Developmental exposure to diethylstilbestrol alters uterine gene expression that may be associated with uterine neoplasia later in life

Abstract

Previously, we described a mouse model where the well-known reproductive carcinogen with estrogenic activity, diethylstilbestrol (DES), caused uterine adenocarcinoma following neonatal treatment. Tumor incidence was dose-dependent reaching >90% by 18 mo following neonatal treatment with 1000 microg/kg/d of DES. These tumors followed the initiation/promotion model of hormonal carcinogenesis with developmental exposure as initiator, and exposure to ovarian hormones at puberty as the promoter. To identify molecular pathways involved in DES-initiation events, uterine gene expression profiles were examined in prepubertal mice exposed to DES (1, 10, or 1000 microg/kg/d) on days 1-5 and compared to controls. Of more than 20 000 transcripts, approximately 3% were differentially expressed in at least one DES treatment group compared to controls; some transcripts demonstrated dose-responsiveness. Assessment of gene ontology annotation revealed alterations in genes associated with cell growth, differentiation, and adhesion. When expression profiles were compared to published studies of uteri from 5-d-old DES-treated mice, or adult mice treated with 17beta estradiol, similarities were seen suggesting persistent differential expression of estrogen responsive genes following developmental DES exposure. Moreover, several altered genes were identified in human uterine adenocarcinomas. Four altered genes [lactotransferrin (Ltf), transforming growth factor beta inducible (Tgfb1), cyclin D1 (Ccnd1), and secreted frizzled-related protein 4 (Sfrp4)], selected for real-time RT-PCR analysis, correlated well with the directionality of the microarray data. These data suggested altered gene expression profiles observed 2 wk after treatment ceased, were established at the time of developmental exposure and maybe related to the initiation events resulting in carcinogenesis.

Figure 1

Global differential gene expression cluster analysis. Panel A: Hierarchical agglomerative clustering (cosine correlation similarity measure) was performed on genes exhibiting a log (ratio) P < 0.001 and fold change of 1.5 between DES-treated uteri and vehicle control at any DES dose. Two hundred thirty-two genes met this criteria and when clustered demonstrate a dose-related dependency on gene expression. The DES-1000 dose exhibits the strongest gene expression differences compared to control. Red indicates higher and green indicates lower transcript levels relative to control, log (ratio) = 0.5 (threefold) shows color saturation. Panel B: Data are plotted as a Venn diagram.

Figure 2

Enrichment for genes involved in cell adhesion, development and cell differentiation. High throughput GOMiner was used to determine common biological themes across at least two of the three DES doses. Significant changes were seen in genes involved in cell adhesion (Panel A); genes involved in growth and development (Panel B); and genes involved in cell differentiation (Panel C).

Figure 3

Comparison of permanently altered DES genes with estrogen regulated genes. Panel A: Genes in common between 24-h estradiol treatment (E, 24 h) and any one dose of DES (P < 0.001) were selected and hierarchical agglomerative cluster (cosine correlation similarity measure) was performed. Red indicates increased and green indicates decreased expression levels relative to control, log (ratio) = 0.5 (threefold) shows color saturation. With gene ontology analysis, the yellow box includes genes involved in biological processes associated with cell cycle, specifically M phase and mitotic spindle formation associations. Panel B: Data are plotted as a Venn diagram.

Figure 4

Comparison of permanently altered DES genes with genes expressed in human uterine adenocarcinoma. Genes found significant in human uterine adenocarcinoma were compared to the three DES doses. Twenty-six genes were found in common. Red indicates increased and green indicates decreased expression levels relative to control, log (ratio) = 0.5 (threefold) shows color saturation.

Figure 5

Verification of microarray analysis with real-time RT-PCR. Scatter plots of Ltf (Panel A), Sfrp4 (Panel B), Tgfb1 (Panel C), and Ccnd1 (Panel D) expression relative to the housekeeping gene18S as determined by RT-PCR. From the left, the doses were control, DES-1, DES-10, and DES-1000 μg/kg/d. Each point represents an individual mouse.