disease. The adverse health effects of psychological stress may be amplified by an unhealthy diet, which has far-reaching effects on health. Stringently controlled long-term studies of the effects of both diet composition and psychological stress are difficult to conduct in humans. Animal models represent a critical adjunct to clinical studies as many confounding factors can be controlled, manipulations such as diet composition can be maintained long-term, and extensive phenotyping not possible in clinical studies can be conducted. Here we described a comprehensive, long-term randomized nonhuman primate trial of the effects of diet composition and psychosocial stress on health in female long-tailed macaques. Our previous work demonstrated that low social status is stressful in macaques and has deleterious effects on health. Likewise, Western diet consumption has deleterious effects on health similar to those observed in humans. In this study we found that Western diet reduces, and Mediterranean diet increases stress resilience in the ANS and the HPA axis. Furthermore, Mediterranean diet delayed the aging of the ANS and HPA axis. Thus, poor diet quality diet may compromise effective responses to psychosocial stress and accelerate aging of stress-responsive systems.

9.2. Mechanisms Underlying Stress and Diet Effects on Stress Reactivity (Figure 4)

Figure 4:

Model of multisystem effects of psychosocial stress and diet composition. Variables in red boxes were measured in nonhuman primates in our randomized preclinical trial. Others were observed in in vitro or rodent studies. SNS: sympathetic nervous system; PNS: parasympathetic nervous system; HRV: heart rate variability; HPA: Hypothalamic-pituitary-adrenal axis

There are several mechanisms through which Western diet composition and psychosocial stress may have additive effects that worsen health (Figure 4). In vitro and rodent data have shown that chronic stress and diet composition have tissue-specific effects on mitochondrial structure and function, and that mitochondria function is a fundamental component of the stress response (Picard and McEwen, 2018a, b). In female macaques, both diet composition and psychosocial stress also had tissue-specific effects on mitochondrial function in skeletal muscle and brain. The biology of stress and diet effects in female primates is complicated by cyclical ovarian steroid production. There are estrogen and progesterone receptors in mitochondria. Estrogens enhance mitochondrial biogenesis and energy-transducing capacity, and facilitate nucleus-mitochondria communication (Guajardo-Correa et al., 2022; Klinge, 2008; Liao et al., 2015). Progesterone stimulates mitochondrial activity, and increases cellular respiration (Behera et al., 2009; Dai et al., 2019; Price and Dai, 2015). Estrogens and progesterone directly modulate the expression of mitochondrial proteins, modulating energy metabolism and mitochondrial biogenesis and its downstream targets (Behera et al., 2009; Ventura-Clapier et al., 2019). In ovariectomized rats, hormone replacement with estradiol and progesterone induced mitochondrial alterations in the CNS that support bioenergetics, reduce oxidative stress, and thus reduce oxidative damage and neuroinflammation (Irwin et al., 2008). Here, female macaque ovarian cyclicity and steroid production were suppressed by psychosocial stress and Western diet, thus reducing the beneficial effects on mitochondrial function.

In female macaques, both Western diet and psychosocial stress altered ANS function, increasing SNS and decreasing PNS outflow, and increased HPA activity (Shively et al., 2020), likely involving alterations in mitochondrial function. Increased SNS and HPA activity are known to alter gut microbiota abundance and activity through neuroendocrine and catecholamine modulation of microbial growth, and reduced GI motility and perfusion, which eventually degrade the gut barrier increasing permeability (Karl et al., 2018). These changes in gut microbiota and permeability can promote monocyte proinflammatory cytokine production. In macaques, Western diet promoted proinflammatory monocyte polarization as well as anxiety (Johnson et al., 2021). In rodents, repeated social defeat causes SNS-dependent monocyte trafficking to the brain which promotes microglial neuroinflammatory signaling and prolonged anxiety. Microglia activation and sympathetic outflow to the peripheral immune system reinforces stress-related behaviors such as anxiety by recruiting inflammatory monocytes to the brain (Wohleb et al., 2014). In our macaque trial, Western diet increased proinflammatory gene transcripts in the temporal cortex, which was accompanied by alterations in brain volumes (Negrey under review, (Frye et al., 2021)). Western diet also decreased gut microbial diversity and increased bacterial abundances, both of which are associated with psychological stress, cognitive dysfunction, and chronic disease in humans, whereas the Mediterranean diet promoted bacterial diversity (Nagpal et al., 2019; Nagpal et al., 2018b). Thus, the deleterious effects of both diet composition and psychosocial stress may be partially mediated through diet effects on the gut microbiome.

9.3. Gaps in Knowledge and Future Studies

While rodent studies have provided important insight about key research questions regarding stress, diet, and health, many observations in rodents have yet to be confirmed in primates. Furthermore, the gut-brain axis has not been well characterized in primates. Critically, the interactive effects of diet and psychosocial stress on cognitive performance remain understudied. Such work is critically important to understanding the functional impacts of diet- and stress-induced perturbations in the periphery. Finally, we do not know whether psychosocial stress has direct effects on gut microbiota or permeability, mitochondrial function, or cortical proinflammatory gene transcription in primates. Finally, current data are insufficient to determine whether there are additive or synergistic effects of psychosocial stress and diet composition on health. This is a critical knowledge gap given that poor quality diet is often characteristic of the most stressed individuals in our society. Studies of the interactive effects of psychological stressors and dietary patterns in macaques are ongoing in our laboratories.

9.4. Diet: A Therapeutic Intervention

The impact of diet composition on anxiety social isolation, and affiliation suggests that diet composition may be manipulated to improve social integration in captive NHP groups. Furthermore, the observations in these NHPs suggest that changes in diet composition could be an effective intervention in humans to reduce stress reactivity, promote gut health, affiliative social behavior and integration, and mitochondrial and ovarian function, and stabilize brain structure during midlife. Population level diet modification in humans has been shown to be feasible as evidenced by 1) the National Cholesterol Education Program which led to decreased dietary fat and cholesterol intake, and circulating cholesterol concentrations (Cleeman and Lenfant, 1998), and 2) the FDA requirement to list trans fats on food labels and reduce trans fats in foods which reduced circulating levels of trans fats by more than 50% (Li et al., 2017). Our findings suggest that population-wide adoption of a Mediterranean-like diet pattern may provide a cost-effective intervention on psychological stress and promote healthy aging with the potential for widespread efficacy.

Highlights.

• Social disadvantage is accompanied by poor diet quality, both of which impact health.

• Social stress and Western diet have additive adverse effects on multiple systems.

• Mediterranean diet protects against many of the deleterious effects of social stress.

• Mediterranean diet may be an efficacious intervention on psychosocial stress.