Interactive influence of sex, stressor timing, and the BclI glucocorticoid receptor polymorphism on stress-induced alterations of long-term memory

Abstract

Certain susceptibility factors, such as genetic variants or specific physiological responses to stress, can dictate the effects of stress on learning and memory. Here, we examined the influence of the BclI polymorphism of the glucocorticoid receptor gene on the time-dependent effects of pre-learning stress on long-term memory. Healthy individuals were exposed to the socially evaluated cold pressor test or a control condition immediately or 30 min before word list learning. Participants’ memory for the words was tested immediately and 24 h after learning, and saliva samples were collected to genotype participants for the BclI polymorphism and to assess cortisol responses to the stressor. Results revealed that stress immediately before learning enhanced memory, while stress 30 min before learning impaired memory; these effects were largely selective to males and non-arousing words. Additionally, stress, independent of when it was administered, enhanced memory in non-carriers of the BclI polymorphism, while impairing memory in carriers; these effects were largely selective to males and participants exhibiting a robust cortisol response to stress. These results provide further evidence for time-dependent effects of stress on long-term memory and suggest that carriers of the BclI polymorphism might be more sensitive to the negative effects of corticosteroids on learning.

1. Introduction

The effects of stress on cognition are adaptive (Diamond & Zoladz, 2016; Zoladz, Park, & Diamond, 2011); they direct our attention to stress-related stimuli, helping us remember the stressful event and what occurs around it. Such effects can aid in survival, especially if a similar episode is experienced later in life. However, the adaptive function of stress-induced alterations of learning and memory can also have inadvertent consequences. It can result in powerful and intrusive traumatic memories that underlie psychological disorders like post-traumatic stress disorder (PTSD). Moreover, devoting a great deal of cognitive resources to the stressful episode can lead to impaired executive functions (e.g., working memory, planning, decision-making), as well as impaired learning and memory for other, unrelated information. Because stress-related alterations of cognition are characteristic of several psychological disorders, developing a better understanding of stress-memory interactions has great scientific and clinical value.

Stress can enhance, impair, or have no effect on learning and memory depending on several factors (e.g., sex, emotional nature of the learned information, stressor timing, etc.) (Diamond, Campbell, Park, Halonen, & Zoladz, 2007; Joels, Fernandez, & Roozendaal, 2011; Schwabe, Joels, Roozendaal, Wolf, & Oitzl, 2012). It is well-documented that post-learning stress enhances memory consolidation and pre-retrieval stress impairs memory, both effects being attributable to an interaction between corticosteroid and noradrenergic mechanisms in the amygdala and hippocampus (Roozendaal, McEwen, & Chattarji, 2009). Pre-learning stress effects on long-term memory are more inconsistent in the literature, yet perhaps most translatable to understanding the mechanisms of traumatic memory formation. One factor that has emerged in the past couple decades as a major determinant of pre-learning stress effects on long-term memory is the timing of stress relative to learning (Diamond et al., 2007; Joels et al., 2011; Schwabe et al., 2012). When a brief stressor is administered immediately before learning, long-term memory is generally enhanced (e.g., Diamond et al., 2007; Quaedflieg, Schwabe, Meyer, & Smeets, 2013; Vogel and Schwabe, 2016; Zoladz, Clark, et al., 2011; Zoladz et al., 2017c, 2014b). On the other hand, when the same stressor is temporally separated from learning (e.g., 30 min before learning), long-term memory is generally impaired (e.g., Quaedflieg et al., 2013; Zoladz, Clark, et al., 2011; Zoladz et al., 2013). Interestingly, in some of our work (Zoladz et al., 2017b, 2013), we have observed that males are more susceptible to these time-dependent effects than females. Such findings are consistent with a wealth of literature revealing significant effects of stress on learning and memory in males and opposite or no effects in females (Andreano and Cahill, 2006; Jackson, Payne, Nadel, & Jacobs, 2006; Payne et al., 2006; Wolf, Schommer, Hellhammer, McEwen, & Kirschbaum, 2001; Zorawski, Blanding, Kuhn, & LaBar, 2006). Moreover, although investigators have sometimes observed selective effects of stress on emotionally-arousing information (e.g., Kuhlmann, Piel, & Wolf, 2005; Payne et al., 2007; Payne et al., 2006), studies from our laboratory (Zoladz et al., 2017b, 2017c, 2014b, 2013), as well as from the laboratory of others (Quaedflieg et al., 2013; Schwabe, Bohringer, Chatterjee, & Schachinger, 2008), examining time-dependent effects of pre-learning stress on long-term memory have not revealed valence- or arousal-dependent effects of stress on learned material or have observed effects that are selective to neutral information.

Investigators have contended that the time-dependent effects of prelearning stress are due to a biphasic influence of stress-induced amygdala activation on hippocampal synaptic plasticity, as well as the temporal profiles of stress-induced noradrenergic and corticosteroid activity (Akirav and Richter-Levin, 2002; Diamond et al., 2007; Joels et al., 2011; Schwabe et al., 2012). Specifically, brief stress experienced immediately before learning enhances long-term memory via the rapid increase in norepinephrine and non-genomic effects of slowly rising corticosteroids exerting excitatory influences on hippocampal synaptic plasticity. In contrast, stress that is temporally separated from learning results in long-term memory impairment due to, at least in part, rising corticosteroid levels exerting delayed, inhibitory influences on hippocampal function.

Our laboratory has been using the pre-learning stress model as a method for better understanding individual differences in stress-induced alterations of learning and memory (Zoladz, Clark, et al., 2011; Zoladz et al., 2014a, 2014b, 2013), which may lend insight into susceptibility factors for traumatic memory formation and psychological illness. One factor that has been associated with the stress response, emotional memory formation, and risk for psychological disorders is the BclI polymorphism of the glucocorticoid receptor (GR) gene (NR3C1), which is located on chromosome 5 in humans (Hauer et al., 2011). The BclI polymorphism is a common polymorphism located in intron B of the NR3C1 gene and consists of a C-to-G nucleotide change. Because the NR3C1 gene results in the production of the glucocorticoid receptor, which responds to cortisol, research has focused on the polymorphism’s influence on cortisol- and stress-related physiological and behavioral processes. Such work has demonstrated the greatest influence of this polymorphism on homozygous carriers of the G allele.

The BclI polymorphism has been associated with lower basal cortisol concentrations (Hauer et al., 2011) and greater GR sensitivity to corticosteroids in healthy (Huizenga et al., 1998; van Rossum et al., 2003) and clinical (Bachmann et al., 2005) samples. Researchers have also shown that homozygous carriers of the G allele exhibit increased (Kumsta et al., 2007; Li-Tempel et al., 2016) or decreased (Hauer et al., 2011; Wust et al., 2004) responses of the hypothalamus-pituitary-adrenal (HPA) axis to stress and greater baseline levels of (Li-Tempel et al., 2016) and stress-induced increases in (Taylor, Larson, & Lauby, 2014) heart rate (HR) and blood pressure (BP). In addition to its association with the stress response, the BclI polymorphism has been implicated in several somatic disorders, such as hypertension (Moreira, Gomes, Mendonca, & Bachega, 2012), insulin resistance (Weaver, Hitman, & Kopelman, 1992), bone resorption (Koetz, van Rossum, Ventz, Diederich, & Quinkler, 2013), and obesity (Buemann et al., 1997; van Rossum and Lamberts, 2004), to name a few.

Homozygous carriers of the BclI polymorphism are at greater risk for major depression (van Rossum et al., 2006) and PTSD (Hauer et al., 2011) and exhibit increased rates of suicide (Park et al., 2016), relative to non-carriers. The elevated risk of psychopathology in BclI carriers might relate to the polymorphism’s aforementioned influence on GR sensitivity, which consequently affects emotional memory formation. In a study examining the interactive influence of PTSD and GR-related polymorphisms on HPA axis activity, Bachmann and colleagues (Bachmann et al., 2005) showed that only PTSD patients who were homozygous carriers of the BclI polymorphism displayed elevated GR sensitivity. Other studies have linked such changes in GR sensitivity with greater emotional memory. For instance, Ackermann and colleagues (Ackermann, Heck, Rasch, Papassotiropoulos, & de Quervain, 2013) reported that healthy homozygous carriers of the BclI polymorphism displayed superior memory for emotional pictures, relative to heterozygous carriers and non-carriers of the polymorphism, and Hauer et al. (2011) showed that homozygous carriers of the polymorphism exhibited lower baseline cortisol levels and more long-term traumatic memories following cardiac surgery and intensive care unit therapy than heterozygous carriers and non-carriers of the polymorphism.

Because of the well-established involvement of corticosteroids in the stress-induced alteration of learning and memory and the emerging association between the BclI polymorphism, GR sensitivity, and emotional memory formation, we examined the influence the BclI polymorphism on the time-dependent effects of pre-learning stress on long-term memory. Participants were exposed to a brief (3-min) stressor immediately or 30 min before learning a list of words and then tested for their memory 24 h later. Based on previous work from our laboratory, we hypothesized that stress immediately before learning would enhance long-term memory, while stress 30 min before learning would impair long-term memory. Such effects were expected to be greater in males and not dependent on the valence or arousal ratings of the words. Based on work linking the BclI polymorphism with increased GR sensitivity, we predicted that stress would exert more pronounced effects on learning and memory in polymorphism carriers.

2. Material and methods

The present manuscript is based on data that overlap with data sets from two previous publications (Zoladz et al., 2017b, 2017c). The unique focus of this paper is a different genetic polymorphism (BclI polymorphism in the NR3C1 gene) and its association with stress-induced changes in learning and memory.

2.1. Participants

Two hundred and thirty-five healthy undergraduate students (97 males, 138 females; age: M = 19.65, SD = 1.59) from Ohio Northern University volunteered to participate in the experiment. The overall sample size for the experiment was based on an a priori power analysis (G*Power 3.1.9.2; University of Kiel, Germany) indicating that in order to attain adequate power (i.e., 1 – β = 0.8) to detect small-to-moderate effect sizes (i.e., partial eta squared = 0.09) for the Stress×Stress Time Point and Stress×Genotype interactions, we would need a total sample of approximately 260 participants. Individuals were excluded from participating if they met any of the following conditions: diagnosis of Raynaud’s or peripheral vascular disease; presence of skin diseases, such as psoriasis, eczema, or scleroderma; history of syncope or vasovagal response to stress; history of any heart condition or cardiovascular issues (e.g., high blood pressure); history of severe head injury; current treatment with psychotropic medications, narcotics, beta-blockers, steroids, or any other medication that was deemed to significantly affect central nervous or endocrine system function; mental or substance use disorder; regular tobacco use; regular use of recreational drugs; regular nightshift work. Participants were asked to refrain from drinking alcohol or exercising extensively for 24 h prior to the experimental sessions; and, to refrain from eating or drinking anything but water for 2 h prior to the experimental sessions. All experimental procedures were approved by the Institutional Review Board at Ohio Northern University, carried out in accordance with the Declaration of Helsinki, and undertaken with the understanding and written consent of each participant. Participants were awarded class credit and $20 cash upon completion of the study.

2.2. Experimental procedures

All experimental procedures took place between 1000 and 1700 in an attempt to control for diurnal fluctuations in cortisol. A timeline of all procedures can be found in Fig. 1.

Fig. 1.

Timeline of the experimental procedures. On Day 1, participants placed their non-dominant hand in ice cold (stress) or warm (no stress) water for 3 min. Stressed participants were also led to believe that they were being videotaped throughout the water bath. Heart rate was continuously recorded throughout the water bath and commenced approximately 1 min prior to participants placing their hand in the water. Subjective pain and stress ratings of the water bath were collected at 1-min intervals. Immediately (“Immed.”) or 30 min (“Delayed”) following the water bath, participants were given a word list to learn, followed immediately by a free recall assessment. Measures of affect (PANAS) and anxiety (SAI) levels were administered before and after the water bath. Saliva samples were also collected before and after the water bath to assess changes in salivary cortisol levels. The next day, participants returned to the laboratory to complete free recall and recognition assessments, which were separated by 15 min. Saliva samples were collected before and after these assessments to assess changes in salivary cortisol levels. Between the two assessments, we collected a saliva sample via the Oragene kit in order to genotype participants for the BclI polymorphism in the glucocorticoid receptor gene (NR3C1).

2.2.1. Socially evaluated cold pressor test (SECPT)

Following completion of a short demographics survey and the collection of baseline physiological and self-report measures (see below), participants were asked to submerge their non-dominant hand in a bath of water for 3 min. Participants who had been randomly assigned to the stress condition (N = 120; 47 males, 73 females) placed their hand in a bath of ice cold (0–2 °C) water, while participants who had been randomly assigned to the control condition (N = 115; 50 males, 65 females) placed their hand in a bath of warm (35–37 °C) water. The water was maintained at the appropriate temperature by a circulating water bath (Cole-Parmer; Vernon Hills, IL). If a participant found the water bath too painful, he or she was allowed to remove his or her hand from the water and continue with the experiment. Based on previous work (Schwabe, Haddad, & Schachinger, 2008), a social evaluative component was added to the cold pressor manipulation. Participants in the stress condition were misleadingly informed that they were being videotaped during the procedure for subsequent evaluation of their facial expressions, and throughout the water bath manipulation, they were asked to keep their eyes on a camera that was located on the wall of the laboratory.

2.2.2. Subjective and objective stress response measures

2.2.2.1. The positive and negative affect schedule (PANAS) and state anxiety Inventory (SAI).

Immediately before and approximately 10 min after the water bath manipulation, participants completed the PANAS (Watson, Clark, & Tellegen, 1988) and the SAI (state portion of the State-Trait Anxiety Inventory) (Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983). This allowed for an analysis of stress-induced changes in affect and anxiety, respectively. We administered the poststress PANAS and SAI 10 min following the water bath manipulation because we wanted to ensure that word list learning and immediate free recall immediately followed the SECPT in the immediate learning conditions (see Section 2.2.3).

2.2.2.2. Subjective pain and stress ratings.

Participants rated the painfulness and stressfulness of the water bath at 1-min intervals on 11-point scales ranging from 0 to 10, with 0 indicating a complete lack of pain or stress and 10 indicating unbearable pain or stress.

2.2.2.3. Cardiovascular analysis.

Heart rate (HR) was measured continuously for approximately 1 min before the water bath until its completion via a BioNomadix pulse transducer (Biopac Systems, Inc.; Goleta, CA) placed on the ring finger of participants’ dominant hand. The pulse transducer was connected to the PPG module of the MP150 Biopac hardware. Average baseline HR (average of 1 min before water bath) and water bath HR (average of water bath) were calculated for statistical analysis.

2.2.2.4. Cortisol analysis.

On Day 1, saliva samples were collected from participants immediately before and 25 min after the water bath to analyze salivary cortisol concentrations. This enabled us to detect stress-induced changes in salivary cortisol levels in participants from both immediate and delayed learning conditions (see Section 2.2.3). It also enabled us to assess salivary cortisol levels for participants in the delayed learning conditions at the time of word list learning. On Day 2, saliva samples were collected from participants immediately before and 25 min after the free recall assessment to analyze salivary cortisol levels. Saliva samples were collected in a Salivette saliva collection device (Sarstedt, Inc., Newton, NC). The samples were stored at −20 °C until being thawed and extracted by low-speed centrifugation. Salivary cortisol levels were then determined by enzyme immunoassay (EIA; Cayman Chemical Co., Ann Arbor, MI) according to the manufacturer’s protocol.

2.2.3. Learning and memory task

Immediately [N = 116; Stress: N = 59 (22 males, 37 females); No Stress: N = 57 (24 males, 33 females)] or 30 min [N = 119; Stress: N = 61 (25 males, 36 females); No Stress: N = 58 (26 males, 32 females)] following exposure to the water bath, participants were presented with a list of 42 words, which were selected from the Affective Norms for English Words (Bradley and Lang, 1999). Based on standardized valence and arousal ratings, we chose 14 neutral, 14 positive, and 14 negative words (7 arousing and 7 non-arousing per category), which, across emotional valence and arousal categories, were balanced for word length and word frequency. As per previous methodology (Schwabe, Bohringer, et al., 2008; Zoladz, Clark, et al., 2011; Zoladz et al., 2014a, 2014b, 2013) and to promote encoding of the words, participants were instructed to read each word aloud and rate its emotional valence on a scale from −4 (very negative) to +4 (very positive) and its arousal level on a scale of 0 (not arousing) to 8 (very highly arousing), with the aid of self-assessment manikins, on a sheet of paper containing the list of words.

Immediately following word list encoding, participants were given 5 min to write down as many words as they could remember from the list of words they just studied (immediate recall). The next day, participants returned to the laboratory to have their memory for the list of words assessed. Participants were again given 5 min to write down as many words as they could remember from the list of words that they studied on the previous day (delayed recall). Fifteen minutes later, participants were given a recognition test. They were presented with a list of words containing 42 “old” words (i.e., words presented on the previous day) and 42 “new” words (i.e., words not presented on the previous day) and were instructed to label each word as “old” or “new.” The “new” words were matched to the “old” words on emotional valence, arousal level, word length, and word frequency. To assess participants’ ability to discriminate between “old” and “new” words, we calculated a sensitivity index (d’ = z[p(hit) – p(false alarm)]) for each category of word to be used for statistical analysis (Wickens, 2002). We also analyzed hits and false alarms separately for completeness.

2.2.4. Genotyping

On Day 2, during the 15-min delay between free recall and recognition testing, a saliva sample was collected from participants via the OGR-500 Oragene (DNA Genotek, Inc.; Ottawa, ON, Canada). The sample was stored at room temperature, until shipped to DNA Genotek, Inc. for genotyping of the BclI polymorphism in the NR3C1 gene for the glucocorticoid receptor. Genotyping was performed by single tube Taqman^® chemistry. The Taqman^® assay is an allele discrimination assay using PCR amplification and a pair of fluorescent dye detectors that target the polymorphism. One fluorescent dye is attached to the detector that is a perfect match to the first allele (e.g., an “G” nucleotide) and a different fluorescent dye is attached to the detector that is a perfect match to the second allele (e.g., a “C” nucleotide). During PCR, the polymerase will release the fluorescent probe into the solution where it is detected using endpoint analysis in a Life Technologies, Inc. (Foster City, CA) 7900HT Real-Time instrument. Primes and probes were obtained through Life Technologies design and manufacturing. Life Tehcnologies Taqman Genotyper v1.0.1 software - Taqman^® single tube assay was used for analysis. The call rate for the polymorphism was 97.6%, and the cluster plot image for the genotyping can be found in Fig. 2.

Fig. 2.

Cluster plot image depicting the genotyping results for rs41423247. Red dots indicate participants who were homozygous for the polymorphism (i.e., GG). Green dots indicate participants who were heterozygous for the polymorphism (i.e., CG). Blue dots indicate participants who were wild type (i.e., CC). Teal dots indicate no template controls. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

2.3. Statistical analyses

Based on previous work establishing an association between the G allele of the NR3C1 gene and HPA axis reactivity to stress, greater GR sensitivity, increased susceptibility to psychological disorders and related phenotypes, and enhanced emotional memory, we divided participants into G allele carriers [heterozygous (N = 95), BclI homozygous (N = 31)] and non-carriers [wild type (N = 108)] for the purpose of data analysis. Although much of the work examining the BclI polymorphism has observed the greatest effects in homozygous carriers of the G allele, we combined heterozygous and homozygous G allele carriers into one group to increase statistical power (i.e., statistical power to detect an effect in homozygous carriers would have been considerably low otherwise). Based on previous work (e.g., Ackermann et al., 2013; Buemann et al., 1997; Koetz et al., 2013; Moreira et al., 2012), we anticipated that approximately half of the sample would carry the G allele. There were 126 carriers [Immediate: N = 67; Stress: N = 31 (14 males, 17 females); No Stress: N = 36 (17 males, 19 females); Delayed: N = 59; Stress: N = 30 (14 males, 16 females); No Stress: N = 29 (12 males, 17 females)] and 108 non-carriers [Immediate: N = 49; Stress: N = 28 (8 males, 20 females); No Stress: N = 21 (7 males, 14 females); Delayed: N = 59; Stress: N = 31 (11 males, 20 females); No Stress: N = 28 (13 males, 15 females)] of the G allele in the experiment. One participant was unable to be genotyped for the BclI polymorphism and was therefore removed from analyses. The data were analyzed with mixed-model ANCOVAs. The between-subjects factors utilized in these analyses were genotype (G carrier, G non-carrier), stress (stress, no stress), stress time point (immediate, delayed), sex (male, female), and the within-subjects factors were word valence (positive, negative, neutral) and arousal (arousing, non-arousing) (for recall and recognition analyses) or measurement time point (for HR, cortisol, and self-report analyses). Because endogenous cortisol levels follow a circadian rhythm and participants were stressed at different times of the day, we included time of day of the experimental sessions as a covariate in all analyses. Alpha was set at .05 for all analyzed, and Fisher’s LSD post hoc tests were employed when the omnibus F indicated the presence of a significant effect.

3. Results

3.1. Genotype characteristics

Chi-square goodness-of-fit analyses revealed that there was no significant deviation from the Hardy-Weinberg equilibrium for the BclI genotype (χ²(1, N = 234) = 1.87, p = 0.17). This suggests that the genotype distribution in our sample did not significantly deviate from the expected genotype distribution in the population.

3.2. Subjective and objective stress response measures

3.2.1. Affect and anxiety levels

3.2.1.1. Positive affect.

Carriers of the BclI polymorphism exhibited lower positive affect than non-carriers at baseline (Genotype×Measurement Time Point interaction: F(1,217) = 4.31, p < 0.05, η² = 0.02). Males exhibited greater positive affect, overall, than females (effect of sex: F(1,217) = 12.60, p < 0.001, η² = 0.06).

3.2.1.2. Negative affect.

Stress led to an increase in negative affect, but only for participants in the delayed learning condition (Stress×Stress Time Point×Measurement Time Point interaction: F(1,217) = 37.57, p < 0.001, η² = 0.15) (see Table 1).

Table 1.

Pre-post changes ( ± SEM) in Day 1 heart rate, affect, and anxiety and Day 2 cortisol.

		Immediate		Delayed
Measure/condition		Pre	Post/during (HR)	Pre	Pre/during (HR)
Day 1 Heart Rate (bpm)
Stress
	Carriers	73.84 (3.20)	98.52 (3.34)*	72.06 (3.20)	95.70 (3.34)*
	Non-carriers	73.97 (3.65)	92.09 (3.82)*	74.92 (3.28)	97.88 (3.42)*
No stress
	Carriers	78.79 (2.92)	86.38 (3.05)	71.69 (3.29)	77.27 (3.44)
	Non-carriers	77.12 (4.04)	81.52 (4.22)	77.21 (3.31)	86.60 (3.46)
Day 1 Positive Affect (PANAS)
Stress
	Carriers	26.27 (1.15)	24.45 (1.26)	28.52 (1.17)	26.35 (1.28)
	Non-carriers	28.98 (1.34)	25.89 (1.46)	28.18 (1.20)	25.50 (1.31)
No stress
	Carriers	29.38 (1.07)	28.85 (1.17)	26.56 (1.20)	24.12 (1.32)
	Non-carriers	29.82 (1.48)	24.57 (1.62)	31.05 (1.21)	29.15 (1.33)
Day 1 Negative Affect (PANAS)
Stress
	Carriers	13.28 (0.66)	14.03 (0.77)	12.86 (0.67)	16.97 (0.78)*
	Non-carriers	14.50 (0.77)	13.91 (0.89)	14.47 (0.69)	18.33 (0.80)*
No stress
	Carriers	14.18 (0.61)	13.72 (0.71)	14.35 (0.69)	11.98 (0.80)
	Non-carriers	14.18 (0.85)	14.71 (0.99)	13.25 (0.69)	11.09 (0.81)
Day 1 Anxiety (SAI)
Stress
	Carriers	35.64 (1.53)	39.08 (1.72)	33.29 (1.55)	46.19 (1.74)*
	Non-carriers	35.28 (1.78)	39.74 (1.99)	35.93 (1.59)	47.85 (1.79)*
No stress
	Carriers	36.21 (1.42)	36.10 (1.59)	37.08 (1.60)	33.80 (1.80)
	Non-carriers	32.50 (1.97)	38.25 (2.20)	32.41 (1.61)	29.44 (1.80)
Day 2 Salivary Cortisol (nmol/l)
Stress
	Carriers	5.19(0.53)	5.57 (0.47)	5.82 (0.55)	6.21 (0.49)
	Non-carriers	5.93 (0.62)	5.93 (0.55)	4.88 (0.56)	5.07 (0.50)
No stress
	Carriers	4.62 (0.49)	5.10 (0.44)	5.09 (0.56)	5.44 (0.49)
	Non-carriers	6.27 (0.68)	6.88 (0.61)	6.14 (0.57)	5.82 (0.51)

^*p < 0.05 relative to no stress (LSD post hoc test).

3.2.1.3. State anxiety.

Stress led to an increase in state anxiety, but only for participants in the delayed learning condition (Stress×Stress Time Point×Measurement Time Point interaction: F(1,217) = 36.41, p < 0.001, η² = 0.14) (see Table 1).

3.2.2. Subjective pain and stress ratings

Stressed participants, independent of genotype, rated the water bath as more painful (effect of stress: F(1,217) = 853.81, p < 0.001, η² = 0.80) and more stressful (effect of stress: F(1,217) = 585.08, p < 0.001, η² = 0.73) than controls (see Table 2). Stressed females also rated the water bath as more painful (Stress×Sex interaction: F(1,217) = 7.42, p < 0.01, η² = 0.03) and more stressful (Stress×Sex interaction: F(1,217) = 11.44, p < 0.001, η² = 0.05) than stressed males.

Table 2.

Pain and stress ratings ( ± SEM) of the water bath manipulation.

DV/condition		Minute 1	Minute 2	Minute 3
MALES
Painfulness (scale of 0–10)
Stress
	Carriers	6.11 (0.33)*	6.18 (0.32)*	6.14 (0.33)*
	Non-carriers	5.97 (0.41)*	6.06 (0.39)*	6.08 (0.41)*
No stress
	Carriers	0.13 (0.07)	0.29 (0.16)	0.39 (0.15)
	Non-carriers	0.19 (0.12)	0.23 (0.12)	0.19 (0.12)
Stressfulness (scale of 0–10)
Stress
	Carriers	5.25 (0.34)*	5.25 (0.35)*	5.14 (0.37)*
	Non-carriers	5.50 (0.42)*	5.27 (0.43)*	5.38 (0.46)*
No stress
	Carriers	0.31 (0.10)	0.27 (0.10)	0.43 (0.14)
	Non-carriers	0.26 (0.15)	0.41 (0.15)	0.42 (0.24)
FEMALES
Painfulness (scale of 0–10)
Stress
	Carriers	7.01 (0.30)*β	7.60 (0.29)*β	8.23 (0.31)*β
	Non-carriers	6.90 (0.28)*β	7.35 (0.27)*β	7.48 (0.28)*β
No stress
	Carriers	0.17 (0.06)	0.23 (0.07)	0.31 (0.10)
	Non-carriers	0.53 (0.28)	0.49 (0.25)	0.53 (0.26)
Stressfulness (scale of 0–10)
Stress
	Carriers	6.47 (0.32)*β	6.97 (0.32)*β	7.60 (0.34)*β
	Non-carriers	6.83 (0.29)*β	7.05 (0.29)*β	7.33 (0.31)*β
No stress
	Carriers	0.39 (0.13)	0.39 (0.13)	0.45 (0.13)
	Non-carriers	0.60 (0.28)	0.60 (0.25)	0.56 (0.26)

^*p < 0.001 main effect relative to the no stress group,

^β = p < 0.001 relative to stressed males (LSD post hoc test).

3.2.3. Heart rate

Stressed participants, independent of genotype, exhibited significantly greater HR during the water bath, relative to controls (Stress×Measurement Time Point interaction, F(1,216) = 55.91, p < 0.001, η² = 0.21) (see Table 1).

3.2.4. Cortisol

On Day 1, stressed participants exhibited greater salivary cortisol levels following the water bath, relative to controls (Stress×Measurement Time Point interaction: F(1,216) = 82.14, p < 0.001, η² = 0.28; Fig. 3). There was no significant influence of genotype on Day 1 or Day 2 (see Table 1) cortisol levels.

Fig. 3.

Salivary cortisol concentrations before and after stress exposure. Stressed participants, independent of stress time point or BclI carrier status, exhibited significantly greater salivary cortisol levels than controls following the water bath manipulation. Data are expressed as means ± SEM. *p < 0.001 main effect relative to no stress.

3.3. Valence and arousal ratings of learned words

3.3.1. Valence ratings

As expected, participants rated negative words more negatively than neutral words, which were rated more negatively than positive words (effect of valence: F(2,434) = 563.82, p < 0.001, η² = 0.72). Participants also rated arousing words more negatively than non-arousing words (effect of arousal: F(1,217) = 17.84, p < 0.001, η² = 0.08). Stressed participants rated positive words more negatively than non-stressed participants (Stress×Valence interaction: F(2,434) = 3.34, p < 0.05, η² = 0.02). Females also rated negative, neutral, and arousing words more negatively than males did (Sex×Valence interaction: F(2,434) = 4.51, p < 0.05, η² = 0.02; Sex×Arousal interaction: F(1,217) = 26.36, p < 0.001, η² = 0.11).

3.3.2. Arousal ratings

As expected, arousing words were given higher arousal ratings than non-arousing words (effect of arousal: F(1,217) = 113.08, p < 0.001, η² = 0.34). Participants rated positive words as more arousing than negative words, which were rated as more arousing than neutral words (effect of valence: F(2,434) = 39.33, p < 0.001, η² = 0.15). Females assigned negative words greater arousal ratings than did males (Sex×Valence interaction: F(2,434) = 7.41, p < 0.001, η² = 0.03). Non-stressed females assigned greater arousal ratings to words, overall, than did stressed females, whereas stressed males assigned greater arousal ratings to words, overall, than did non-stressed males (Stress×Sex interaction: F(1,217) = 7.13, p < 0.01, η² = 0.03). Non-stressed carriers of the BclI polymorphism assigned negative words greater arousal ratings than did non-stressed non-carriers of the polymorphism (Stress×Genotype×Valence interaction: F(2,434) = 4.77, p < 0.01, η² = 0.02).

3.4. Memory testing

3.4.1. Immediate recall

Participants recalled more positive and negative words than neutral words (effect of valence: F(2,434) = 13.36, p < 0.001, η² = 0.06). They also recalled more arousing words than non-arousing words, particularly when the words were positive (Valence×Arousal interaction: F(2,434) = 14.83, p < 0.001, η² = 0.06). There were no significant effects of stress or genotype on immediate free recall (Fig. 4).

Fig. 4.

Immediate recall performance. When analyzing the data as a function of stress condition, there were no significant differences in immediate recall (a). However, after breaking down stressed participants into cortisol responders and non-responders, the analyses revealed that stress enhanced immediate recall in non-carriers of the BclI polymorphism who exhibited a robust cortisol response to the stressor (b). Data are expressed as means ± SEM. *p < 0.05 relative to non-responders and no stress (LSD post hoc test).

3.4.2. Delayed recall

3.4.2.1. Raw data.

Participants recalled more positive words than negative words, which were recalled more than neutral words (effect of valence: F(2,434) = 6.55, p < 0.01, η² = 0.03). They also recalled more arousing words than non-arousing words, particularly when they were positive (Valence×Arousal interaction: F(2,434) = 12.72, p < 0.001, η² = 0.06). Females recalled more words than males, particularly when the words were positive or neutral (Sex×Valence interaction, F(2,434) = 3.17, p < 0.05, η² = 0.01). Stress immediately before learning selectively enhanced recall in males, relative to non-stressed males, but had no effect on females (Stress×Stress Time Point×Sex interaction: F(1,217) = 6.80, p < 0.01, η² = 0.03; Fig. 5). Stress 30 min before learning impaired recall in males, relative to stressed females, who tended to exhibit enhanced memory relative to non-stressed females (p = 0.08). Independent of time point, stress enhanced recall in male non-carriers of the BclI polymorphism, relative to non-stressed non-carriers of the polymorphism (Stress×Sex×Genotype interaction: F(1,217) = 3.97, p < 0.05, η² = 0.02). In contrast, stress impaired recall in male carriers of the polymorphism, relative to stressed male non-carriers, stressed female carriers, and stressed female non-carriers. None of the effects of stress, sex, and genotype on memory depended on word valence or arousal level.

Fig. 5.

Delayed recall performance, expressed as raw data (a, b) and percent of immediate recall (c, d). Stress immediately before learning selectively enhanced recall in males (a), while stress 30 min before learning selectively impaired recall in males. Stress, independent of when it was administered, selectively enhanced recall in male non-carriers of the BclI polymorphism (b, c), an effect which was positively associated with cortisol response to the stressor (d). Data are expressed as means ± SEM. *p < 0.05 relative to non-stressed males; **p < 0.05 relative to stressed females; ***p < 0.05 relative to non-stressed male non-carriers; τ=p < 0.05 relative to all other stress conditions; β=p=0.08 relative to non-stressed females (LSD post hoc tests).

3.4.2.2. Percent of immediate recall.

Upon analyzing delayed recall as a percent of the number of words recalled on Day 1, stress, independent of when it was administered, enhanced recall in male non-carriers of the BclI polymorphism, relative to non-stressed non-carriers of the polymorphism (Stress×Sex×Genotype interaction: F(1,217) = 4.16, p < 0.05, η² = 0.02; Fig. 5). None of the effects of stress, sex, and genotype on memory depended on word valence or arousal level.

3.4.3. Recognition memory

3.4.3.1. Discrimination index (d’).

Overall, participants recognized more neutral words than positive words and more positive words than negative words (effect of valence: F(2,434) = 8.90, p < 0.001, η² = 0.04). Participants also recognized more arousing than non-arousing words, particularly if they were negative or neutral (Valence×Arousal interaction: F(2,434) = 5.99, p < 0.01, η² = 0.03). Stress immediately before learning enhanced recognition of non-arousing words, while stress 30 min before learning impaired recognition of non-arousing words (Stress×Stress Time Point×Arousal interaction: F(1,217) = 6.44, p < 0.05, η² = 0.03). Stress, independent of time point, enhanced recognition memory in non-carriers of the BclI polymorphism, but it impaired recognition memory in carriers (Stress×Genotype interaction: F(1, 217) = 8.50, p < 0.01, η² = 0.04; Fig. 6). Stress immediately before learning enhanced recognition memory in males, while stress 30 min before learning impaired recognition memory in males; such effects were not observed in females (Stress×Sex×Stress Time Point interaction: F(1,217) = 7.02, p < 0.01, η² = 0.03).

Fig. 6.

Recognition memory performance. Stress immediately before learning enhanced recognition memory of non-arousing words, while stress 30 min before learning impaired recognition of non-arousing words (a). Stress immediately before learning selectively enhanced recognition memory in males, and stress 30 min before learning selectively impaired recognition memory in males (b). Stress, independent of when it was administered, selectively enhanced recognition memory in non-carriers of the BclI polymorphism (c), which appear to be positively associated with cortisol response to the stressor (d). Data are expressed as means ± SEM. *p < 0.05 relative to respective no stress group (LSD post hoc tests).

3.4.3.2. Hit rate.

Participants exhibited greater hit rates for arousing words than non-arousing words (effect of arousal: F(1,217) = 8.30, p < 0.01, η² = 0.04). Stress, independent of time point, led to greater hit rates in non-carriers of the BclI polymorphism, but it led to lower hit rates in carriers (Stress×Genotype interaction: F(1,217) = 10.54, p < 0.01, η² = 0.05). Stress immediately before learning led to greater hit rates in males, while stress 30 min before learning led to lower hit rates in males; such effects were not observed in females (Stress×Sex×Stress Time Point interaction: F(1,217) = 7.40, p < 0.01, η² = 0.03).

3.4.3.3. False alarm rate.

Participants exhibited greater false alarm rates for negative words than for positive words, and the false alarm rate for positive words was greater than it was for neutral words (effect of valence: F(2,434) = 18.65, p < 0.001, η² = 0.08). Participants also exhibited lower false alarm rates for arousing words than for non-arousing words, particularly if they were negative or neutral (Valence×Arousal interaction: F(2,434) = 8.90, p < 0.001, η² = 0.04). There were no significant effects of stress or genotype on false alarm rate.

3.4.4. Analyses based on cortisol response to the stressor

Because previous work has shown that stress and its time-dependent effects on learning and memory can be influenced by the magnitude of one’s cortisol response to stress and because the BclI polymorphism has been associated with altered HPA axis responses to stress, we divided stressed participants into cortisol “responders” and “non-responders” and compared their memory performance following stress exposure. Specifically, stressed participants exhibiting a cortisol increase of at least 2.5 nmol/l following the SECPT were considered responders (N = 57; 25 males, 32 females); all other stressed participants were considered non-responders (N = 62; 22 males, 40 females). This cutoff criterion corresponds to an elevation of approximately 1 μg/dL of serum or plasma cortisol and is thought to reflect a cortisol secretory episode that would occur following a stressor (Kirschbaum, Pirke, & Hellhammer, 1993; Smeets et al., 2012; Zoladz et al., 2017a, 2014a).

Mixed-model ANOVAs were performed on the memory data, as described above, except “responder” was included in place of “stress” as one of the between-subjects factors. The analyses revealed a trend that stress, independent of when it was administered, selectively enhanced immediate recall in non-carriers of the BclI polymorphism who exhibited a robust cortisol response to the stressor (i.e., cortisol responders) (Responder×Genotype interaction: F(2,208) = 2.92, p = 0.056, η² = 0.03; see Fig. 4). A similar effect was observed for delayed recall. When analyzing delayed recall as a percent of the number of words recalled on Day 1, stress selectively enhanced recall in male non-carriers of the BclI polymorphism who exhibited a robust cortisol response to the stressor (Responder×Sex×Genotype interaction: F(2,208) = 3.26, p < 0.05, η² = 0.03; Fig. 5). The analyses of recognition memory indicated that stress selectively enhanced recognition memory in non-carriers of the BclI polymorphism who exhibited a robust cortisol response to the stressor (Responder×Genotype interaction: F(2,208) = 4.45, p < 0.05, η² = 0.04; Fig. 6). These analyses also revealed that stress immediately before learning selectively enhanced recognition of non-arousing words in cortisol responders (Responder×Stress Time Point×Arousal interaction: F(2,208) = 4.21, p < 0.05, η² = 0.04). None of the effects of stress, sex, and genotype on memory depended on word valence or arousal level. Because some of the effects for memory performance were selectively observed in cortisol responders, we performed correlational analyses to assess whether stress-induced cortisol levels were associated with recall or recognition memory. However, none of these analyses revealed statistically significant relationships.

4. Discussion

Examining how certain susceptibility factors influence stress-induced alterations of physiology and behavior could provide important insight into mechanisms underlying the onset of stress-related psychological disorders. We have taken an interest in how such factors affect stress-memory interactions, which could help us understand who is more susceptible to producing traumatic memories. The purpose of the present study was to examine the influence of the BclI polymorphism of the glucocorticoid receptor gene (NR3C1) on the time-dependent effects of pre-learning stress on long-term memory. The results revealed that stress administered immediately before learning enhanced long-term memory, while stress administered 30 min before learning impaired long-term memory. These effects were, for the most part, selective to males and non-arousing information. Additionally, independent of when it was administered, stress enhanced memory in non-carriers of the BclI polymorphism but impaired memory in carriers. These effects were often selective to males and/or participants who exhibited a robust cortisol response to the stressor. Collectively, these findings provide further support for the time-dependent effects of stress on declarative learning and memory and suggest that carriers of the BclI polymorphism of the glucocorticoid receptor gene might be more sensitive to the negative influences of stress on learning and memory.

4.1. Time-dependent effects of stress on learning and memory

Researchers have shown that when stress is administered immediately before or after learning, long-term memory is generally enhanced (Cahill, Gorski, & Le, 2003; Diamond et al., 2007; Nielson and Arentsen, 2012; Nielson and Powless, 2007; Nielson, Yee, & Erickson, 2005; Quaedflieg et al., 2013; Vogel and Schwabe, 2016; Zoladz, Clark, et al., 2011; Zoladz et al., 2017c, 2014b). In contrast, when a stress experience is temporally separated from learning, long-term memory is generally impaired (Quaedflieg et al., 2013; Zoladz, Clark, et al., 2011; Zoladz et al., 2013). The immediate, enhancing effects of stress on learning appear to occur because there is a convergence of stress-induced neuromodulators in cognitive brain areas and because the stress is close enough in time to learning to become part of the learning context (Joels, Pu, Wiegert, Oitzl, & Krugers, 2006). Indeed, when stress is administered immediately before or after learning, the observed enhancement of memory is sometimes selective to context-related information (e.g., stress-related words) (Smeets et al., 2009). These immediate, enhancing effects are driven by synergistic activity of norepinephrine and corticosteroids in cognitive brain areas, such as the hippocampus and amygdala (De Voogd, Klumpers, Fernandez, & Hermans, 2017; Diamond et al., 2007; Hermans, Battaglia, et al., 2014; Joels et al., 2011; Schwabe et al., 2012). When stress is temporally separated from learning, learning occurs when the brain is in a “memory storage” mode (Schwabe et al., 2012) and, in order to facilitate future survival, cognitive brain areas are focused on consolidating the memory of what occurred during the stress experience. The negative effects of stress in this case are believed to depend on delayed activity of corticosteroids.

In the present study, stress immediately before learning selectively enhanced delayed recall and recognition in males, while stress 30 min before learning selectively impaired delayed recall and recognition in males. The observation of selective effects in males is not surprising. On several occasions, we have observed differential effects of pre-learning stress on memory in males and females and, as a result, have contended that the temporal dynamics of pre-learning stress effects on long-term memory might differ across the sexes (Zoladz et al., 2017b, 2014b, 2014c, 2013). In the present study, the selective enhancement of long-term memory in males could be attributable, in part, to better long-term memory in non-stressed females. This could have made a stress-induced enhancement of long-term memory in females more difficult to observe (e.g., a ceiling effect of sorts). In contrast, the effects of stress 30 min before learning appeared to be opposite across the sexes. While such stress impaired memory in males, females exhibited a statistical trend suggestive of enhanced memory. Collectively, the sex-dependent nature of our findings further support the notion that the time-dependent effects of stress on declarative learning differ between the sexes and that females are less sensitive to the effects of stress on learning and memory, particularly its impairing effects.

We also observed that stress immediately before learning selectively enhanced recognition of non-arousing words in participants who exhibited a robust cortisol response to the stressor, while stress 30 min before learning selectively impaired recognition of non-arousing words. These effects on recognition appeared to be driven by stress-induced alterations of participant hit rates. It is not uncommon to observe selective effects of stress on neutral, non-arousing information. Indeed, several studies have reported selective stress-induced enhancements and impairments of memory for neutral information (e.g., Payne et al., 2007, 2006; Schwabe, Bohringer, et al., 2008). One concern is how the enhancing effect of immediate stress was associated with cortisol levels. Previous work has established that circulating corticosteroids can exert rapid, non-genomic effects on synaptic plasticity by acting on membrane-bound receptors (Joels, Sarabdjitsingh, & Karst, 2012; Karst et al., 2005). Indeed, the rapid enhancing effects of stress and amygdala activation on hippocampus-dependent learning and synaptic plasticity have been associated with corticosteroid-dependent mechanisms (Akirav and Richter-Levin, 1999, 2002). In the present study, the enhancing effects of cortisol would have had to occur very quickly, as there was only an approximately 8-min gap of time between the onset of stress and the completion of word list learning. Although this seems perhaps too fast for corticosteroids to underlie the observed enhancement, previous work has shown that exogenous corticosteroid administration can exert cognitive effects in less than 10 min. For instance, investigators found that i.v. administration of cortisol 8 min prior to memory testing exerted dose-dependent, curvilinear effects on memory retrieval (Schilling et al., 2013). Others also revealed that i.v. administration of cortisol disrupted prepulse inhibition within 7 min of infusion (Richter et al., 2011).

4.2. BclI polymorphism and stress effects on learning

The influence of the BclI polymorphism on stress-induced alterations of learning and memory were independent of stressor timing. Stress selectively enhanced immediate and delayed recall in male non-carriers of the BclI polymorphism and enhanced recognition memory in male and female non-carriers of the polymorphism. Further analyses suggested an involvement of stress-induced corticosteroid activity in these effects. Specifically, the memory enhancements in stressed non-carriers of the polymorphism were selective to participants exhibiting a robust cortisol response to the stressor (i.e., responders). Because participants’ cortisol responses to the stress did not vary as a function of genotype, these findings suggest that the same stress-induced elevation in cortisol may have impacted learning and memory differently in carriers and non-carriers.

These findings suggest that carriers of the BclI polymorphism are resistant to the enhancing effects of immediate pre-learning stress on long-term memory. This finding bears striking resemblance to previous work from our laboratory in which memory performance in carriers of three different FKBP5 polymorphisms was also unaffected by immediate pre-learning stress (Zoladz et al., 2017c). The commonality between these polymorphisms is that they all influence corticosteroid receptor sensitivity, stress responsivity, and susceptibility to psychological illness. Based on the present findings, we would contend that BclI carriers exhibit heightened sensitivity to stress, particularly stress-induced corticosteroid activity.

The findings that stressed non-carriers, but not carriers, exhibited enhanced long-term memory may seem to be at odds with previous work, as other investigators have reported enhanced emotional memory in BclI carriers. Ackermann and colleagues found that homozygous carriers of the BclI polymorphism displayed greater recall of emotionally arousing pictures than heterozygous carriers and non-carriers of the polymorphism (Ackermann et al., 2013). Moreover, Hauer et al. reported that homozygous carriers of the BclI polymorphism exhibited more long-term traumatic memories following cardiac surgery and intensive care unit therapy than heterozygous carriers and non-carriers of the polymorphism (Hauer et al., 2011). It may be the case that the BclI polymorphism, under baseline conditions, sensitizes receptors to corticosteroid activity (Huizenga et al., 1998; van Rossum et al., 2003), which results in ideal biological conditions for learning and memory. However, when stress is added to the system, the polymorphism might result in an excessive sensitivity to corticosteroids, one that pushes the effect on cognition to the right of an inverted U-shaped relationship between corticosteroid activity and cognition (Diamond, Bennett, Fleshner, & Rose, 1992; Domes, Rothfischer, Reichwald, & Hautzinger, 2005; Lupien and Lepage, 2001; Lupien et al., 2002; Salehi, Cordero, & Sandi, 2010).

Stress also selectively impaired delayed recall and recognition memory in carriers of the BclI polymorphism. The impairment of delayed recall was selective to males. These observations are consistent with the idea presented above, that carriers of the BclI polymorphism exhibit greater sensitivity to the negative effects of stress on long-term memory. Because carriers of the BclI polymorphism exhibited greater corticosteroid receptor sensitivity, any stress, regardless of when it is administered, could result in learning that is suboptimal.

4.3. Theoretical explanations

Researchers examining the effects of stress on cognition have distinguished between two large scale neural networks, the salience network and the executive control network, that are differentially influenced by stress (Hermans, Henckens, Joles, & Fernandez, 2014; Hermans et al., 2011; Homberg, Kozicz, & Fernandez, 2017; Young et al., 2017). The salience network, which drives bottom-up processing and attention to salient (e.g., threatening) stimuli in the environment, consists of structures such as the cingulate gyrus, amygdala, hippocampus, and striatum. The executive control network, which drives top-down processing and working memory, planning, and decision making, consists of structures in the prefrontal and parietal cortices. Investigators found that stress increases the functional connectivity of the salience network, at the expense of the executive control network (Hermans, Henckens, et al., 2014; Hermans et al., 2011; Homberg et al., 2017; Young et al., 2017). Changes in salience network connectivity were correlated with stress-induced increases in cortisol, noradrenergic activity, and subjective measures of affect. Thus, we propose that there is a brief (i.e., minutes) window of time following stress onset during which the memory processing ability of select cognitive brain areas in the salience network, such as the amygdala and hippocampus, is enhanced (De Voogd et al., 2017; Diamond et al., 2007; Hermans, Battaglia, et al., 2014; Joels et al., 2011; Schwabe et al., 2012). This enhancement would result from increased noradrenergic activity and non-genomic corticosteroid actions at membrane-bound receptors, both of which significantly increase amygdala and hippocampus excitability (Hermans, Battaglia, et al., 2014; Karst, Berger, Erdmann, Schutz, & Joels, 2010). If learning that is dependent on such brain regions occurred during this time frame, long-term memory would be facilitated. The window of time following stress during which such learning would be enhanced is still not completely defined, but at some point (likely more than 10–20 min after stress, depending on several factors related to stress intensity and individual differences) (Hermans, Henckens, et al., 2014; Zoladz, Clark, et al., 2011; Zoladz et al., 2017b, 2017c, 2014b, 2014c, 2013), the memory processing ability of these brains areas would be inhibited. This inhibition likely results from gradually increasing levels of circulating corticosteroids, which facilitate recovery from stress by reducing salience network activity (Hermans, Henckens, et al., 2014; Homberg et al., 2017). This would theoretically allow cognitive brain areas from the salience network to focus on and store the information that was encoded during the stress experience.

It is nonetheless important to point out that the delayed effects of stress on learning and memory that we have observed here and in previous work may not be the result of gene-dependent corticosteroid activity, which takes at least an hour to manifest (Hermans, Henckens, et al., 2014). Instead, we posit that “rapid” corticosteroid activity might be exerting time-dependent effects on learning and memory. Research has shown that acute stress causes organisms to switch from a cognitive-based (hippocampus-dependent) learning strategy to a more habit-based (striatum-dependent) learning strategy (Schwabe and Wolf, 2013). This switch is likely an adaptive response to stress that allows an organism to behave automatically during threatening situations, freeing up resources to deal with the stressor. Recent work has shown that the shift from a declarative memory strategy to a procedural strategy is dependent on corticosteroid activity, particularly at the mineralocorticoid receptor (MR) (Schwabe, Schachinger, de Kloet, & Oitzl, 2010; Schwabe, Tegenthoff, Hoffken, & Wolf, 2013; Vogel et al., 2016). Furthermore, the shift appears to occur rapidly, as acute stress results in increased amygdala-striatum connectivity, along with a concomitant decrease in amygdala-hippocampus connectivity, less than 20 min after stress onset (Vogel et al., 2015). Thus, we propose that immediately following stress, rapid increases in noradrenergic and non-genomic corticosteroid activity leads to enhanced amygdala-hippocampus connectivity, which results in enhanced declarative learning and memory. However, after some period of time (e.g., 10–20 min), rising corticosteroid levels act on membrane-bound receptors to switch learning and memory strategies to a procedural mode. This results in impaired declarative learning and memory.

What the present and previous work from our laboratory reveals is that the temporal dynamics of pre-learning stress effects on learning and memory depend on several factors, including sex, physiological response to the stressor, and genetic characteristics of the individual. Here, we have shown that carriers of the G allele of the BclI polymorphism do not exhibit enhanced long-term memory when stressed is administered immediately before learning and may exhibit enhanced sensitivity to the impairing effects of stress. Because carriers of the BclI polymorphism exhibit greater corticosteroid receptor sensitivity, the rapid physiological responses that follow stress onset would have greater effects, thus causing the stress-induced switch in learning strategy to occur more rapidly and result in greater inhibitory effects on hippocampus-dependent learning and memory. This means that carrying the BclI polymorphism could result in incomplete, perhaps fragmented, memories following stress.

5. Limitations

The present findings are noteworthy because, along with several previous publications from our laboratory, they highlight the influence of genetic variants on stress-memory interactions. Nonetheless, our findings are preliminary, and there are some limitations of this work that deserve attention. First, our overall sample ended up being slightly underpowered, according to the a priori power analysis that was performed. This could explain why some observations were statistical trends. Moreover, some of the effects that were observed between males and females and between cortisol responders and non-responders were based on small cell sizes. Thus, the effects should be considered preliminary and interpreted cautiously. Second, some of the observed effects were spread across recall and recognition assessments. We have observed similar effects in previous work, and it is possible that the effects spread across recall and recognition memory were simply different manifestations of similar stress-induced alterations of encoding and/or consolidation. Third, we only examined cortisol levels at two time points on Day 1 because we were primarily interested in obtaining a baseline sample, as well as a sample that would illustrate peak stress-induced cortisol levels (particularly at the time right before the delayed stress group learned). Obtaining additional samples would have allowed us to assess the influence of genotype on cortisol recovery following stress and perhaps how that interacted with learning and memory. Fourth, we did not time word list learning in the present study, and one could argue that such time might differ between stress and control groups. However, data from previous work employing identical methodology has revealed no significant effect of stress on word list learning duration. Lastly, most studies examining the influence of the BclI polymorphism on physiology and behavior have not combined heterozygous G allele carriers with homozygous G allele carriers; instead, these studies have observed effects primarily for homozygous G allele carriers. We combined the two groups in the present study to increase statistical power. However, it is possible that homozygous G allele carriers were the driving force behind any observed gene-dependent effects.

6. Conclusions and future directions

We have replicated previous work by demonstrating that prelearning stress time-dependently influences long-term memory. Specifically, stress administered immediately before learning enhanced long-term memory, while stress that was temporally separated from learning impaired long-term memory. We have also revealed that the BclI polymorphism, which results in greater corticosteroid receptor sensitivity, can influence these effects. Independent of when it was administered, stress enhanced immediate and delayed recall in male non-carriers and enhanced recognition memory in male and female non-carriers; these effects were selectively observed in participants who exhibited a robust cortisol response to the stressor. Stress also impaired delayed recall in male carriers and recognition memory in male and female carriers. Collectively, these findings suggest that carriers of the BclI polymorphism might be more sensitive to the negative effects of corticosteroids on learning and memory processes.

As researchers continue to examine the effects of stress and corticosteroid administration on cognition and the functional connectivity of cognitive brain areas, it is important to keep in mind that stress is not synonymous with HPA axis activity. Indeed, the time-dependent influences of stress versus corticosteroids on cognitive processes appear to differ. For instance, investigators consider the “rapid” effects of corticosteroids to last up to 75 min (Henckens, van Wingen, Joels, & Fernandez, 2010, 2011, 2012; van Ast, Cornelisse, Meeter, Joels, & Kindt, 2013), but studies have revealed both “rapid” (i.e., enhancing) and “delayed” (i.e., impairing) effects of stress within this time frame (Quaedflieg et al., 2013; Schwabe, Bohringer, et al., 2008; Zoladz, Clark, et al., 2011; Zoladz et al., 2017b, 2017c, 2014c, 2013). Thus, future work needs to clarify the time course of stress-induced enhancements versus impairments of long-term memory, as well as the mechanisms underlying this transition. It is also important to emphasize the continued sex-dependent effects of stress on learning and memory. Future work should elucidate the mechanisms responsible for males and females exhibiting different time-dependent effects of stress on long-term memory.