Hedgehog signaling is required for endometrial remodeling and myometrial homeostasis in the cycling mouse uterus

Abstract

Decades of work demonstrate that the mammalian estrous cycle is controlled by cycling steroid hormones. However, the signaling mechanisms that act downstream, linking hormonal action to the physical remodeling of the cycling uterus, remain unclear. To address this issue, we analyzed gene expression at all stages of the mouse estrous cycle. Strikingly, we found that several genetic programs well-known to control tissue morphogenesis in developing embryos displayed cyclical patterns of expression. We find that most of the genetic architectures of Hedgehog signaling (ligands, receptors, effectors, and transcription factors) are transcribed cyclically in the uterus, and that conditional disruption of the Hedgehog receptor smoothened not only elicits a failure of normal cyclical thickening of the endometrial lining but also induces aberrant deformation of the uterine smooth muscle. Together, our data shed light on the mechanisms underlying normal uterine remodeling specifically and cyclical gene expression generally.

Keywords: Biological sciences; Cell biology; Developmental biology.

Graphical abstract

Figure 1

Cyclical mouse uterine remodeling occurs concomitant with transcriptional changes (A‒J) H&E staining of mouse uterine sections at (A) proestrus, (B) estrus, (C) metestrus, and (D) diestrus where the dotted white line denotes boundary between the endometrium and myometrium. Scale bar = 250μm. Quantitation of (E) uterine area, (F) lumen area, (G) endometrial area, (H) myometrial area, (I) uterine gland number, and (J) luminal epithelial height during those same time points. Each point on the graphs represents one uterus and is the mean quantification of 4 or more sections; all error bars are standard deviation of the mean. (K‒O) (K) Principal-component analysis (PCA) plot of 3′ TagSeq dataset. Volcano plots showing differential gene expression between (L) diestrus and proestrus, (M) proestrus and estrus, (N) estrus and metestrus, and (O) metestrus and diestrus. (P) Heatmap of the most variable differentially expressed genes at each estrous cycle stage, where each column represents one sample. All heatmaps use z-scores of the DESeq2 variance stabilized counts. t tests were performed, where ∗ = p < 0.05; n.s. = not significant.

Figure 2

Development genes are cyclically expressed in the uterus during the estrous cycle (A‒F) GO terms of differentially expressed genes enriched at (A) proestrus, (B) estrus, and (C) metestrus. Heatmaps of (D) Hox genes, (E) Wnt signaling, and (F) TGF-β signaling. See also Table S2. Each column represents one sample, and all heatmaps use z-scores of the DESeq2 variance stabilized counts.

Figure 3

The Hedgehog signaling pathway is cyclically regulated across the estrous cycle (A) The biochemical pathway of Hh signaling. (B) Heatmap of the core components of Hh signaling across the estrous cycle, where the Z score of the DESeq2 variance stabilized counts is used. Each column represents one sample. See also Table S2. (C‒M) qPCR of the same core components of Hh signaling with n = 8 samples. See also Table S1. t tests were performed, where ∗ = p < 0.05; n.s. = not significant.

Figure 4

Hedgehog signaling is required for cyclical uterine remodeling (A) Experimental flow to determine role of Hh signaling in the adult uterus during the estrous cycle. (B) qPCR for Smoothened in the controls and Smo cKOs. (C) qPCR for Gli1 in the controls and Smo cKOs. (D) Breeding trial of controls and Smo cKOs. (E) Estrous cyclicity of controls and Smo cKOs. (F‒I) H&E-stained uterine sections from control and Smo cKOs. (J‒L) Quantitation of (J) total uterine area, (K) endometrial area, and (L) myometrial-to-endometrial area ratio. Each point on the graphs represents one uterus and is the mean quantification of 4 or more sections. t tests were performed, where ∗ = p < 0.05; n.s. = not significant. Scale in image = 250μm. All error bars are standard deviation of the mean.

Figure 5

The myometrium is significantly folded in the absence of Smoothened (A‒E) (A) Quantitation of longitudinal muscle thickness. Each point on the graph represents one uterus and is the mean quantification of 4 or more sections. (B–E) Masson Trichrome histology of the mouse uterus. (B′–E′) Outline of sections highlighting the ruffled nature of the Smo cKO. (B″–E″) Best fit ellipse of the outline. (F) The smoothness of each sample was determined by taking the ratio of the true outline to the best fit ellipse. Each point on the graph represents one uterus and is the mean quantification of 4 or more sections. t tests were performed, where ∗ = p < 0.05; n.s. = not significant. Scale bar = 250μm. All error bars are standard deviation of the mean.

Figure 6

Myometrial fibers are significantly bigger in the absence of Smoothened (A‒D) Masks of the longitudinal muscle fibers were produced using Ilastik. (A′–D′) Masson Trichrome histology was used to identify the myometrial muscle fibers. (A″–D″) Longitudinal fibers were identified throughout the entire section. (E‒G) (E) Quantitation of the fiber number in each sample. Each point on the graph represents one uterus and is the mean quantification of 4 or more sections. (F) Quantitation of the fiber area in each sample. Each point on the graph represents one uterus and is the mean quantification of 4 or more sections. (G) Fiber areas were binned in 1000μm² increments. t tests were performed, where ∗ = p < 0.05 and n.s. = not significant. Scale in main image = 50μm. Scale in ‘and “images = 250μm. All error bars are standard deviation of the mean.