1 INTRODUCTION

In humans, stress represents a major vulnerability factor for developing different neuropsychiatric disorders such as depression, anxiety and schizophrenia. Consequently, different preclinical stress models have been developed for studying biobehavioural factors underlying psychopathology (Italia et al., 2020). For these models to be considered reliable and informative, multiple methodological, empirical, and theoretical criteria should be fulfilled to ensure that the obtained results represent to some extent the behavioural and physiological processes occurring in humans (Belzung & Lemoine, 2011; Willner, 2017a). In this regard, criticisms of models of acute and chronic stress have arisen over the last decades, mainly because (1) the stressors used are excessively severe and unnatural (e.g., electric shocks, immobilisation and tail pinch) and (2) the protocols are scheduled to vary randomly in frequency and intensity, allegedly to better resemble the stress of modern human life (for details, see Section 5). Although these protocols reliably induce stress responses and cause several pathological phenotypes, it can be argued that these approaches do not reflect the stressome (i.e., the whole spectrum of stressors to which an organism is naturally exposed) of neither rodents nor humans. The latter inevitably leads us to ask which are the main sources of stress in our current modern life? Are the most common chronic, noncommunicable diseases derived from severe and unusual stressors (e.g., coronary and metabolic diseases, autoimmune conditions and neuropsychiatric disorders)? Although there are no straight answers, it is not difficult to associate most mental, cardiovascular and metabolic diseases with our new “natural” habitat: an impoverished environment and stressful lifestyle in which minor, recurrent and often predictable events accrue during the lifespan (Juster et al., 2010; McEwen & Wingfield, 2003). In this scenario, most preclinical models might have failed to represent the stress associated with our modern human lifestyle, leading us to reexamine one of the most fundamental, sometimes neglected but easy to model stress stimulus in humans: the living environment. Therefore, the aims of the present review are (1) to characterize some of the main sources of stress derived from living in urban areas, (2) to highlight the impact on welfare and mental health of being exposed to the so-called “green factor” and other enriching contexts in humans, (3) to criticise the rodent models of stress, (4) and to discuss how housing conditions models can be adapted to study the stressful effects of our modern lifestyles.

2 THE STRESS OF LIVING IN URBAN AREAS IN HUMANS

The number of people living in urban areas increased dramatically over the last 70 years. In 1950, 30% of the worldwide population resided in cities. By 2018, the percentage increased to 55% and it is expected to reach 68% by 2050 (United Nations, Department of Economic and Social Affairs, Population Division, 2018). The land used for urban expansion also increased during that time twice as fast as the population grew, with an expected fourfold increment from 2000 to 2050 (Angel et al., 2011). Conversely, green spaces within cities have reduced progressively, with only 10%–25% of the area of some European towns (e.g., Warsaw, Dublin and Rome) being covered with parks and forests (Wicht & Kuffer, 2019). As modern cities transformed into densely populated areas with limited access to green spaces, different health problems associated with this new human habitat have appeared (Evans, 2003). For example, office jobs—highly concentrated in urban areas—demand increasingly longer sedentary periods with fewer opportunities for physical activity (Biernat et al., 2010; Thorp et al., 2012). Office workers spend up to 9.7-hr sitting but less than 30-min walking during workdays, with >35% of them reporting low physical activity throughout the week (Biernat et al., 2010). About 75% of working hours are dedicated to uninterrupted sedentary activities, with less than 2.5% of daily time being reserved for moderate-to-high intensity physical activity (Thorp et al., 2012). Moreover, people living in crowded urban areas with limited access to public transportation and a reduced number of parks report lower levels of physical activity (Sallis et al., 2016), with those spending less than 30 min per week in green spaces being more likely to experience depression and high blood pressure (Shanahan et al., 2016). Besides the negative effects of a sedentary lifestyle, new technologies commonly used in working and recreational spaces have also limited the psychological benefits of face-to-face interactions (Kowert et al., 2014), with the combination of high screen time and low physical activity increasing the likelihood of experiencing depressive and anxiety symptoms, low self-esteem and life dissatisfaction (Hrafnkelsdottir et al., 2018). Furthermore, when people spend some time doing physical activities outdoors, their benefits are sometimes overshadowed by the increased exposure to airborne contaminants such as particulate matter, highly concentrated in cities (Tainio et al., 2016; Zhang et al., 2018). Particulate matter consists mostly of secondary inorganic components (e.g., ammonium sulfate and ammonium nitrate), crustal materials and secondary biogenic aerosols, which may induce cellular oxidation at the concentrations found in densely populated areas (e.g., Paris, London, Milan; Daellenbach et al., 2020). Fine-mode secondary organic aerosols—which have the greatest oxidative potential—come from residential biomass burning and vehicular emissions highly concentrated in cities that produce levels of particulate matter threefold higher than those found in rural areas (Daellenbach et al., 2020). As exposure to particulate matter is linked to systemic inflammation (Zhang et al., 2018), respiratory (Jo et al., 2017) and cardiovascular diseases (Liang et al., 2020) and lower mental health status (Dzhambov et al., 2018), outdoor activities in densely populated cities may have mixed health consequences.

Other pollutants such as artificial lighting and noise are also abundant in urban areas and are associated with poor well-being (Griefahn et al., 2008; Hunter & Hayden, 2018). At nights, high levels of artificial illumination observed in cities and urban neighborhoods can disrupt circadian rhythms and affect sleep (Hunter & Hayden, 2018). In fact, artificial lighting within homes extends over 4 hr after evening, creating a prolonged twilight capable of inducing melatonin suppression, especially in houses with energy-efficient lightning (i.e., LEDs and fluorescent lights; Cain et al., 2020). As a result, such extended exposure to artificial illumination increases wakefulness after bedtime and worsens quality and quantity of sleep (Cain et al., 2020). Residential noise also compromises general health by increasing subjective annoyance and irritation, which in turn affects the perception of the restorative quality of neighborhoods and discourages physical activity (Dzhambov et al., 2018). Moderate-to-high levels of noise coming from construction sites, commercial establishments and trains increase levels of stress and anxiety in city dwellers (Ma et al., 2018). The greater the noise experienced by people living closer to main roads, the higher the likelihood of suffering mental health disturbances (Ma et al., 2018). Likewise, city dwellers moving to a location with undesirable environmental noise conditions are up to 50% more likely to report mental health problems (Ma et al., 2018). Moreover, the combination of artificial illumination and noise constitutes one of the primary causes of insomnia in urban dwellers (Griefahn et al., 2008; Hunter & Hayden, 2018). When these factors combine with high levels of particulate matter in urban environments with reduced green spaces, a synergetic negative effect is produced on mental health (Klompmaker et al., 2019).

It is not a coincidence, therefore, that all these environmental stressors (e.g., noise, artificial lighting and airborne contaminants) have been constantly associated with higher risks of cardiovascular diseases, diabetes, obesity and psychiatric disorders (Cho et al., 2015; Evans, 2003; Gruebner et al., 2017; Hunter & Hayden, 2018; Klompmaker et al., 2019), which are referred to as some of the main consequences of allostatic overload derived from our modern lifestyle (Juster et al., 2010). Allostatic load relates to the cumulative cost to the body of allostasis, which is the process of maintaining stability through change while adapting to predictable and unpredictable events (McEwen & Wingfield, 2003). In fact, the so-called type 2 allostatic load is extensively associated with our new habitat (i.e., urban environments) and lifestyle and encompasses two fundamental characteristics: a positive energy balance (i.e., high caloric intake but low expenditure through physical activities) and exposure to low intensity but predictable chronic stressors. Most of these stressors are associated with living in crowded cities and having low socioeconomic status and include poor housing and working conditions, limited access to education, health services, recreational opportunities and exercise (Juster et al., 2010; McEwen & Wingfield, 2003). For instance, in the United States, most vulnerable groups (e.g., lower-income earners, African Americans and Latinos) live in neighborhoods with few or no green spaces, which in turn reduces the chances of outdoor activities (Astell-Burt et al., 2014; Landau et al., 2020). Furthermore, those disadvantaged groups have poor educational spaces and limited access to healthcare, jobs and beneficial social networks (Firebaugh & Acciai, 2016) and are also less likely to visit green and enriching environments within metropolitan areas (Hess, 2020; Wang et al., 2018). In fact, the exposure to poor environments and neighborhoods within cities has increased over the last 30 years, with people from rural and nonurban areas showing as half the exposure to such contexts as that of city dwellers (Hess, 2020). Thus, populations living in disadvantaged environments are more likely to be both segregated and isolated from more stimulating contexts, which also may exacerbate their exposure to other stressors commonly found in poor neighborhoods (e.g., violence and crime). In this regard, the poor socio-affective interactions and loneliness experienced by socially isolated people adversely affect the optimal development and mental health of urban dwellers. Social isolation (SI) suffered by institutionalised children (i.e., orphanages) negatively affects communication and social and motor skills and induces immunological changes and epigenetic alterations in a plethora of genes (e.g., RB1 and HSPA9) related to children's behavioural impairments (Naumova et al., 2019). In middle-aged and elderly population, perceived loneliness predicts depressive symptoms upon different individual and social parameters (Cacioppo et al., 2010), with people living in urban areas being more likely to experience SI (Menec et al., 2019) and loneliness (Broek, 2017; van den Abshire et al., 2020) compared to those living in rural areas. In young and middle-aged adults, periods of SI as short as 10 hr can induce loneliness, evoke “social craving,” and increase brain activity of dopaminergic areas associated with motivated behaviours (i.e., substantia nigra pars compacta and ventral tegmental area; Tomova et al., 2020). In contrast with acute SI, chronic loneliness induces the opposite effect: It reduces “social craving” and the activity of those brain areas, which might lead to social withdrawal and affective disturbances (Tomova et al., 2020). Although many socioeconomic (e.g., neighbourhood features and income) and individual conditions (e.g., age, sex and partners) modulate the association between urbanicity, SI, loneliness and human well-being, it is undeniable that the impact these factors can have on life quality and optimal development across the lifespan (Buecker et al., 2021; Menec et al., 2019). Furthermore, as marginalising factors are long-lasting and almost immutable for some people (e.g., poor social relationships, reduced environmental stimulation and sedentarism), living in urban areas might constitute a real impoverishing condition for a huge proportion of human populations. It is not surprising, therefore, that people living in those contexts are more likely to develop different neuropsychiatric disorders such as psychosis, schizophrenia and depression (Gruebner et al., 2017; Hartig & Kahn, 2016; Padhy et al., 2014) and also to report poorer mental well-being (Li & Liu, 2018). From a stress standpoint, the cumulative cost to overcome modern-life stressors surpasses individuals' capacities and induces physiological disturbances and pathologies, especially in the most vulnerable population (e.g., minorities, low income and migrants).

As a byproduct of the SARS-CoV2 pandemic that started in late 2019, the exposure to modern-life stressors has worsened by the exacerbation of “captivity” conditions and economic emergencies resulting from sanitary restrictions. Social interactions and access to recreational spaces such as parks and playgrounds have been abruptly restricted in multiple countries, potentiating the negative effects of our habitat and lifestyle with unforeseen consequences. For instance, during the first 30 days after the pandemic declaration, there was an average reduction of 27% in physical activity worldwide, with some countries showing a maximum reduction of 48.7% in the following months (Tison et al., 2020). In terms of mental well-being, a Dutch population sample with no previous history of mental health disorders was assessed 3 months after the pandemic outbreak, showing increased loneliness and symptoms of depression and anxiety, with those having previous disorders reporting poorer stress-coping abilities to deal with the pandemic and enhanced fear about it (Pan et al., 2020). Similarly, the prevalence of different neuropsychiatric disorders, including anxiety and depression, substantially increased in Germany during the same period attributable to the psychological distress produced by the pandemic (Munk et al., 2020). Remarkably, during the third month after the pandemic declaration, people of France and Switzerland decreased vigorous physical activity but increased both sedentary time and moderate physical activity (e.g., walking) during their leisure time (Cheval et al., 2020). However, those with a greater increase in physical activity showed significant improvements in their physical health, whereas those with higher levels of sedentarism experienced poorer physical and mental health (Cheval et al., 2020). Intake of ultraprocessed food has also increased globally during the pandemic, especially in adolescents and in Latin American countries, with physical activity also being reduced during this time, particularly in women (Ruíz-Roso et al., 2020). For the younger population, fundamental “enriching” spaces such as daycare centres and schools suddenly closed, with up to 54% of the global student population suffering limited social contact, physical activity and sensorimotor and cognitive stimulation resulted from these closures (United Nations Educational, Scientific and Cultural Organization, 2020). In these scenarios, it is possible that the consequences of such environmental deprivations would be noticed over a long period of time, with a deeper impact on children from unprivileged families or deprived during the most sensitive developmental periods.

Although most evidence suggests that living in urban areas may encompass a larger exposure to multiple, long-lasting stressors with a potential negative impact on human health (Table 1); the opposite panorama can be also true, namely, that living in rural areas may produce similar (American Psychological Association, 2010; Bigbee, 1987, 1990; Carey et al., 2017; Elgar et al., 2003; Uphold et al., 2005) or even higher (Yeresyan & Lohaus, 2014) levels of stress than those observed in urban regions under certain circumstances. This effect is possible in two ways: Some rural areas offer few opportunities for academic, sports and cultural and leisure activities, which makes the lifestyle less stimulating and, to some extent, stressful. Conversely, some urban cities provide a broad range of all sorts of enriching activities, which may offset the negative effects associated with dense-populated areas. Unfortunately, only a small proportion of city dwellers are exposed in enough frequency and intensity to these stimuli (e.g., academic, sports, cultural and leisure activities) to be considered “enriching,” as all these factors are highly dependent on socioeconomic status. Some evidence suggests that it is not the level of perceived stress that matters most but the type of stressors typically presented in each region (American Psychological Association, 2010; Bigbee, 1987, 1990), which may lead to differential effects on health. In this sense, urban and rural regions should not be considered strictly opposites.

3 THE IMMERSION IN GREEN ENVIRONMENTS AS A STRATEGY FOR STRESS RELIEF IN HUMANS

The idea of using green areas and forests for alleviating modern-life stress is not new. During the 1980s, the concept of Shinrin-Yoku (i.e., forest-bathing) was coined in Japan for describing the immersion in natural spaces as mandatory for preventive health care (Hansen et al., 2017). However, studies investigating the effects of natural contexts on brain and behaviour became more systematic until the late 1990s and early 2000s. Recently, important advances in the development of technologies for assessing those effects in vivo have been made (Hansen et al., 2017; Summers & Vivian, 2018). The evidence provided by these studies indicates that being exposed to and performing activities in green environments such as parks and woods is associated with improvements in mental health and well-being (Mitchell et al., 2015). People living in areas with higher vegetation coverage and animal diversity are less likely to experience stress, anxiety and depressive symptoms, with those spending longer time outdoors getting more benefits out of these activities (Cox et al., 2017). Taking short walks (i.e., 15 min) in green-covered parks is perceived as restorative and improves positive mood compared to walking in urban “grey” areas (Stigsdotter et al., 2017). In middle-aged people, taking short walks (i.e., 20–30 min) in green spaces reduces self-reported stress and increases positive mood and feelings of subjective energy (Coventry et al., 2019). Short forest walks (i.e., 20 min) also induce subjective feelings of calmness and are perceived as comfortable, while contemplating the forest landscape after walking exerts the same effects and reduces the salivary concentration of the stress hormone cortisol (Park et al., 2007). Similarly, the sole contemplation of urban green spaces can exert positive effects on mood, as suggested by a frontal lobe alpha asymmetry—associated with positive emotionality—obtained from electroencephalographic (EEG) recordings during exposures to such spaces (Olszewska-Guizzo et al., 2020). In this line, elderly people walking within urban green spaces (i.e., 10–15 min) showed lower levels of low beta EEG activity—associated with increased alertness and arousal—compared to that evoked by walking over urban busy areas (Neale et al., 2020). Longer sessions of “green walking” (i.e., 1.5–2.5 hr) also enhance positive mood states, reduce anxiety and lower systolic blood pressure in middle-aged women (Chen et al., 2018). Similar “green” outdoor activities contributed to reducing maladaptive attentional focus on negative thoughts (i.e., rumination) and the associated activity in the subgenual prefrontal cortex in young adults (Bratman et al., 2015). When people are tracked during their daily activities, their spontaneous transit into urban green areas is associated with a higher sensation of emotional well-being that corresponds with the amount of vegetation coverage in their surroundings (Tost et al., 2019). At a brain level, those experiencing intense positive sensations while passing through those areas also showed lower prefrontal activity in response to negative socioemotional cues—a neural trait associated with mental health risk (Tost et al., 2019). Moreover, people living close to green spaces are more likely to engage in physical activity (Sallis et al., 2016), with those who exercise in parks (i.e., running) experiencing a reduction in arousal, frustration and engagement (i.e., direct attention) as shown by EEG measurements taken during the activity (Aspinall et al., 2015). In young adults, exercising in green spaces also induces a higher self-perception of robustness compared to exercising indoors, with people having increased positive expectations about the “green exercise” showing the greatest benefits (Flowers et al., 2018).

Besides the immediate, short-term benefits associated with human activities within urban green areas and forests, people moving from cities to more green environments experience a positive impact on their mental health over the following years after dwelling in the new location (Astell-Burt et al., 2014). Two-year-old children raised in contexts with high green coverage show better physical and mental development (Liao et al., 2019), as well as improved attention and reduced impulsivity when assessed 2 and 5 years later (Dadvand et al., 2017). In adults, exposure to urban green areas has a protective effect on individuals' well-being, with those living closer to parks and forests being less likely to commit suicide (Helbich et al., 2020). Moreover, greater availability of parks and urban green spaces during childhood and adulthood is associated with lower cognitive decline during the elderly (Cherrie et al., 2018). In contrast, upbringing (i.e., from birth to 15 years old) or living in urban areas (i.e., after 15 years old) increases the activation of the amygdala and the perigenual anterior cingulate cortex in response to social stress, with such activation increasing proportionally with the amount of time living in urban areas (Lederbogen et al., 2011). Moreover, the stress-evoked activity in those brain regions was lower in people living in smaller towns and even lower in those living in rural areas (Lederbogen et al., 2011). Interestingly, elderly population (i.e., 61–82 years) living close to forests (i.e., 1-km radius) showed better structural integrity of their amygdala—indicative of lower, age-related neurodegeneration—with those dwelling near to urban green areas showing no changes in the integrity of this region (Kühn et al., 2017). In this regard, changes in the amygdala and the perigenual anterior cingulate cortex activity resulting from living in urban areas might be associated with higher rates of mental health conditions observed in city dwellers as these brain regions participate both in the stress response and the aetiology of many neuropsychiatric disorders (Lederbogen et al., 2011). In Europe, people report higher life satisfaction when living in areas with easy access to green spaces and with high biodiversity (Methorst et al., 2020). Similar results are also reported in an Australian population sample, where the number of parks and the closeness to those areas is positively associated with mental well-being (Wood et al., 2017). Spending daily time enjoying nature is also associated with increased levels of positive affect and decreased levels of negative affect, even when most of this time is spent in social activities rather than in physical activity (Anderson, 2020). However, physical activity was still associated with increased levels of positive affect irrespective of whether it was performed while enjoying nature or not (Anderson, 2020). When analyzing data from multiple countries (Amano et al., 2018) and including different studies (Chiabai et al., 2020), the evidence is consistent about how much urban green spaces can benefit human health and well-being (Table 2). Yet different factors such as income, age and country economy modulate the magnitude and direction of the association between green exposure and health parameters (Amano et al., 2018; Chiabai et al., 2020).

The common factors of what constitutes a positive, enriching environment for humans seem to include access to natural green spaces, regular physical and recreational outdoor activities, positive and meaningful social interactions and minimal exposure to pollutants (e.g., artificial lighting, noise and particulate matter). For these environmental factors to have noticeable benefits on human health and well-being, they need to be consistently present during a sufficient time. Accordingly, formal programs of environmental enrichment (EE) in humans have been implemented to promote optimal development and to prevent or palliate different disorders by increasing social interaction, physical activity and sensorimotor stimulation (Queen et al., 2020). These EE programs used to include a combination of social contact, multisensory stimulation, games, hikes, gymnastics lessons, going up and down stairs, music and arts (Aronoff et al., 2016; Raine et al., 2003; Woo & Leon, 2013; Woo et al., 2015). These interventions have been successfully incorporated as an adjuvant treatment for antisocial behaviour, schizophrenia, autism spectrum and attention deficit/hyperactivity disorders (Ball et al., 2019; Raine et al., 2003) and also reduce age-related cognitive decline and increase well-being both in healthy and unhealthy individuals (de Vries et al., 2003; Queen et al., 2020; Santini et al., 2020). Exercise and physical activity are two fundamental components of EE (Queen et al., 2020; Simpson & Kelly, 2011) capable of potentiating brain functions in healthy and unhealthy humans (Fernandes et al., 2017; Stillman et al., 2020). As the “green factor” (i.e., the contact with nature) positively affects brain function and mental health (Kuo, 2015; Mantler & Logan, 2015), it has been recently suggested that it should constitute a key component of all EE programs in humans (Queen et al., 2020). Although scarce, some evidence supports the positive interaction between exercising or doing physical activity and the green factor, suggesting that its occurrence in natural environments might produce even greater benefits on humans' well-being (Aspinall et al., 2015; Flowers et al., 2018; Neale et al., 2017; Stigsdotter et al., 2017).

4 RAT MODELS OF HOUSING CONDITIONS RESEMBLE DIFFERENT SCENARIOS OF HUMANS LIVING IN URBAN AREAS

Laboratory animals also benefit from the immersion in environments resembling their wild habitat. As in humans, physical activity, social interaction and sensorimotor stimulation are the main components of EE (e.g., large multilevel cages filled with different toys and many conspecifics) in rats (Figure 1g), exerting multiple changes in brain and behaviour (Crofton et al., 2015). Seminal studies back in the 1950s and 1960s showed that rearing rats in EE alter brain chemistry and morphology compared with group-housed individuals living in standard laboratory cages (i.e., standard housing: SH; Figures 1f and 2b; Krech et al., 1960; Rosenzweig et al., 1962). Although initially received with skepticism (Rosenzweig et al., 1972), such findings were extensively replicated from then on and have served as a foundation for the study of behavioural and brain changes in response to experience (Krech et al., 1960; Rosenzweig et al., 1962; Van Praag et al., 2000). Mounting evidence shows that EE exerts a wide range of physiological effects including changes in neurotransmitters concentration (e.g., increased norepinephrine or serotonin concentrations on prefrontal cortex, hippocampus and ventral striatum; Brenes et al., 2008, 2009), modifications of neuronal structure (Landers et al., 2011; Leggio et al., 2005) and the enhancement of hippocampal neurogenesis, cell survival and long-term potentiation (Brenes et al., 2016; Kempermann et al., 1997; Stein et al., 2016). In addition, a huge amount of studies indicates that EE alters the expression of genes regulating multiple neural mechanisms such as plasticity and epigenetic profile (Hüttenrauch et al., 2016; Rampon et al., 2000; Rojas-Carvajal et al., 2020; Sun et al., 2010). These structural, physiological and functional hallmarks underlie the EE-induced behavioural enhancements on many learning and memory paradigms (Brenes et al., 2008; Leggio et al., 2005; Pamplona et al., 2009; Rojas-Carvajal et al., 2018; Sampedro-Piquero et al., 2014; Sun et al., 2010) and on social behaviours (McQuaid et al., 2018). In addition, EE ameliorates neurobehavioural outcomes associated with neurological disorders such as Alzheimer, Parkinson's and Huntington's disease (Jeong et al., 2011; Jungling et al., 2017; Mazarakis et al., 2014). Remarkably, EE also exerts a strong modulation of the stress response (McQuaid et al., 2018; Moncek et al., 2004; Peña et al., 2009) and induces antidepressant- and anxiolytic-like effects (Benaroya-Milshtein et al., 2004; Brenes et al., 2008, 2009; Peña et al., 2006, 2009), with both protective and remediating actions against stress-induced behavioural alterations (Hellemans et al., 2004; Lehmann & Herkenham, 2011; Du et al., 2012; Grippo et al., 2014; Vega-Rivera et al., 2016). As a comprehensive description of biobehavioural changes induced by EE is beyond the scope of this work, please review the literature suggested here (Gelfo et al., 2018; Girbovan & Plamondon, 2013; Gubert & Hannan, 2019; Kempermann, 2019; Sampedro-Piquero & Begega, 2017; Simpson & Kelly, 2011; Smail et al., 2020).

Traditionally, the positive effects of EE are interpreted as consequences of the increased and positive stimulation obtained from living in those environments. However, the evidence rather suggests that SH induces a myriad of deficits that are ameliorated by exposing animals to environments vaguely mimicking some features of their natural habitat (Nithianantharajah & Hannan, 2006). So much so that short EE periods (e.g., during some hours, days or randomly exposed for weeks) or minimally enriched cages (e.g., two rats enriched in standard cages) are still able to produce behavioural and neural improvements compared with SH (Ashokan et al., 2016; Brenes et al., 2016; Koe et al., 2016; Rojas-Carvajal et al., 2020; Stein et al., 2016; Widman & Rosellini, 1990; Zaias et al., 2008). In this line, rearing rodents in SI (Figures 1e and 2a) for extended periods aggravate the deleterious effects of captivity, which may lead to a vast array of negative behavioural and neural consequences even when compared with SH animals (for review see Biggio et al., 2019; Fone & Porkess, 2008; Hueston et al., 2017; Lukkes et al., 2009; Mumtaz et al., 2018; Zorzo et al., 2019). SI is a low intensity but usually prolonged stressor capable of disrupting the hypothalamic-pituitary-adrenal (HPA) axis activity and induces maladaptive stress responses (Boero et al., 2018; Serra et al., 2005). At the physiological level, SI alters the activity of several neurotransmitters and neuromodulators (Heidbreder et al., 2000; Jones et al., 1990; Scaccianoce et al., 2006), as well as the expression profiles of many genes associated with neural plasticity and other relevant cellular processes (Murphy et al., 2010; Sun et al., 2018). Moreover, SI negatively affects dendritic arborisation, spine density and morphology (Biggio et al., 2019; Silva-Gómez et al., 2003; Wang et al., 2012), as well as neuronal functions including long-term potentiation (Quan et al., 2010; Roberts & Greene, 2003). Behaviourally, SI animals show hyper-reactivity to novelty (Del Arco et al., 2004; Varty et al., 2000), a reduction of memory and visuospatial learning performance (Bianchi et al., 2006; Hellemans et al., 2004) and alterations in reward-related processes (Hall et al., 1997; Jones et al., 1990). Furthermore, increased anxiety- and depressive-like behaviours are consistently found in SI rats (Amiri et al., 2015; Brenes et al., 2008, 2020; Lukkes et al., 2009; McCool & Chappell, 2009), acting as risk factors for the development of particular endophenotypes of several stress-related disorders such as drug dependence (Butler et al., 2016; Ding et al., 2005; McCool & Chappell, 2009), psychosis and schizophrenia (Alquicer et al., 2008; King et al., 2009). It is worth noting that when SI animals are socially (e.g., temporarily housed in SH) or physically (e.g., given access to treadmill exercise) enriched, the adverse effects of SI are ameliorated or reversed (Biggio et al., 2019; Brenes et al., 2020). In other words, providing animals with the minimal substrates and materials promotes the emission of species-specific behaviours (e.g., foraging, forming social hierarchies, galloping, building nests and hideouts) capable of reducing the stress and impoverishment of captivity.

Based on the inoculation stress hypothesis, chronic exposure to EE will inoculate individuals allowing them to develop resilience against future stress exposures while enhancing numerous brain functions (Crofton et al., 2015). From this perspective, EE can be considered as a source of mild stress that puts in motion a wide range of species-specific behaviour that cannot be otherwise triggered under SH or SI conditions. In fact, the three main components (i.e., physical activity, social interaction and sensorimotor stimulation) of EE are well-known for inducing a stress response increasing the corticosterone levels (Fediuc et al., 2006; Fokkema et al., 1988; Larsson et al., 2002; Piazza et al., 1991; Raz, 2013). At the same time, these components are rewarding activities evoking positive emotionality in rats (Heyse et al., 2015; Rojas-Carvajal et al., 2020), which places the EE as a form of positive stress (i.e., eustress) with many beneficial effects (Bassett & Buchanan-Smith, 2007; Koolhaas et al., 2011; Sambrook & Buchanan-Smith, 1997). The latter does not mean that living in an EE cage is not a demanding activity. Indeed, rodents should learn to deal with a larger and more complex environment including objects and materials that change from time to time. In such a complex environment, rodents are also enabled to better delimitate the territory by building nests and hiding food on certain “home base” locations. The fact that rats can be manipulable objects and materials at different configurations induces exploration and increases the ability to control and predict what is happening in the surroundings, which in turn reduces anxiety and improves animals' welfare (Abou-Ismail & Mendl, 2016; Bassett & Buchanan-Smith, 2007; Darwish et al., 2001). Also, food and water in the EE cage might not be as easily accessible as in SH cages, and social structure and hierarchy can become rather intricate as the number of mates per cage is much higher. Such opportunities to display foraging behaviours and having diverse social interactions can buffer the HPA axis and ameliorate stress responses (Giralt & Armario, 1989; Hoffman-Goetz et al., 1992), which is known to increase well-being (Uvnäs-Moberg, 1997). Physical activity, on the other hand, provokes multiple physiological alterations including changes in monoaminergic neurotransmission (Arnold et al., 2020; Dremencov et al., 2017), a reduction of the HPA axis activity and stress responses (Campeau et al., 2010; Hare et al., 2014; Pietrelli et al., 2018) and the induction of neural plasticity (Dahlin et al., 2019; Stranahan et al., 2007; Trinchero et al., 2019). As a result, physical activity is related to lower anxiety- and depressive-like behaviours (Brenes et al., 2020; Fuss et al., 2010; Greenwood et al., 2003; Pietrelli et al., 2018) and enhanced learning and memory (Diederich et al., 2017; Gibbons et al., 2014). In fact, physical activity and exercise are the most studied components of EE both in humans and rats, with positive effects well-documented in both species (Bidzan-Bluma & Lipowska, 2018; Bliss et al., 2021; Chen et al., 2020; Heinze et al., 2021; Heyse et al., 2015). Interestingly, a recent report showed that EE exerts broader and stronger antidepressant-like effects compared to those induced by exercise alone, which still induces greater benefits than those caused by the administration of the antidepressant fluoxetine (Brenes et al., 2020). Moreover, social and physical EE induces slightly greater improvements on novelty habituation, learning and memory, brain plasticity and ultrasonic communication compared to those exerted by either of these components alone (Brenes et al., 2016). Conversely, the impoverished environment offered by SH and SI prevents the animals from fulfilling many of their ethological needs and deprived them of essential sources of stimulation for optimal development. Thus, this passive but prolonged source of distress imprints deleterious effects on rats. Furthermore, as “natural” or “seminatural” stimuli within EE cages evoke a more complex behavioural repertoire than artificial items (Lambert et al., 2016; Rojas-Carvajal et al., 2020), it is possible that the “green factor” also plays a key role in the effects induced by EE.

5 CURRENT ANIMAL MODELS OF STRESS MAY NOT REPRESENT THE STRESS DERIVED FROM OUR MODERN LIFESTYLE

Considering that rodents are normally housed in stressful and impoverished conditions (i.e., SH or SI), it is not surprising, therefore, that most preclinical models of stress include severe stressful stimuli to provoke stress responses and stress-related sequels prominent enough to be distinguishable from those caused by their standard living conditions. If SI rodents are used as control groups, the chances to detect differences in the stress response are even lower. Because there is only a small room for further deterioration, stress models have come to exhibit weak or moderate effects with sometimes lack of consistency within and among labs despite using severe stress protocols (Willner, 2017a). In this sense, stress models have unintentionally come to represent the extreme spectrum of the stress response, which might correspond neither with the natural response of rodents in their wild habitat nor with the human stress response that they intended to model in the first place. Due to the severity and chronicity of the stress protocols (Table 3), most models come closer to represent the effects of the so-called physical or systemic stress (Godoy et al., 2018), which comprises unpredictable events such as diseases, accidents, traumas and bereavement (Juster et al., 2010; McEwen & Wingfield, 2003). This claim becomes conspicuous after an in-depth analysis of the stressors used in common stress models (Table 3). A huge body of evidence on neurobehavioural responses has been produced over the last decades by exposing animals acutely or chronically to highly disturbing stimuli such as forced swim (Molendijk & de Kloet, 2019), electric shocks (Bali & Jaggi, 2015), physical immobilisation (Glavin et al., 1994), cold or heat (Cure, 1989; Fukuhara et al., 1996), noise (Campeau & Watson, 1997), predator odours (Adamec et al., 2005), alteration of illumination rhythms (Vernikos-Danellis et al., 1970) and deprivation of food, water or sleep (Aguilera et al., 1993; Andersen et al., 2005; Dallman et al., 1999). In addition, a number of highly stressful social paradigms involving territory defence (e.g., resident-intruder paradigm), hierarchy formation (e.g., social defeat) or SI have also been widely and successfully used (Blanchard et al., 2001; Fone & Porkess, 2008). Even the chronic unpredictable mild stress model, a model, allegedly designed to study the effects of moderate stress, includes several of the aforementioned stressors such as SI, food/water/sleep deprivation, immobilisation, forced swim and cold exposures (Willner, 2017a, 2017b). Many of these stressors produce substantial physiological alterations, induce a negative energy balance (i.e., body mass loss), trigger flight/fight/freeze or despair responses and are associated with type 1 allostatic overload (Godoy et al., 2018; Lupien et al., 2009; McEwen & Wingfield, 2003). Although investigating the neurobiological underpinnings of this particular stress response is of special relevance, it does not necessarily translate to the most common stress situations provoked by our modern lifestyle and living conditions. Actually, apart from devastating cases such as torture, starvation or the atrocities experienced in wartime, only few of the above-referenced stressors match with those experienced by humans in daily life. Most people will face psychological instead of physical stress, including social instability (e.g., couples breakdown, family disintegration, bullying and problems at work and social violence), economic issues, lack of opportunities and reduced prospects and more recently, the constant bombing of (miss)information, violence and pressure through social media and the internet.

Based on most theoretical models of psychopathology (Boyce & Ellis, 2005; Monroe & Simons, 1991; Zubin & Spring, 1977), vulnerability to psychiatric disorders is conferred by a complex interplay between genetic predispositions and early environmental factors (e.g., stress), which gradually aggregate to other precipitating events during development (Kendler et al., 2004; Kessler et al., 2010). Mounting evidence indicates that early childhood and adolescence are the most sensitive periods for stress to induce more profound and enduring sequels in brain and behaviour, which translates into higher risk for mental health disorders (Chaby et al., 2017; Lupien et al., 2009; Nanni et al., 2012). For example, adversity during childhood increase risk of major depression (Heim et al., 2010; Nanni et al., 2012; Nelson et al., 2002), bipolar disorder (Gilman et al., 2012; Horesh et al., 2011), anxiety disorders (Heim et al., 2010; Sareen et al., 2013), drug dependence (Kendler et al., 2000; Nelson et al., 2002), schizophrenia (Akbarian, 2010; Brown, 2011) and Borderline personality disorder during adulthood (Bierer et al., 2003; Golier et al., 2003). In general, susceptibility factors crystalize during early stages in a subgroup of vulnerable subjects until a first episode emerges. Due to the overwhelming evidence linking stress with psychopathology, it can be said that the enormous interest in modelling stress in laboratory animals stems from the fact that it is the most important risk factor modulating mental health, especially during early life. Thus, animal models including stress as an etiological factor should be implemented in young subjects for translational and validity purposes. Surprisingly, the vast majority of the preclinical studies are conducted in adult rodents.

It is true that the severity of the adverse events (e.g., sexual, physical, psychological abuse and negligent care) will substantially increase risk of mental disorders (Cong et al., 2012; Kendler et al., 2000). For instance, the average age for the first diagnosis of mood disorders, including depression, is 30 years (Kessler et al., 2005), but the occurrence of sexual abuse shifts age of onset of psychiatric disorders to early adolescence, increasing the duration and severity of the disorder as well (Cong et al., 2012; Kendler et al., 2000; Teicher et al., 2009). Subjects who had onset of sexual abuse between 3 and 13 years of age developed depression or posttraumatic stress disorders (PTSD) approximately 7–9 years later (Teicher et al., 2009). However, the lifetime prevalence of nonsexual, severe traumatic events (e.g., held captive/tortured/kidnapped and war experience) is less than 2%, whereas sexual-related events (e.g., rape and sexual assault) show a prevalence of approximately 9% in women and 1%–3% in men (Breslau et al., 1998). Although up to 90% of the population is exposed to moderate trauma during their lifetime (e.g., unexpected death of a loved one, witnessing injury or shocking events and suffering a motor vehicle accident; Breslau et al., 1998), the worldwide lifetime and 12-month prevalence of PTSD in trauma-exposed individuals is only 5.6% and 2.8%, respectively (Koenen et al., 2017). Even in persons who met criteria for the PTSD, the remission rate during the first year is 66% (Breslau et al., 1998). In a 10-year follow-up study, early separation or childhood abuse increased risk of major depression only in those with family history of mood disorders (Zimmermann et al., 2008). Likewise, in cases with less severe forms of child sexual abuse, the relationship between adverse events and mental disorders was modest or absent, indicating that reaching a certain level of distress is required for most psychiatric disorders to manifest (Kendler et al., 2000). All this evidence suggests that (1) suffering a traumatic event is not sufficient to develop mental disorders, (2) most common psychiatric conditions would originate from the combination of mild stressors with biological susceptibility factors that accrue during the lifespan and (3) most people used to learn, to a certain extent, how to master most common forms of stress without showing long-lasting health consequences (Rutter, 2012). Thus, implementing stress models with severe stimuli would not necessarily resemble the etiological factors associated with the vast majority of mental health conditions, except for PTSD and other trauma-related disorders. In terms of timing and duration, many stress protocols are scheduled to vary randomly in frequency and intensity according to the view that stressors in humans are rather unpredictable and of an ever-changing nature. However, most stressors associated with psychopathology originate in the household, during early life, and are perpetuated by parents and close relatives, being therefore easily foreseeable and less variable than normally thought. Some of these family-related stressors include marital discord, low social class/income, paternal criminality, maternal mental disorder and foster placement (Amrock & Weitzman, 2014; Biederman et al., 1995; Björkenstam et al., 2017; Kendler et al., 2003; Reid & Crisafulli, 1990; Rutter et al., 1975; Tseng et al., 2017). Other stressors come from the community and the educational systems—such as bullying and discrimination due to ethnicity and sexual orientation—that have been associated with negative mental health outcomes (Berger & Sarnyai, 2015; Lee et al., 2016; Pachter et al., 2018). In general, those stressors coming from the nearby social surroundings are not severe traumatic events but mild, identifiable and predictable situations lingering on over the years. Therefore, it is the aggregation of minor stressful events, rather than the presence of any single factor, that impairs development and confers vulnerability for mental health disorders (Rutter, 2012; Rutter et al., 1975).

Although preclinical research in stress should have been considering all these abovementioned elements when developing animal models of stress, there is a lack of matching between clinical and preclinical research that could undermine the validity and reproducibility of the stress models reducing in consequence their translatability to humans. Classically, animal models should meet at least three criteria to be considered valid (Belzung & Lemoine, 2011; Willner, 1984). Face validity refers to the phenomenological similarities (i.e., symptomatology) between the model and the human disease or condition. Construct validity focuses on the theoretical rationale of the model, that is, the similarity in the etiological processes underlying the development of the disease or condition. Predictive validity is the extent to which a model allows screening potential treatments and implies the ability to discriminate between old and new therapeutic compounds. Although reformulated and extended in recent years (Belzung & Lemoine, 2011), these criteria still remain at the centre of the discussion of the validity of animal models of neuropsychiatric disorders. Because different stressors activate differently the stress response and induce different downstream physiological and behavioural alterations (Lupien et al., 2009), it becomes of great importance to use stressors as close as possible to those affecting humans; otherwise, face, construct and predictive validity of the model may be seriously jeopardised.

6 MODELLING THE STRESS OF OUR MODERN LIFESTYLE IN RATS

7 CONCLUSIONS

Stress is a common risk factor for mental, cardiovascular and metabolic diseases, which are more frequent in urban areas than in rural areas. The impoverished environment and the demanding lifestyle associated with living in cities might constitute our species' primary source of stress. Objective or perceived SI, poor housing and working conditions, sedentarism, obesity and limited access to recreational activities and exercise are some of the main factors characterizing our new “normal” habitat. In contrast, social and physical activities, exercise and living close to and performing activities in green environments have systematically shown to improve physical and mental health. Despite the evidence that our stressful lifestyle results from the accumulation of minor, recurrent and often predictable events (type 2 allostatic load), most stress models have focused on studying the short-term consequences of severe and unusual stressors (type 1 allostatic load). The reason may be that SH or SI, which are used as control groups, already produces a basal stress condition difficult to exacerbate if stressors are not intense enough. According to the well-known EE effects, it is clear that rearing animals in SH or SI induce a complex stress-related phenotype that might represent some negative features of our modern habitat and lifestyle, whereas EE might represent the “average” living conditions of urban dwellers. In this regard, we propose using a more standard and simplified form of EE (Figure 2c) as a new reference condition (i.e., the control group) against which SH and SI will be compared (i.e., the stress groups). These housing conditions could be combined with other stressors to test particular hypotheses. In general, the housing conditions model would be easy to implement and replicate, would produce more translational results and accomplish some ethical commitments by promoting the refinement of procedures to model stress, diminishing animal suffering, enhancing animal welfare and eventually reducing the number of experimental animals needed.

CONFLICT OF INTEREST

The authors declare no competing interests.

AUTHOR CONTRIBUTIONS

MR-C, JB and AS-C conceived the subject of the manuscript. MR-C prepared a draft version of the manuscript. JB and AS-C wrote the final version, which was reviewed and approved by all authors.

ACKNOWLEDGEMENTS

This work was funded by the grants 723-B9-197, 837-B8-123, 837-B7-603, and 837-C0-606 of the University of Costa Rica. A special acknowledgment to Daniela Sandoval, B.A who confectioned the artwork used in the manuscript. [Correction added on XX April 2021, after first online publication: Acknowledgements have been added.]