The effects of early-life and intergenerational stress on the brain

Abstract

Stress experienced during ontogeny can have profound effects on the adult phenotype. However, stress can also be experienced intergenerationally, where an offspring's phenotype can be moulded by stress experienced by the parents. Although early-life and intergenerational stress can alter anatomy, physiology, and behaviour, nothing is known about how these stress contexts interact to affect the neural phenotype. Here, we examined how early-life and intergenerational stress affect the brain in eastern fence lizards (Sceloporus undulatus). Some lizard populations co-occur with predatory fire ants, and stress from fire ant attacks exerts intergenerational physiological and behavioural changes in lizards. However, it is unclear if intergenerational stress, or the interaction between intergenerational and early-life stress, modulates the brain. To test this, we captured gravid females from fire ant invaded and uninvaded populations, and subjected offspring to three early-life stress treatments: (1) fire ant attack, (2) corticosterone, or (3) a control. Corticosterone and fire ant attack decreased some aspects of the neural phenotype while population of origin and the interaction of early-life stress and population had no effects on the brain. These results suggest that early-life stressors may better predict adult brain variation than intergenerational stress in this species.

1. Introduction

Individuals must contend with unpredictable or adverse stressors in their environment such as predation, agonistic encounters with conspecifics and rapid abiotic change. These stressors can present immediate and potentially continual threats to survival and thus represent strong selective agents, favouring individuals that direct effort towards managing those stressors. In most vertebrates, exposure to stressors activates the hypothalamic–pituitary–adrenal (HPA) axis, inducing the release of glucocorticoid hormones (GCs) that organize and drive appropriate physiological and behavioural responses. However, prolonged stress and release of GCs can lead to allostatic overload and potentially come at the cost of other fitness-promoting activities such as reproduction, cognition or immunity [1–4]. Thus, in most vertebrates, there appears to be a trade-off between energy devoted to an intense or prolonged stress response and other less immediately critical processes.

These trade-offs are exemplified by the relationship among stress, cognition and the brain. When an individual is chronically stressed, energy devoted to maintaining cognitive functioning and the brain may decrease [5,6]. This can occur within an organism's lifespan, where exposure to stressors during early life can affect cognition and the neural phenotype, and these changes can persist into adulthood [7–9]. However, these neural responses may also be adaptive, as they allow individuals to mount the appropriate response to environmental stressors while directing less energy towards less critical processes such as cognition [10].

Similarly, studies in rodents and fish have demonstrated intergenerational effects of stress, where stress experienced by a parent can affect the behaviour and physiology of subsequent offspring, even if the offspring do not directly experience the stressor [10–17]. This is assumed to be an adaptive response, as it prepares offspring for a consistently stressful environment [18]. Studies on the effects of intergenerational stress on offspring neural attributes are not as well studied as other factors (e.g. survival, birthweight), yet they support that stress experienced by parents, even before conception, can mould offspring neural attributes. However, the directionality of intergenerational stress effects on the brain varies [13,19–22]. Because of the relative paucity of studies examining intergenerational stress and the brain, it remains equivocal how and in what contexts intergenerational stress modulates the neural substrate.

The effects of early life and intergenerational stress on the brain are primarily studied in the mammalian hippocampus, a plastic brain region linked to modulation of the stress response, as well as spatial learning and memory. The hippocampus is critical in regulating the stress response, by inhibiting the HPA axis and decreasing stress response duration [23–25]. Mechanistically, GCs can exert their effects in the hippocampus via the high density of GC receptors on hippocampal neurons [26–28]. However, severe stress or abnormally high levels of GCs lead to neurodegeneration in the hippocampus, concomitant with impaired cognition and a protracted stress response [29], and these effects can last into adulthood [30]. Because of the plasticity of the hippocampus and the robust relationship between the stress response and the mammalian hippocampus, this area of the brain is often the target of studies on the neural effects of stress.

While some studies have elucidated some of the outcomes of early life and intergenerational stress on the hippocampus, most notably in model rodent species, there is little understanding of the fundamental functional and evolutionary significance of this relationship across the vertebrate lineage. Understanding this relationship using a comparative approach is necessary to identify the generalizable or divergent effects of stress on brain structure and function [31]. Reptiles may serve as a reference group for comparative endocrinological and neurobiological studies, as they are the most basal extant amniote and therefore represent an important evolutionary link in the vertebrate lineage. In reptiles, the structure and physiology of the HPA axis is like other vertebrates [32,33], suggesting a conserved stress response. However, the behavioural and physiological stress response in reptiles is also subjected to modulation based on a myriad of factors including season, reproductive state and sex [34,35]. Similarly, a recent review suggests that HPA activation in non-model vertebrates vary quite significantly based on species, individual and season, with few generalizable patterns in the stress response in the wild [36]. Therefore, the evolutionary and adaptive significance of the stress response has not been fully delineated, particularly in non-model species in the field.

Likewise, the evolution of brain structures and functions continues to be intensely debated [37]. While the reptilian brain does not have a true hippocampus, the medial and dorsal cortices of reptiles exhibit structural and functional similarities to the hippocampi of other vertebrates, including functions related to the stress response [38,39]. The neurons of the cortices also have a high density of GC receptors, and chronic stress can modulate receptor expression and neurotransmitter release in the cortices [40–43]; however, cortical modulation of the stress response is unknown. Taken together, as in other vertebrates, the stress response and the reptilian cortices are likely to be linked, although there is a relative paucity of studies delineating this relationship.

Currently, there are no studies examining the effects of intergenerational stress on the reptilian cortices. More broadly, to our knowledge, there are also no studies in any vertebrate species examining how intra- and intergenerational stress interact to affect the resultant hippocampus or hippocampal homologue. To address these gaps, this study examined the interaction between intra- and intergenerational stress contexts by leveraging a natural system of native lizards and invasive fire ants. The eastern fence lizard (Sceloporus undulatus) is a wide-ranging species found in the eastern United States [44]. For over 70 years, some populations of eastern fence lizards have co-occurred with non-native fire ants (Solenopsis invicta) [45,46]. In these populations, fire ants frequently attack lizards, and their venom can kill adults and juveniles [47]. These potentially lethal interactions induce a strong stress response in the lizards, reflected by increased circulating GCs after an attack [48]. Further, lizards in invaded populations have behavioural, morphological and physiological adaptations that correspond to defence against fire ant attacks, some of which are attributable to the intergenerational effects of stress [49–54]. Thus, in invaded populations, environmental stressors in the form of fire ant attacks correlate with adaptive phenotypic traits associated with defence against fire ant attack, and these traits arise both intra- and intergenerationally.

However, it is unclear if fire ant attacks during development have organizational effects on the cortical phenotype. Because attacks are a common, immediate threat to survival, lizards may prioritize mounting a stress response at the cost of cortical maintenance and growth. Similarly, intergenerational effects in the form of population of origin (invaded versus uninvaded) could affect the cortical phenotype by preparing offspring for a stressful environment. Finally, intra- and intergenerational stress may interact to induce changes in the cortices that may not be apparent by examining each stress context separately. Thus, the goal of this study was three-fold: (1) examine the effects of early-life, non-lethal stressors (either fire ant attack or GC application) on the adult medial and dorsal cortices; (2) determine if populations that have evolved under different stress contexts exhibit differential cortical phenotypes; and (3) ascertain if intra- and intergenerational stress contexts interact to shape the adult cortical phenotype. Further, our design allows us to determine if these differences are mediated by GCs alone or in conjunction with physical attack by fire ants. We predicted that early-life stress exposure in the form of GCs and fire ant attacks would decrease the adult cortical phenotype, specifically volume and neuron soma volume in the medial cortex. However, we also predicted that the effects of intergenerational stress may differentially modulate this relationship [51,54], with no a priori assumptions on directionality.

2. Material and methods

(a) . Subjects

Lizards were from a subset of animals used in a previous study on early-life stress; thus collection, housing and stress treatments were identical to that study [51]. Briefly, gravid females (n = 86) were collected from six populations in the southeastern United States during April and May of 2012. In three of these populations, lizards have co-occurred with invasive fire ants since the 1950s (invaded) while the other three populations had no record of invasion by fire ants at the time of collection (uninvaded) [45,46]. Females were brought into the laboratory and placed in plastic oviposition enclosures (56 × 40 × 30 cm). Each enclosure contained a shelter, a water dish and moist sand for oviposition. Ambient light was on a 12 : 12 light : dark schedule, supplemented with 6 h of a 60 W incandescent bulb for thermoregulation at one end of the enclosure during daylight hours. Crickets dusted with calcium and vitamins were offered every other day and water was available ad libitum.

Oviposition enclosures were checked twice a day and eggs were immediately collected. Eggs from each clutch were placed in a plastic container (11 × 11 × 7.5 cm) containing moist vermiculite, covered in plastic wrap and sealed with a rubber band [55]. Containers were placed in an incubator (29°C ± 1°C) and rotated every other day to prevent within-incubator effects. Containers were checked twice daily, and any hatchlings were immediately removed and toe-clipped for individual identification. Hatchlings were housed in enclosures similar to the oviposition enclosures except the floor was lined with paper towels; all other husbandry aspects were identical to those used in the oviposition enclosures. Each hatchling enclosure housed six individuals hatched within 4 days of each other, two from each treatment group (see below). Hatchlings from uninvaded and invaded sites were not housed together.

(b) . Stress treatments

Using a split-clutch and matched-sex design, hatchlings were assigned to one of three treatment groups: control, corticosterone (CORT, the primary adrenal GC in reptiles [56,57]) and fire ants (FA). Stress treatment regimens were performed weekly between 2 and 43 weeks of age. Once a week, individuals were placed in a sand-lined arena for 30 s (either with or without fire ants) followed by a pipetted topical application of 3 µl of sesame oil (either with or without CORT) to their backs. In the control treatment (n = 9), lizards were placed in the empty arena for 30 s followed by application of the sesame oil vehicle. Doing so controlled for the effects of handling, arena exposure and CORT vehicle exposure. In the CORT treatment (n = 13), lizards were placed in the empty arena for 30 sec to control for handling effects and arena exposure. CORT ( ≥ 92%, Sigma C2505, St Louis, MO) dissolved in sesame oil was applied immediately after arena exposure [58]. Dosages ranged from 0.6 to 1 µg CORT g^–1 based on body mass. This dosage is ecologically relevant, as it produces plasma CORT level profiles similar to the CORT levels found in wild lizards 30 min after exposure to fire ants [49,50]. Finally, in the FA treatment (n = 14), a lizard was placed in the arena with 15–20 fire ants. Ants were allowed to attack the lizard and the trial ended 30 s after the first ant contacted the subject. After 30 s, all ants were removed from the lizard. This treatment represents an ecologically relevant, non-lethal stressor that is sufficient to raise CORT levels to those measured in the field [50]. The sesame oil vehicle was then applied to control for the effects of handling and the CORT vehicle.

(c) . Brain analyses

After subjects reached reproductive maturity based on snout–vent length (SVL) [45,59]) and 24–30 weeks after treatments ended (variation due different hatch dates), all individuals were sacrificed with a lethal overdose of tricaine methanesulfonate on the same day. Euthanasia consisted of two intracoelomic injections [60,61]. The first injection was a 500 mg kg⁻¹ 1% sodium-bicarbonate buffered solution (dosage based on weight) to induce general anaesthesia. Euthanasia was completed by a second lethal injection of 1.0 ml unbuffered 50% solution. Heads were removed and post-fixed in 10% methanol-free formalin (from paraformaldehyde) for 2 weeks. The brains were then removed and cryoprotected in 15% sucrose for 24 h, followed by 30% sucrose solution for an additional 24 h, then stored at −80°C until sectioning. Brains were sectioned on a cryostat (Leica CM 3050S, −20°C) in the coronal plane every 50 µm and every other section was mounted and Nissl-stained with thionin. Slides were coded thus neural attributes were measured blind to population of origin and stress treatment.

We used standard, unbiased stereological techniques (StereoInvestigator, Microbrightfield, Williston, VT; Leica M400B microscope) optimized for this species (all coefficient of errors <0.011 [62]) to estimate medial and dorsal cortical volumes and neuron soma volume in the medial cortex. We also measured telencephalon volume (minus medial and dorsal cortical volumes) to assess if population of origin and stress treatment effects were specific to the cortices or representative of global changes within the brain [63–65]. The left and right hemispheres of the medial cortex, dorsal cortex and the remainder of the telencephalon were contoured in their entirety at 5×. Cortical volume estimations were generated with the Cavalieri procedure [66] using a 200 µm grid. Finally, we estimated medial cortical neuron soma volume with the optical fractionator nucleator procedure using four rays and a 250 µm grid, at 100×, measuring three randomly selected cells per counting frame. Due to histological artefacts, not all cortical parameters could be analysed in all subjects.

(d) . Statistical methods

There were no significant volumetric differences between the left and right hemispheres for any of the neural variables (paired t-tests: medial cortex, t₃₅ = −0.424, p = 0.674; dorsal cortex, t₃₄ = −1.224, p = 0.229; remainder of the telencephalon, t₃₆ = −1.494, p = 0.144; medial cortex neuron soma volume, t₃₃ = 0.082, p = 0.467). Consequently, left and right hemispheres were summed (or averaged for neuron soma volume) and subsequent analyses were performed on the pooled data. Two individuals were missing either a left or right cortex. Because there were no differences between left and right hemispheres for any of the measured parameters in the other subjects, data for total cortical volume and soma size were doubled from the unaffected cortical hemisphere.

Homogeneity of variances was assessed with Levene's test, and all data conformed to the assumption (all p > 0.05). We used a general linear model (GLM) to assess differences in the remainder of telencephalon volume with SVL as a covariate to ascertain the effects of population of origin and stress treatment on overall brain size, outside of allometry. One individual from the FA treatment did not have an SVL measurement and was eliminated from this analysis. We also used GLMs to assess the effects of population of origin and stress treatment on the volumes of the dorsal and medial cortices, using the remainder of the telencephalon volume as a covariate. Doing so assured that these results were specific to changes in the cortices of the telencephalon, rather than global changes in the brain. To ascertain variation in medial cortex neuron soma volume based on population of origin and stress treatment, we used GLM with medial cortex volume as a covariate. Finally, we performed analyses without covariates to ascertain any statistical differences when not controlling for the covariates. As previously reported, baseline CORT was not associated with population of origin or stress treatment [51], so CORT was not used as a covariate in the brain analyses. All analyses were conducted with SPSS (version 25.0, IBM Corp., Armonk, NY) with α = 0.05; Tukey's HSD post-hoc comparisons were applied when appropriate.

3. Results

There were no statistically significant differences in the volume of the remainder of the telencephalon due to stress treatment, population of origin or the interaction between stress treatment and population of origin (SVL covariate: F_1,28 = 1.759, p = 0.196; stress treatment: F_2,28 = 1.749, p = 0.192; population of origin: F_1,28 = 2.369, p = 0.135; interaction: F_2,28 = 0.029, p = 0.971; figure 1).

Figure 1.

Average telencephalon volume (mm³) ± SE across three early-life stress treatment groups (control, individuals treated with exogenous corticosterone (CORT), and individuals exposed to sublethal fire ant attack), originating from two different populations (fire ant invaded versus uninvaded). No statistically significant differences were found across stress treatments, population of origin, or the interaction of the two (all p > 0.135).

When controlling for the remainder of the telencephalon (telencephalon covariate: F_1,29 = 31.508, p < 0.001), we found that stress treatment had a significant effect on the medial cortex volume (F_2,29 = 5.319, p = 0.011) while population of origin and the interaction between stress treatment and population of origin did not (population of origin: F_1,29 = 0.056, p = 0.815; interaction: F_2,29 = 0.126, p = 0.882). Post-hoc comparisons indicated that the control group had significantly larger medial cortices than either the CORT (p = 0.020) or FA treatments (p = 0.030); the CORT and FA treatments did not differ (p = 0.969) (figures 2a and 3a).

Figure 2.

The relationship between remainder of the telencephalon volume (mm³) and (a) medial and (b) dorsal cortex volume (mm³) in eastern fence lizards (Sceloporus undulatus). Individuals originated from two different populations (fire ant invaded versus uninvaded) and were subjected to three early-life stress treatments (control, individuals treated with exogenous corticosterone (CORT), and individuals exposed to sublethal fire ant attack).

Figure 3.

(a) Medial and (b) dorsal cortex volume (mm³) ± SE across three early-life stress treatment groups (control, individuals treated with exogenous corticosterone (CORT), and individuals exposed to sublethal fire ant attack), originating from two different populations (fire ant invaded versus uninvaded). When controlling for the remainder of the telencephalon, stress treatment had a significant effect on both cortices (medial: p = 0.011, dorsal: p = 0.024); population of origin and the interaction between stress treatment and population of origin did not (all p > 0.169). Individuals in the control group had larger medial cortices than those in the CORT (p = 0.002) or fire ants (p = 0.03) treatments. The control group had larger dorsal cortices than those in the CORT treatment (p = 0.06) and trended in that direction in the fire ants treatment (p = 0.087). There were no differences between CORT and fire ants treatment groups for either cortical region (medial: p = 0.969, dorsal: p = 0.402).

When controlling for the remainder of the telencephalon (telencephalon covariate: F_1,29 = 22.033, p < 0.001), we found that stress treatment also had a significant effect on the dorsal cortex volume (F_2,29 = 4.264, p = 0.024) while population of origin and the interaction between stress treatment and population of origin did not (population of origin: F_1,29 = 1.995, p = 0.169; interaction: F_2,29 = 0.100, p = 0.905). Post-hoc comparisons indicated that the control group had significantly larger dorsal cortices than the CORT treatment (p = 0.006) and trended in that direction when compared to the FA treatment (p = 0.087); the CORT and FA treatments did not differ (p = 0.402) (figures 2b and 3b).

When controlling for medial cortex volume (medial cortex covariate: F_1,29 = 0.209, p = 0.651), we found that stress treatment, population of origin, and the interaction between stress treatment and population of origin did not significantly affect medial cortex neuron soma size (stress treatment: F_2,29 = 0.196, p = 0.823; population of origin: F_1,29 = 2.328, p = 0.138; interaction: F_2,29 = 0.514, p = 0.604).

Elimination of the covariates from all the above analyses yielded similar statistical results and therefore did not alter the interpretation of the results.

4. Discussion

We found that weekly early-life stressors in the form of CORT and sub-lethal attack by fire ants decreased relative medial and dorsal cortical volumes in adult eastern fence lizards. Because we did not find differences in remainder of the telencephalon volume, these results were not attributable to global changes in the brain and thus were specific to the cortices. There were no statistically significant differences between CORT and FA exposure on the cortical phenotype. While early-life stressors affected overall medial cortex volume, this was not due to variation in neuron soma size in the medial cortex. Although early-life stressors did decrease medial and dorsal cortical volumes, intergenerational effects (population of origin, invaded versus uninvaded) did not affect cortical or neuron soma volumes, nor did they differentially interact with the early-life stressors to modulate the resultant cortical phenotypes.

(a) . The effects of early-life stress on the neural phenotype

The postnatal brain undergoes significant developmental changes from birth until adulthood. During ontogeny, an individual is susceptible to environmental experiences that can have major organizational effects on the brain [67,68]. Further, early-life stress and GC exposure have long-term effects on cognition and the brain, particularly in the hippocampus, a plastic region of the mammalian brain involved in spatial memory, cognition, and the stress response [69]. In mammals, when early-life stress or exposure to GCs is mild and brief, hippocampal attributes and spatial cognition tend to be enhanced, and therefore short-term exposure to stressors can be beneficial. Conversely, long-term or severe early-life stress, or long-term exposure to GCs, push an individual into allostatic overload, resulting in cognitive deficits and decreased hippocampal attributes, which can persist into adulthood [70–72]. Taken together, there is ample evidence that early-life stress modulates the resultant mammalian hippocampal phenotype, although the direction of the effect is determined by the intensity and longevity of the stressor.

The relationship between stress and the brain is thought to be mediated through the production and regulation of GCs. In vertebrates, GCs cross the blood–brain barrier and bind to GC receptors that are heavily concentrated in hippocampal neurons [26,27]. GC binding initiates a multitude of changes within the mammalian hippocampus including alteration of long-term potentiation, neuron firing rate, dendritic branching, neurogenesis, and volume [69,73–76], which are then mirrored by changes in cognitive functioning [30,69,71]. These studies suggest that GCs are one of the mechanistic intermediaries between stress, cognition and the neural substrate supporting cognitive processes in mammals [6,29,71,77,78].

In line with previous studies on early-life stress and the brain, this study found that long-term (42 weeks), weekly exposure to a potentially lethal predator or an ecologically relevant dose of CORT during early life is sufficient to decrease medial and dorsal cortical volumes. Further, there were no differences between cortical attributes in the CORT and FA treatments, and GCs were similarly elevated in both treatments [49,50]. This supports previous research in rodents indicating that GCs are the likely mechanism that mediates stress and the neural phenotype. However, it is unclear if and how cortical-based cognitive processes or the stress response may be affected by decreased cortical volume in this species. This is important, in that differential stress responses, neural compensation and brain function may exist in populations that experience different environmental stressors, especially in populations that encounter chronic, predictable stressors in the wild.

(b) . The effects of intergenerational stress on the neural phenotype

If an offspring's environment is likely to match that of the parents, parents may adaptively prepare offspring for a consistently stressful environment [18]. While studies on the effects of intergenerational stress on morphology and behaviour are becoming more commonplace, a clear understanding of how intergenerational stress can affect the neural architecture and cognition of offspring is unclear. Several studies in rats and mice have examined the effects of parental pre-conception stress on the brain and cognition of offspring, but the directionality of the effects remains equivocal. For example, male rodents who experienced traumatic stress produced offspring with decreased hippocampal synaptic attributes, decreased dendritic spine density and long-term memory impairments [12,19]. However, one study found that rat mothers stressed before conception produced male offspring with increased hippocampal neuron dendritic length, spines and dendritic complexity [20]. Finally, two studies found that rat parents that were stressed before conception produced offspring with increased anxiety, an altered CORT response, and changes in hippocampal DNA methylation [21,22]. Although there is a relative paucity of studies examining intergenerational effects of stress on the brain, all have shown some demonstrable effect of intergenerational stress on the rodent hippocampus. At present, it is unclear how different stress contexts modulate neural variation, and alter the directionality of those effects, across generations, and if intergenerational stress similarly affects the brain in other vertebrates, especially in the field. Interestingly, in our study species, elevated glucocorticoids during gestation have sex-specific effects on offspring telomere lengths [79] and DNA methylation [80], suggesting potential mechanisms [81].

While previous studies in rodents demonstrated intergenerational effects of stress on the brain, we did not find any effect on the cortices. Within each treatment group, offspring from invaded and uninvaded populations had similar medial and dorsal cortical volumes and medial neuron soma volumes. Further, population of origin did not interact with early-life stress treatments to differentially modulate the cortical phenotype. One possibility is that stress of capture and rehousing gravid females induced abnormally high levels of GCs that transferred to the egg yolks. High maternal GC levels can pass into egg yolks in lizards [52,82–85], and this may have affected the cortical phenotype of offspring, suppressing any intergenerational differences. However, previous studies using the exact same subjects found effects of intergenerational stress, but not early-life stress, on other aspects of the offspring phenotype. Offspring from invaded populations had increased CORT reactivity, increased back limb length and adaptive hemagglutination ability, grew more quickly, and exhibited decreased survival in the lab [51,53,86], suggesting that CORT deposition in eggs due to capture was not sufficient to suppress intergenerational effects in those measured variables. Alternatively, we speculate the lack of intergenerational cortical differences may be due to selection against fixed cortices. The mammalian hippocampus is one of the most plastic regions of the brain, exhibiting the capacity for a broad range of structural remodelling including the production and incorporation of new neurons into the existing neural architecture, synaptic modifications, and dendritic branching and retraction. Much of this plasticity is due to within lifetime experiences such as stress, motor activity, learning and memory use, and environmental enrichment [87]. Because of the varied functions of the mammalian hippocampus, there may be other, more dominating selective forces that select against a fixed phenotype, to cope with changes in the environment. Further, compared to mammals, squamate reptiles have higher levels of neurogenesis and a much greater ability to significantly remodel the brain after injury than do mammals [88,89]. The unique neural plasticity of this taxonomic group may further counter selection for a fixed cortical phenotype, as lack of plasticity may be maladaptive over the lifetime of the individual.

A non-mutually exclusive explanation concerns the environment in which offspring were raised. Decreased environmental complexity within the laboratory creates lower demands on learning and memory, which is correlated with decreased brain attributes, particularly in the mammalian hippocampus and hippocampal homologues in other vertebrates [90]. The effects of environmental complexity are even more apparent when comparing laboratory-reared and field-caught conspecifics; previous studies in mammals, birds, reptiles and fish found that individuals either hatched/born or housed long-term in the laboratory have smaller hippocampi/cortices and decreased neurogenesis than their field-caught counterparts [91–99]. Thus, while CORT and FA treatments were sufficient to decrease cortical volume in the lab, the ability to detect the effects of intergenerational stress on the brain, if they exist, may have been diminished due to cortical development in a simplistic laboratory environment.

(c) . The effects of CORT versus an ecologically relevant stressor

Stress and GCs exhibit an inverted-U relationship with mammalian hippocampal attributes and cognition [8,27], where predictable, moderate levels of stress can up-regulate mammalian hippocampal neurogenesis and enhance hippocampal-based cognition whereas unpredictable and intense chronic stress down-regulates neurogenesis and impairs cognition [100]. However, stress is also a contextual experience, and this relationship may not be static over the lifetime [29]. Varying the stress context can change the salience and emotional valence of the stressor and the subsequent physiological response [101,102]. For example, an individual may habituate to a predictable, controllable or continual stressor, reflected by a dampened HPA response to that stressor [103–106]. Interestingly, the contextual importance of stress has been observed in previous studies on eastern fence lizards. In one study, lizards were either intentionally exposed to fire ant attacks or housed with fire ants in semi-natural enclosures. The lizards that were intentionally exposed to fire ant attacks had elevated CORT levels compared to lizards housed with fire ants, suggesting that a more naturalistic encounter may be perceived as less stressful or may eventually dampen the stress response [50]. Another study found that lizards from invaded populations in the FA treatment had high CORT reactivity during restraint but not when acutely exposed to fire ants [51]. Here, it appears that the novelty or lack of controllability of the restraint-induced stressor better predicts an elevated stress response. Finally, adult male fence lizards exposed to two different yet ecologically relevant stressors (high temperature and fire ants) differed in CORT reactivity, again suggesting that stress context influences stress responses [107]. Taken together, the stress context can modulate stressor saliency and the subsequent HPA response. Dampening the stress response to a consistent, mild stressor could be advantageous, as less energy would be expended during future encounters with relatively benign stressors.

Although our previous study found variation in CORT reactivity between invaded and uninvaded individuals, CORT reactivity did not differ when exposed to an ecologically relevant stressor [51]. This could possibly explain why cortical attributes did not differ between CORT and FA treatments. Once a week exposure may not be enough to create different stress contexts, reflected by comparable CORT reactivity and neural attributes between the two stress treatments. Similarly, over time stressors can become predictable and individuals can adjust their physiological stress response to stressors [2]. For example, rats with repeated exposure to an intruder have lower c-fos expression in the brain compared with males who experience only one intrusion [104]. In our study, the stress contexts were different, but the predictability and non-lethality of the stressors were identical. This may have created a stress response but one that did not vary between the CORT and FA contexts. Examining individuals at regular time points across the duration of the study would elucidate if any variation that may have existed between the CORT and FA treatments converged.

(d) . Finer-scale cortical changes

We found that early-life stress treatments can decrease cortical volume, yet the functional relevance of brain region volume is unclear. Probably, finer-scale anatomical changes such as neuron number, neuron soma size and dendritic branching underlie volumetric changes, and assessing these parameters may be more informative for understanding the cellular basis and functional significance of cortical volume changes [108]. Previous studies found variation in avian hippocampal volume with concurrent changes in hippocampal neuron soma sizes [109,110]. Similarly, studies have found variation in brain region volume and neuron soma size within the region of interest in lizards [65,111], suggesting that neuron size may partially contribute to volumetric changes. However, we also measured variation in neuron soma size and found that there is not a concomitant change in neuron soma volume in the medial cortex. Therefore, there are probably other finer-scale neural attributes, possibly dendritic branching or neuron number, that exhibit plasticity and contribute to cortical volume variation based on early-life stressors.

(e) . Adaptive stress response and the brain

Clinical research has deepened our understanding of the relationship among stress, GCs, cognitive functioning, and the mammalian hippocampus [112]. However, a laboratory setting cannot capture the myriad of stressors an individual experiences in the field, nor delineate how those stressors may differentially affect the HPA axis. Stress comes from many facets in the natural environment including predation, abiotic factors, and interactions with conspecifics, and these stressors can be novel, familiar, consistent, sporadic, controllable, uncontrollable or some combination thereof. Many vertebrates also exhibit substantial variation in the stress response based on individual, species, sex, age, life history stage and season [36,113,114]. Because of the inherent dynamic nature of the HPA axis and the myriad of intrinsic and environmental factors that modulate the stress response, it may not be surprising that baseline and stress-induced GC levels in wild vertebrates exhibit low repeatability, and the effects of GCs on fitness have been equivocal [114]. Therefore, a unified explanation of the adaptive and evolutionary significance of the vertebrate stress response has yet to be described.

What is apparent, however, is that the stress response can differentially adapt to a stressor based on the context [115]. Changes in the stress response may include a longer latency to initiate the stress response, a blunted stress response, or a stress response that recovers more quickly [116,117]. Some stressful stimuli are predictable over time (e.g. interaction with territorial neighbours) and individuals will have repeated exposures to these non-lethal stressors. Eventually, this may lead to an adjustment in, or dampening of, the stress response [2,116–119]. For example, laboratory mice subjected to repeated, sublethal stress developed a dampened stress response [120], suggesting an adaptation of the HPA axis to that stressor. Similarly, in brown anoles (Anolis sagrei), lizards from an urban population had a blunted stress response and a lower concentration of cortical CORT after an environmental stressor compared to a forest population [121]. In eastern fence lizards, individuals from populations that have evolved and adapted to the stress of fire ant attacks may exhibit physiological adaptation to a repeated ecologically relevant stressor, reflected by a blunted stress response. If this response is naturally selected, it may be plausible that intergenerational effects would then also be diminished. Alternatively, stress from fire ant attack in invaded populations may not be consistent or strong enough to warrant a fixed response in the brain or, assuming stressors are strong enough, there may not have been enough evolutionary time for the cortices to become fixed.

Currently, it remains unclear if environmental stressors in the field select for differential physiological stress responses or adaptive phenotypic plasticity [29,122,123]. Similarly, because we have comparatively little research of the importance of the hippocampus/hippocampal homologues in modulating the stress response in wild animals, it is unclear if the neural mechanisms underlying the stress response parallel that found in laboratory rodents. Taken together, we are just beginning to understand the evolutionary trajectory of the stress response in natural systems, with even more to discover about the co-evolution of the stress response and the brain. Leveraging natural populations that are evolving or have evolved in differing stress contexts can provide insight on the evolution and plasticity of the stress response and the hippocampus/hippocampal homologues, as well as aid in predicting how a population may respond to novel environmental stressors over time.

Ethics

All animals were collected under appropriate state permits and all procedures were approved by the Institutional Animal Care and Use Committee at The Pennsylvania State University (no. 35780).

Data accessibility

Data files can be accessed on ScholarSphere: https://doi.org/10.26207/72qc-af46 [124].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

L.D.L.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, supervision, visualization, writing—original draft, writing—review and editing; G.L.M.: conceptualization, data curation, funding acquisition, investigation, methodology, project administration, writing—review and editing; T.R.R.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, writing—review and editing; A.S.L.: data curation, formal analysis, investigation, methodology, writing—review and editing; T.L.: conceptualization, funding acquisition, investigation, project administration, resources, supervision, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was supported in part by the National Science Foundation (grant no. DGE1255832 to G.L.M.; IOS1051367 to T.L. and S.A.C.).