Why the estrous cycle matters for neuroscience

Behavioral analyses

We previously compared diestrus females, proestrus females, and males across three different tests for anxiety-related behavior, including open field, light dark box, and elevated plus maze (Fig. 2A) [5]. Across all tests, diestrus females exhibited higher anxiety indices than proestrus females, while a sex difference was found between diestrus and male groups only (Fig. 2A) [5]. Specifically, in the open field, there was a significant effect of group on the time spent in the center [F(2,39) = 5.93, P = 0.006], with post hoc test showing diestrus females spending less time in the center compared to both proestrus females (P = 0.006) and males (P = 0.044, Fig. 2A) [5]. In the light dark box test, we found a significant difference between groups in the time spent in the light compartment [F(2,37) = 21.63, P < 0.001], which was driven by diestrus females spending less time in the light than both proestrus females (P < 0.001) and males (P < 0.001, Fig. 2A) [5]. Finally, in the elevated plus maze test, we saw a significant effect of group on the time spent in the open arms of the maze [F(2, 37) = 5.33, P = 0.009], with diestrus females spending less time in the open arms compared to proestrus females (P = 0.008) and there was a similar trend in the diestrus–male comparison (P = 0.084, Fig. 2A) [5].

Notably, when the two female groups are merged and compared to males (Fig. 2A), none of the behavioral comparisons between males and females reached statistical significance including the open field test [t(30.06) = 1.05, P = 0.303], the light–dark box [t(26.08) = 1.75, P = 0.092], and the elevated plus maze [t(22.05) = 0.71, P = 0.483].

Taking the light–dark box test further as an example, we visualized the normal distributions of the data, comparing mixed female and male group distributions, as well as distributions of separate diestrus, proestrus, and male groups (Fig. 2B). We found a substantial overlap between males and mixed-females (76%), as previously reported for many neurobehavioral measures [35]. However, when females are separated by the estrous cycle stage, there is a high overlap between proestrus and males only (82%), but little overlap between males and diestrus (34%) and even less overlap within females, between proestrus and diestrus groups (20%) (Fig. 2B).

We then addressed data variability between both males and merged females, as well as between proestrus, diestrus, and males for all three anxiety tests. For the time spent in the center of the open field, we found equal variance between proestrus females, diestrus females, and males [F(2, 39) = 0.80, P = 0.456; Levene’s test]; we also found equal variance [F(1, 40) = 1.21, P = 0.279; Levene’s test] and equal distribution shapes (D = 0.2, P = 0.823; Kolmogorov–Smirnov test) between merged females and males. We then looked into the time spent in the light compartment of the light–dark box test and found equal variance between proestrus, diestrus, and males [F(2, 37) = 0.46, P = 0.633; Levene’s test]; we also found equal variance [F(1, 38) = 1.63, P = 0.209; Levene’s test] and equal distribution shapes (D = 0.25, P = 0.591; Kolmogorov–Smirnov test) between merged females and males. Finally, we examined the time spent in the open arms of the elevated plus maze and found equal variance between proestrus, diestrus, and males [F(2, 37) = 0.83, P = 0.443; Levene’s test]; we also found equal variance [F(1, 38) = 0.01, P = 0.925; Levene’s test] and equal distribution shapes (D = 0.14, P = 0.986; Kolmogorov–Smirnov test) between merged females and males.

Overall, these data show that including the estrous cycle stage as a variable allows us to find the sex difference in anxiety-related behavior, which would be masked if the mixed female group was compared to males. Interestingly, we also found that the significant effect of the estrous cycle was not accompanied by the increased female variability compared to males, for any of the measured outcomes. In fact, we see similar variability between male and female groups, whether taking into account the estrous cycle or not.

Analysis of dendritic spine density

To extend our study to other neurobehavioral outcomes, we performed similar analyses of dendritic spine density in the ventral hippocampus (Fig. 3). We previously analyzed spine density in males, proestrus females, and diestrus females and found a significant group effect [F(2, 597) = 1907, P < 0.001], with proestrus females having a higher density than both diestrus females (P < 0.001) and males (P < 0.001; Fig. 3A) [5]. In this example, females either have significantly higher, or equal, dendritic spine density in comparison to males depending on their estrous cycle stage. Importantly, when the two female groups are merged, this dynamism in the sex difference is lost and merged females are observed to have higher spine density than males [t(540.91) = 16.04, P < 0.001; Fig. 3A].

When we analyzed this data using normal distributions (Fig. 3B), we found a partial overlap between males and mixed females (38%). However, after separating females by their estrous cycle stage, we found a large overlap between males and diestrus females (90%), and virtually no overlap between males and proestrus females (1%) or within females, between proestrus and diestrus (1%), illustrating how the information about the estrous cycle gives new insight into the data.

We also tested data variability for dendritic spine density in the ventral hippocampus. We found unequal variance between males, proestrus females, and diestrus females [F(2, 597) = 5.65, P = 0.004; Levene’s test], as well as between males and merged female groups [F(1, 598) = 530.88, P < 0.001; Levene’s test]. We also found that distribution shapes were unequal between merged females and males (D = 0.49, P < 0.001).

Overall, this data provides an example where females have higher variability than males, and a sex difference can be found without accounting for the estrous cycle. However, having the information about the estrous cycle explains where the sex-based variability is coming from and allows for a mechanistic insight, which is that the sex difference is driven by sex hormone changes in females.

Gene expression analysis

We further looked into our molecular data, including ventral hippocampal gene expression of two genes: Ptprt, (encoding protein tyrosine phosphatase receptor type T), involved in the development of dendrite spines [36]; and Htr2b (encoding serotonin receptor 2b), important for anxiety-related behavior [37] (Fig. 4).

For Ptprt, we observed a similar pattern that we observed with the dendritic spine density data (Fig. 3). Comparing diestrus, proestrus, and males, we found a significant effect of group on Ptprt expression [F(2, 21) = 483.2, P < 0.001], with proestrus females having higher expression than both diestrus females (P < 0.001) and males (P < 0.001), and with males having higher expression than diestrus females (P = 0.002; Fig. 4A) [5]. When the two female groups are merged, this dynamic sex difference is reduced to the merged female group exhibiting higher overall Ptprt expression compared to males [t(16.18) = 2.72, P = 0.015; Fig. 4A]. We also created distributions for this dataset (Fig. 4B), and found that there is a small overlap between males and mixed females (23%), with the distribution of mixed females appearing notably flatter. After separating the female groups, there is a modest overlap between diestrus and males (35%), and there is no overlap between proestrus and males (0%) or within females, between proestrus and diestrus (0%), indicating these groups form entirely distinct populations in measures of ventral hippocampal Ptprt expression.

For the second gene, Htr2b, we found a significant difference between diestrus females, proestrus females, and males [F(2, 21) = 12.87, P < 0.001], with diestrus females having higher expression than both proestrus (P < 0.001) and male (P = 0.013) groups (Fig. 4A) [5]. When the two female groups were merged, however, we found no difference between males and females [t(21.61) = 0.65, P = 0.520; Fig. 4A].

We then looked into data variability, both among proestrus, diestrus, and male groups, as well as between the merged female and male groups. For expression of Ptprt in the ventral hippocampus, we found equal variance between proestrus females, diestrus females, and males [F(2, 21) = 0.19, P = 0.829; Levene’s test]; however, variance was unequal between merged females and males [F(1, 22) = 269.3, P < 0.001; Levene’s test], while distribution shape between these two groups were equal (D = 0.5, P = 0.126; Kolmogorov–Smirnov test). For expression of Htr2b in the ventral hippocampus, we found equal variance between proestrus, diestrus, and male groups [F(2, 21) = 1.96, P = 0.166; Levene’s test]; we also found equal variance [F(1, 22) = 3.99, P = 0.058; Levene’s test] and equal distribution shape (D = 0.38, P = 0.424; Kolmogorov–Smirnov test) between merged females and males.

In summary, we found Ptprt expression to follow the same pattern that we see with the structural dendritic spine phenotype; we found more variability in females than in males and that the sex difference, detectable when females are merged, is actually driven by the estrous cycle stage. With Htr2b expression, we see the pattern that we observed with anxiety-related behavior; males and females show similar variability and sex difference can only be detected when there is information about the estrous cycle stage.

Analysis of the 3D genome interactions

Finally, we explore our previously published data derived from the unbiased chromosome conformation (Hi-C) assay (Fig. 5) [34]. This assay detects 3D genome interactions throughout the genome, and here we focus on CTCF loops (Fig. 1E), which allow long-range interactions between distant genomic regions, important for higher-order chromatin organization and gene regulation [38]. We explored these chromatin loops in sorted ventral hippocampal neurons and made the following comparisons: diestrus vs. male groups, as well as merged female (diestrus + proestrus) vs. male groups (Fig. 5). Importantly, we found an increased ability (1.65 times) to call sex-specific loops when comparing diestrus to males (260 differential loops), as opposed to comparing mixed females to males (158 differential loops; Fig. 5A) [34].

To illustrate this with an example, we present a loop involving Adcyap1 (Fig. 5B) [34], an important stress- and estrogen-sensitive gene implicated in anxiety-related behavior [39, 40]. This 2-Mb loop is stronger in proestrus and males than in diestrus (Fig. 5B), and this is associated with differential Adcyap1 expression among the three groups (Fig. 5C). Interestingly, this differential loop is also found in the mixed-female to male comparison (Fig. 5D) [34], further showing that the sex-specific dynamism that we observed, with proestrus becoming more similar to male Adcyap1 in terms of gene looping and gene expression, is only detectable if we monitor the estrous cycle stage.

Overall, this data indicates that accounting for the estrous cycle stage in females helps identify sex differences in chromatin looping of relevance to chromatin organization and gene expression.

Data variability across neurobehavioral measures

Finally, we decided to test our data variability across all neurobehavioral measures—behavior, hippocampal dendritic spine density, and gene expression—using the coefficient of variation (CV = standard deviation/mean), as a measure of relative variability, previously described in the meta-analyses performed by Prendergast et al. [22] and Becker et al. [21] (Fig. 6). When we calculated and compared the CV value for each group across the 6 datasets described here, we found no difference in variability between females and males whether females were separated by estrous cycle stage [F(2, 15) = 0.514, P = 0.608, Fig. 6A], or merged into one female group [t(9.94) = 1.87, P = 0.092; Fig. 6B].

In sum, our data are consistent with the data previously reported in mice and rats that females are, on average, not more variable than males [21, 22, 26]. However, our data also clearly show that this finding is not, at all, predictive of whether the estrous cycle plays an important role in regulating the outcome of interest.

留言 (0)

沒有登入
gif