Critical brain circuits at the intersection between stress and learning

Abstract

The effects of stressful life experience on learning are pervasive and vary greatly both within and between individuals. It is therefore unlikely that any one mechanism will underlie these complicated processes. Nonetheless, without identifying the necessary and sufficient circuitry, no complete mechanism or set of mechanisms can be identified. In this review, we provide two anatomical frameworks through which stressful life experience can influence processes related to learning and memory. In the first, stressful experience releases stress hormones, primarily from the adrenals, which directly impact brain areas engaged in learning. In the second, stressful experience indirectly alters the circuits used in learning via intermediary brain regions. Importantly, these intermediary brain regions are not integral to the stress response or learning itself, but rather link the consequences of a stressful experience with circuits used to learn associations. As reviewed, the existing literature provides support for both frameworks, with somewhat more support for the first but sufficient evidence for the latter which involves intermediary structures. Once we determine the circumstances that engage each framework and identify which one is most predominant, we can begin to focus our efforts on describing the neuronal and hormonal mechanisms that operate within these circuits to influence cognitive processes after stressful life experience.

Figure 1. Adrenalectomy, but not demedullation, prevents the stress-induced enhancement of hippocampal dependent trace eyeblink conditioning (Beylin and Shors, 2003)

Stress increased the percentage of CRs emitted in sham operated but not in adrenalectomized (ADX) male rats (A). Note that basal levels of corticosterone were provided in drinking water. Demedullation (Demed), which leaves the adrenal cortex and corticosterone production intact, while removing adrenal catecholamine production, failed to prevent the stress-induced enhancement of trace conditioning, indicating that these effects are specific to corticosterone (B). Data are represented as Mean ± SEM percentage of CRs averaged across 300 training trials. Asterisks indicate a significant difference (p<0.05).

Figure 2. This is a schematic representation of the two models of stress and learning interactions

In model 1 stress hormones directly impact learning circuitry (A). In model 2 stress hormones act via intermediary structures to impact learning circuitry (B).

Figure 3. Opposite effects on stress on hippocampal-dependent trace conditioning and dendritic spines in the hippocampus in males vs. females (Shors et al., 2001a; (Waddell et al., 2008)

Under unstressed conditions, females emit more CRs than males. However, stressor exposure enhances trace conditioning in males, but impairs it in females (A). Similarly under unstressed conditions, females in proestus have greater spin density than males. Stressful experience increases spine density in males, while it decreases spine density in females (B). The altered spine density following stress is observed on apical dendrites of CA1 pyramidal neurons (C).

Figure 4. The hippocampus is required for stress to modulate learning, even when learning itself is independent of the hippocampus (Bangasser and Shors, 2007)

Hippocampal lesions prevented the stress-induced enhancement of delay eyeblink conditioning in male rats (A) and the stress-induced impairment of conditioning in female rats (B). Data are represented as Mean ± SEM percentage of CRs over 600 training trials (150 trials per day).

Figure 5. Activation of the amygdala during stressor exposure is required for stress to modulate learning (Waddell et al., 2008)

Temporary inactivation of the amygdala during stressor exposure prevented enhanced conditioning in males (A) and impaired conditioning in female rats (B). Data are represented as Mean ± SEM percentage of CRs over 600 training trials (150 trials per day).

Figure 6. The BNST is required for stress effects on learning in masculinized, but not in cycling (i.e., normal) female rats (Bangasser et al., 2005)

In cycling females, BNST inactivation at any timepoint failed to prevent impaired conditioning (A). Just like in males, masculinized females require their BNST during training for enhanced conditioning after stress (B). Data are represented as Mean ± SEM percentage of CRs over 600 training trials (150 trials per day).

Figure 7. This figure illustrates how the hippocampus, amygdala, and BNST could affect the circuitry necessary for classical eyeblink conditioning

Solid lines represent possible connections in males and females, dashed lines represent possible connections in males only.