Embryo implantation evolved from an ancestral inflammatory attachment reaction

Abstract

The molecular changes that support implantation in eutherian mammals are necessary to establish pregnancy. In marsupials, pregnancy is relatively short, and although a placenta does form, it is present for only a few days before parturition. However, morphological changes in the uterus of marsupials at term mimic those that occur during implantation in humans and mice. We investigated the molecular similarity between term pregnancy in the marsupials and implantation in eutherian mammals using the gray short-tailed opossum (Monodelphis domestica) as a model. Transcriptomic analysis shows that term pregnancy in the opossum is characterized by an inflammatory response consistent with implantation in humans and mice. This immune response is temporally correlated with the loss of the eggshell, and we used immunohistochemistry to report that this reaction occurs at the materno-fetal interface. We demonstrate that key markers of implantation, including Heparin binding EGF-like growth factor and Mucin 1, exhibit expression and localization profiles consistent with the pattern observed during implantation in eutherian mammals. Finally, we show that there are transcriptome-wide similarities between the opossum attachment reaction and implantation in rabbits and humans. Our data suggest that the implantation reaction that occurs in eutherians is derived from an attachment reaction in the ancestral therian mammal which, in the opossum, leads directly to parturition. Finally, we argue that the ability to shift from an inflammatory attachment reaction to a noninflammatory period of pregnancy was a key innovation in eutherian mammals that allowed an extended period of intimate placentation.

Keywords: evolution; inflammation; marsupial; placenta; pregnancy.

Fig. 1.

Relationship between major groups of mammals. Oviparity is still observed in the extant monotremes. Viviparity evolved in the stem therians and is present in almost all mammals today; however, marsupial pregnancy is short, and an extended pregnancy is observed only in eutherian mammals.

Fig. 2.

Histological examination of the fetal–maternal interface in the opossum through the second half of gestation. H&E-stained uterine sections from females 7 (A), 11.5 (B), 12.5 (C), and 13.5 (D) d post copulation. Parturition occurs 14.5 d post copulation. Before day 12.5 there is an intact shell coat (SH) that separates the embryonic trophoblast (T) from the uterine luminal epithelial cells (arrows). On day 13.5 of gestation large vesicles actively bud from the apical surface of the uterine lumen. (Scale bars: 100 μm.) Arrows point to uterine luminal epithelial cells. E, developing egg/embryo; G, uterine glandular tissue.

Fig. 3.

Correlations of transcriptome samples. (A) A heatmap showing the correlations between each pair of transcriptome samples. (B) Principal component analysis of uterine gene expression between samples at different stages of the reproductive cycle. LG, late gestation (postattachment, day 13.5 of pregnancy); MG, midgestation (preattachment, day 7 of pregnancy); NP, nonpregnant tissue. (C) Heatmap of clustering of differentially expressed genes in each uterine tissue sample. Gene-expression values are normalized by using a z-score transformation on TPM. (D) Number of differentially expressed genes in pairwise comparisons between each reproductive stage.

Fig. 4.

Semantic clustering of significantly overrepresented GO terms in differentially expressed genes in comparisons between nonpregnant (A) and midgestation (B) tissue and between preattachment (C) and postattachment (D) tissue. GO analysis was performed by GOrilla (77), and semantic clustering was achieved by ReviGO (78). In each gene set there is significant enrichment of GO terms.

Fig. 5.

Expression of key markers (as labeled in A–F) of implantation in M. domestica through key days of pregnancy, as measured by real-time qPCR. The time of shell-coat breach is marked by a red arrow, and parturition is marked by a star. Expression levels are reported relative to expression in nonpregnant uterine tissue and to the internal reference gene TBP. The change in mean value is indicated by a black line.

Fig. 6.

Immunolocalization of key implantation markers to the uterine tissue of M. domestica. Images present the localization of PTGS2, also known as COX2 (A–C), HBEGF (D–F), MUC1 (G–I), and PTGES (J–L). Uterine tissue was collected at different stages of the reproductive cycle from nonpregnant (A, D, G, J), preattachment (B, E, H, K), and postattachment (C, F, I, L) females. Localization was visualized with 3,3′-diaminobenzidine-tetrahydrochloride (DAB), which appears as a brown precipitate in sections. (Scale bars: 100 μm.) Arrowheads point to uterine luminal epithelial cells. G, uterine glandular tissue; T, trophoblast tissue.

Fig. 7.

Transcriptomic comparisons between the opossum attachment reaction and eutherian implantation. (A) Comparison between known inflammatory molecules produced during implantation in several eutherian lineages and their production during attachment in the opossum. PG, prostaglandin. References: sheep (–81), pig (, –86), horse (–89), dog (90, 91), mouse (–94). (B and C) The timing of implantation marker expression relative to the menstrual cycle in humans (B) and in opossum in days following copulation (C). (B) The periods of expression of key markers of implantation during the menstrual cycle are indicated by the horizontal bars (5). The window for successful implantation (orange bar) occurs when there is coexpression of each protein. (C) The expression of key implantation markers in different stages of the reproductive cycle in M. domestica. No expression (<3 TPM) is indicated by empty circles; solid circles indicate that expression was observed (>3 TPM); large circles indicate that expression was very high (>30 TPM). (D–G) Statistical tests of overlap between transcriptome-wide changes in gene expression during implantation in eutherians and during the opossum attachment reaction. (D) Overlap between genes enriched in the rabbit implantation site compared with the interimplantation site (41) and opossum attachment. (E–G) Overlap of genes more highly expressed in the human endometrium during the implantation window (–47) and during opossum attachment. P values were calculated using the χ² test.

Fig. 8.

Comparison between the key stages of pregnancy in the opossum and human. The eutherian reproductive condition has evolved by inserting an anti-inflammatory period of gestation between the inflammatory attachment reaction and the inflammatory parturition reaction. Dashed arrows represent the two hypotheses that could explain how secondary inflammation arose in eutherians: 1) eutherian pregnancy can be understood either through the insertion of an antiinflammatory phase of pregnancy into what was one inflammatory process in the ancestral therian; or 2) by the addition of both an antiinflammatory phase and an independently derived secondary inflammation at parturition.