Unexpected Role of Physiological Estrogen in Acute Stress-Induced Memory Deficits

Abstract

Stress may promote emotional and cognitive disturbances, which differ by sex. Adverse outcomes, including memory disturbances, are typically observed following chronic stress, but are now being recognized also after short events, including mass shootings, assault, or natural disasters, events that consist of concurrent multiple acute stresses (MAS). Prior work has established profound and enduring effects of MAS on memory in males. Here we examined the effects of MAS on female mice and probed the role of hormonal fluctuations during the estrous cycle on MAS-induced memory problems and the underlying brain network and cellular mechanisms. Female mice were impacted by MAS in an estrous cycle-dependent manner: MAS impaired hippocampus-dependent spatial memory in early-proestrous mice, characterized by high levels of estradiol, whereas memory of mice stressed during estrus (low estradiol) was spared. As spatial memory requires an intact dorsal hippocampal CA1, we examined synaptic integrity in mice stressed at different cycle phases and found a congruence of dendritic spine density and spatial memory deficits, with reduced spine density only in mice stressed during high estradiol cycle phases. Assessing MAS-induced activation of brain networks interconnected with hippocampus, we identified differential estrous cycle-dependent activation of memory- and stress-related regions, including the amygdala. Network analyses of the cross-correlation of fos expression among these regions uncovered functional connectivity that differentiated impaired mice from those not impaired by MAS. In conclusion, the estrous cycle modulates the impact of MAS on spatial memory, and fluctuating physiological levels of sex hormones may contribute to this effect.SIGNIFICANCE STATEMENT: Effects of stress on brain functions, including memory, are profound and sex-dependent. Acute stressors occurring simultaneously result in spatial memory impairments in males, but effects on females are unknown. Here we identified estrous cycle-dependent effects of such stresses on memory in females. Surprisingly, females with higher physiological estradiol experienced stress-induced memory impairment and a loss of underlying synapses. Memory- and stress-responsive brain regions interconnected with hippocampus were differentially activated across high and low estradiol mice, and predicted memory impairment. Thus, at functional, network, and cellular levels, physiological estradiol influences the effects of stress on memory in females, providing insight into mechanisms of prominent sex differences in stress-related memory disorders, such as post-traumatic stress disorder.

Keywords: estrogen; hippocampus; memory; sex differences; stress; synapses.

Figure 1.

Spatial memory impairment following MAS is limited to mice entering proestrus. A, For the Objection Location Memory (OLM) task, mice were habituated to the apparatus several days before MAS. At 2 h after MAS, mice were trained and then memory was tested 24 h later. B, Estrus control, estrus MAS, and early proestrus control mice preferentially explored the object in a novel location, whereas early proestrus MAS mice explored both objects equivalently (n = 9-11/group). C, For the spatial Y-maze task, 2 h after MAS, mice were trained in the apparatus with one arm closed. After 1 h, mice were reintroduced to the maze with the previously closed arm (the novel arm) now open. D, Most mice of both control groups and most estrus MAS mice entered the novel arm as their first choice, whereas the first entry being the novel arm for MAS early proestrus mice was below chance. E, Estrus control, estrus MAS, and early proestrus control mice entered the novel arm more frequently than the familiar arm, whereas early proestrus MAS mice entered the novel and familiar arms equally (n = 7–9 per group). *p < 0.05 (post-test). Points represent scores of individual animals. Connected points are matched samples within an animal. Error bars indicate ± SEM.

Figure 2.

Early proestrous mice (those with impaired memory following MAS) have higher levels of circulating estradiol. Estrous cycles were monitored via vaginal cytology and at the time of the experiment mice were divided into groups that were early proestrus or estrus. A, Vaginal cytology classifications were done according to relative presence of nucleated epithelial, cornified epithelial, or leukocytes in the sample. High E2 mice had a majority nucleated cells, whereas low E2 mice had smears consisting of almost entirely cornified cells (n = 11-18/group). B, Mice classified as early proestrus according to their vaginal smears had higher average estradiol in serum samples as measured by an estradiol ELISA compared with estrus mice (n = 11-18 mice/group). C, High E2 mice had higher average uterine indices (uterus weight/body weight × 100) (n = 8-17 mice per group). D, The amount of estradiol within a sample had a significant positive correlation with the percentage of the smear that consisted of nucleated epithelial cells. E, Estradiol had a negative correlation with cornified cells. From this point on, early proestrus mice were classified as high E2 and estrus mice were classified as low E2. ^#p < 0.05, main effect. Points represent scores of individual animals. Error bars indicate ± SEM.

Figure 3.

Dendritic spine loss in CA1 of the dorsal hippocampus is induced by MAS in high E2 but not low E2 female mice. Dendritic spines were visualized in mice expressing YFP in pyramidal cells under the control of the Thy1 promoter. A, Under control conditions, mice with higher levels of estrogen had more thin spines but no difference in mushroom or total (thin + mushroom) spines. B, High E2 control mice had thin (thin arrows) and mushroom spines (thick arrows) in the stratum radiatum of CA1. C, After MAS, high E2 mice selectively lost thin spines. Framed areas are enlarged in B-2 and C-2. Scale bars: B-1, C-1, 10 µm; B-2, C-2, 2 µm. D, Following 2 h MAS, there was no significant difference in total number of spines between control or MAS mice in either cycle phase. E, Thin spines were greater in high E2 control mice than low E2 control mice, but these spines were selectively reduced following MAS in the high E2 phase. F, Mushroom spines remained intact following MAS in either cycle phase (n = 3-5 mice/group). *p < 0.05 (post-test). Points represent scores of individual animals. Error bars indicate ± SEM.

Figure 4.

MAS-induced memory impairments are not explained by differential activation of the dorsal hippocampus. Activation of the dorsal hippocampus and hypothalamic paraventricular nucleus (PVN) was assessed by quantifying fos⁺ cells in control or MAS mice in either cycle phase. A, fos⁺ cells did not differ with cycle phase or MAS in the CA1 region. B, Numbers of fos⁺ cells also did not differ with cycle phase or MAS in the CA2 and CA3 regions. C, In the DG, there was an effect of cycle on fos⁺ cells, with cell number distinguishing low E2 and high E2 in control but not MAS mice. D, E, fos⁺ cells in the hypothalamic PVN were more abundant following MAS in both groups, with a greater increase in activation in the high E2 group (n = 5-7 mice per group). Scale bar, 100 µm. *p < 0.05 (post-test). Points represent scores of individual animals. Error bars indicate ± SEM.

Figure 5.

Neuronal activation across the brain varies with cycle phase and in response to MAS. fos⁺ cells were quantified in the BLA, MeA, BNST, MS, LS, and the PVT. A, Whereas there were fewer fos⁺ cells in BLA at baseline in the high E2 group, MAS resulted in a significant increase in fos⁺ cells in this group only. B, In the MeA, MAS increased the number of fos⁺ cells significantly in high E2 mice but not in low E2 mice. C, In the BNST, MAS increased fos⁺ cell numbers in the low E2 group only. D, In the MS, there was a main effect of MAS that did not differ between mice at different cycle phases. E, In the LS, there was an increase in fos⁺ cells following MAS in the high E2 group, but not the low E2 group. F, In the PVT, the number of fos⁺ cells was augmented by MAS in the high E2 group only. G, Graphic summary of differences in fos counts across brain regions (n = 5-7 mice per group). *p < 0.05 (post-test). Points represent scores of individual animals. Error bars indicate ± SEM.

Figure 6.

Correlated neuronal activity is influenced by estrous cycle phases and MAS. Patterns of neuronal activity were inferred by computing the Spearman correlations of scaled counts of fos⁺ cells among all brain regions. Within conditions, correlation matrices were computed for the following: A, High E2 control. B, High E2 MAS mice. C, Low E2 control. D, Low E2 MAS. As an example of MAS-induced changes of correlated activity, the yellow rectangles represents correlations with the PVN. Many PVN correlations are negative in high E2 control but shift to positive with MAS, whereas such a change is not evident among low E2 conditions.

Figure 7.

Comparing correlated neuronal activity among groups uncovers connectivity patterns that may contribute to MAS-induced memory impairments. To compare the differently active networks between conditions, Spearman correlations (see Fig. 6) were transformed to z scores that were compared between pairs of groups. Differential connectivity networks were constructed that indicated relationships that were increased (blue) or decreased (red) in Group 1 (first group listed) compared with Group 2 (second group listed). Line thickness indicates the intensity of this difference. Comparative functional networks were constructed for the following: A, Control: high E2 compared with low E2. B, High E2: MAS compared with control. C, Low E2: MAS compared with control. D, MAS: high E2 compared with low E2 mice.