Comparative temporal and dose-dependent morphological and transcriptional uterine effects elicited by tamoxifen and ethynylestradiol in immature, ovariectomized mice

Abstract

Background: Uterine temporal and dose-dependent histopathologic, morphometric and gene expression responses to the selective estrogen receptor modulator tamoxifen (TAM) were comprehensively examined to further elucidate its estrogen receptor-mediated effects. These results were systematically compared to the effects elicited by the potent estrogen receptor ligand 17alpha-ethynylestradiol (EE) to identify pathways similarly and uniquely modified by each compound.

Results: Three daily doses of 100 microg/kg TAM elicited a dose-dependent increase in uterine wet weight (UWW) in immature, ovariectomized C57BL/6 mice at 72 hrs with concurrent increases in luminal epithelial cell height (LECH), luminal circumference and glandular epithelial tubule number. Significant UWW and LECH increases were detected at 24 hrs after a single dose of 100 microg/kg TAM. cDNA microarray analysis identified 2235 differentially expressed genes following a single dose of 100 microg/kg TAM at 2, 4, 8, 12, 18 and 24 hrs, and at 72 hrs after three daily doses (3 x 24 hrs). Functional annotation of differentially expressed genes was associated with cell growth and proliferation, cytoskeletal organization, extracellular matrix modification, nucleotide synthesis, DNA replication, protein synthesis and turnover, lipid metabolism, glycolysis and immunological responses as is expected from the uterotrophic response. Comparative analysis of TAM and EE treatments identified 1209 common, differentially expressed genes, the majority of which exhibited similar profiles despite a temporal delay in TAM elicited responses. However, several conserved and treatment specific responses were identified that are consistent with proliferation (Fos, Cdkn1a, Anapc1), and water imbibition (Slc30a3, Slc30a5) responses elicited by EE.

Conclusion: Overall, TAM and EE share similar gene expression profiles. However, TAM responses exhibit lower efficacy, while responses unique to EE are consistent with the physiological differences elicited between compounds.

Figure 1

Tamoxifen-induced dose dependent and temporal changes in uterine weight. Graphs illustrate fold-change increases in uterine wet (open) and blotted (solid) weight. A) Tamoxifen elicits a dose dependent uterotrophic response (EC₅₀ = 33.7 μg/kg) and achieves maximal induction of approximately 5-fold following three daily doses (3 × 24 hrs) of 100 μg/kg TAM. Significant increases (p < 0.05, n = 5) are denoted by an asterisk (*). In contrast, 100 μg/kg EE (positive control) maximally induced uterine wet weight 11-fold (*, p < 0.05, n = 5) with significant water imbibition (#; p < 0.05, n = 3), while TAM only achieved 50% uterotrophic efficacy and no water imbibition. B) A single dose of 100 μg/kg TAM significantly increased uterine wet weight as early as 24 hrs after administration. No significant water imbibition was observed at any time point.

Figure 2

Uterine histology. Hematoxylin and eosin stained sections of uterine tissue at 100× magnification after three daily doses of A) sesame oil, B) 1 mg/kg TAM and C) 100 μg/kg EE. TAM and EE treatment induced increases in luminal epithelial cell height. Luminal circumference is increased to a greater degree by EE than TAM. Bars represent 20 μm.

Figure 3

Tamoxifen-induced temporal gene expression patterns. Five k-means clusters best represent the general temporal patterns for the 2941 features differentially expressed following TAM treatment. Note the 8 hr delay in gene expression response especially in comparison to EE elicited gene expression [16] is speculated to be due to the delayed absorption of TAM. Inset numbers indicate the number of features represented by each cluster. Black pseudolines indicate the general profile represented within each cluster.

Figure 4

Quantitative real-time PCR verification of selected TAM-induced genes. Overall, the microarray results for 14 TAM- and EE-induced genes were verified using QRT-PCR. The verified genes represent various affected pathways and different temporal patterns of expression. Overall, there was good correlation (average ρ = 0.8) between microarray (lines) and QRT-PCR (bars) data. Examples for six of the genes are illustrated. Statistically significant QRT-PCR differences (p < 0.05, n = 4) due to treatment are denoted by an asterisk (*).

Figure 5

Immunohistochemical detection of differential Pcna protein levels due to TAM. Twelve-hour vehicle (A) and TAM (B) treated uteri sections were immunohistochemically stained (NovaRED^®) with Pcna specific antibodies. Treated samples have darker nuclear staining, indicating greater levels of Pcna protein expression, in agreement with the histological assessment and changes in gene expression associated with cell proliferation. Increased Pcna expression is more pronounced in the luminal and glandular epithelium, and stroma (arrows). Tissues were counter-stained with hematoxylin. Images are representative of four biological replicates. Bars represent 20 μm.

Figure 6

Temporal comparison of genes commonly activated by TAM and EE. Hierarchical clustering of 1209 TAM- and EE-regulated genes (y-axis) identifies subsets of similar profiles according to time and treatment (x-axis). The dendrogram indicates that early responses (4 hrs) to ethynylestradiol (E) are most similar to 8 and 12 hrs tamoxifen (T) responses demonstrating temporally displaced TAM activation consistent with the delayed absorption of TAM. However, temporal displacement of TAM elicited responses is not maintained as EE and TAM responses cluster together at 24 and 72 hrs.

Figure 7

Examples of TAM and EE differential gene expression classifications. Examples of representative genes classified as Similar or Efficacious based on microarray data only. QRT-PCR analysis confirmed the classifications of these genes. In some cases (e.g., Cdkn1a) a gene classified as Similar may also be classified as EE-Efficacious based on QRT-PCR results due to data compression inherent in microarray data. Statistically significant differences (p < 0.05, n = 4) due to treatment are denoted by an asterisk (*).

Figure 8

Identification of unique EE and TAM differentially expressed genes. Treatment specific differentially expressed genes were identified by excluding a list obtained using a more relaxed criteria (P1(t) > 0.9; |fold change| ≥ 1.4) for one treatment from the differentially expressed genes identified using the standard criteria (P1(t) > 0.999; |fold change| ≥ 1.5) of the second treatment to identify gene expression changes that were more likely to be unique to one treatment. (A) A liberal list of TAM-induced genes identified, using a relaxed criteria of P1(t) ≥ 0.9 and |fold change| ≥ ± 1.4, was excluded from the EE differentially expressed gene list using the standard selection criteria of P1(t) ≥ 0.999 and |fold change| ≥ ± 1.5 to identify 240 genes more likely to be differentially expressed by EE alone. (B) Using a similar approach, a list of 60 genes more likely to be differentially expressed by TAM alone was generated. Lists of EE and TAM specific genes are provided in Additional files 4 and 5.

Figure 9

Temporal expression profiles of TAM and EE-specific genes. Graphical representation of genes exhibiting compound-specific responses demonstrated profiles which were distinctly different in pattern and magnitude compared to its non-responsive counterpart. These examples further illustrate that the filtering conditions used were adequate to identify differential responses by TAM and EE.