Transcriptome analysis of the hormone-sensing cells in mammary epithelial reveals dynamic changes in early pregnancy

Abstract

Background: Alveoli, the milk-producing units of the mammary gland, are generated during pregnancy by collaboration of different epithelial cell types. We present the first analysis of transcriptional changes within the hormone sensing population during pregnancy. Hormone-receptor positive (HR+) cells play a key role in the initiation of alveologenesis as they sense systemic hormonal changes and translate these into local instructions for neighboring HR- cells. We recently showed that IGF2 is produced specifically by HR+ cells in early pregnancy, but is undetectable in the virgin state. Here, we define the transcriptome of HR+ cells in early pregnancy with the aim to elucidate additional changes that are unique for this dynamic developmental time window.

Results: We harvested mammary glands from virgin, 3-day and 7-day pregnant mice and isolated a few hundred hormone-sensing cells per animal by FACS for microarray analysis. There was a high concordance between animals with a clear induction of cell cycle progression genes at day 3 of pregnancy and molecules involved in paracrine signalling at day 7.

Conclusions: These findings underscore the proliferative capacity of HR+ cells upon specific stimuli and elucidate developmentally-restricted changes in cellular communication. Since the majority of breast cancers are HR+, with a variable proportion of HR+ cells per tumor, we anticipate that this data set will aid further studies into the regulation of HR+ cell proliferation and the role of heterotypic signalling within tumors.

Figure 1

Pregnancy induces proliferation in both the HR+ and HR- cells. (A) Representative images of carmine-alum-stained whole mounts of mammary glands from virgin, 3-day and 7-day pregnant FVB/N mice. Scale bar, 1 mm. (B) Confocal immunofluorescence of mammary gland sections from virgin, 3-day and 7-day pregnant mice stained for the proliferation marker 5-ethynyl-2’-deoxyuridine (EdU, green), Estrogen Receptor (ER, red) and the luminal cell marker cytokeratin-8 (CK8, blue). Scale bar, 10 μm. The exposure time for the ER signal was increased in pregnant samples to allow robust identification of ER+ cells (see Additional file 3 for images with a fixed exposure time). (C) Quantification of the proportion of luminal cells that is proliferating, separated by cell type based on ER expression. ER+ and ER- luminal cells start proliferating at day 3 of pregnancy, by day 7 the majority of proliferating luminal cells are ER-. Bars represent individual animals.

Figure 2

Purification of HR+ cells by FACS in early pregnancy. (A) The luminal population of mammary epithelial cells (CD24^hiCD49f^lo) was separated into hormone-sensing cells (Sca1^hi CD49b^lo; purple gate) and alveolar progenitor cells (Sca1^lo CD49b^hi; orange gate). (B) Quantitative RT-PCR on 500 directly-lysed cells per sample for markers that distinguish the two luminal cell types. Values are relative to mRNA levels in HR+ cells from virgin animals and normalised to HPRT expression. Error bars indicate standard deviation for 3 individual mice.

Figure 3

Transcriptome analysis clusters HR+ cells by developmental state. (A) Principal Component Analysis (PCA) was performed on three biological replicates of HR+ cells derived from virgin (Vir), 3-day (3D) and 7-day (7D) pregnant mice based on 19,821 probes (see Methods for details). A scatterplot of the first two principal components demonstrates a separation by pregnancy along the first principal component (PC1) and by duration of pregnancy along the second principal component (PC2). (B) Unsupervised hierarchical clustering based on 1328 probes (probes with an absolute fold-change > =2 and false discovery rate (FDR) of 10% or less) shows that the biological replicates cluster according to developmental state. (C) Venn diagram illustrating the number of genes that were changed more than 2-fold with a FDR of 10 in the 3 replicates of HR+ cells isolated at day 3 of pregnancy (3D, light grey) compared to virgin samples and the overlap of this gene set compared to differentially expressed genes at day 7 of pregnancy (7D, dark grey, more than 2 fold change compared to virgin).

Figure 4

Heat maps of transcriptional changes in HR+ during early pregnancy. (A) Heat map of Illumina gene probes generated by GenePattern software showing transcriptional changes during early pregnancy. (B-C) Heat maps generated with an input of the 25 most up-regulated and 25 most down-regulated genes in each category based on the values of their Pi scores (see Methods). Twenty-five gene probe sets that are induced (B) and reduced (C) most strongly in HR+ cells specifically at day 3 of pregnancy. (D-E) Gene probe sets that are most strongly induced (D) and reduced (E) in early pregnancy. (F-G) Gene probe sets that are induced (F) and reduced (G) most strongly in HR+ cells specifically at day 7 of pregnancy.

Figure 5

Pathway analysis of changes in HR+ cells during early pregnancy. (A) Bar chart of the transcriptional changes of genes involved in IGF signaling. Values are normalized Log2 values of Illumina probe sets, error bars denote standard deviation of three biological replicates. The dotted red line indicates background noise. (B) Ingenuity Pathway Analysis showing canonical pathways that are most significantly changed at day 3 of pregnancy (blue) and at day 7 of pregnancy (green) compared to virgin samples. The number of genes in each pathway that was significantly changed in the HR+ samples is indicated at the top of each bar.

Figure 6

Top molecules that are differentially expressed in HR+ cells in early pregnancy. (A) Molecules identified by Ingenuity Pathway Analysis that are most strongly down- and upregulated in HR+ cells at day 3 of pregnancy compared to virgin samples. (B) Molecules identified by Ingenuity Pathway Analysis that are most strongly down- and upregulated in HR+ cells at day 7 of pregnancy compared to virgin samples.

Figure 7

Quantitative polymerase chain reaction (qPCR) validation of the transcriptional changes in HR+ in early pregnancy. (A) Relative mRNA expression of cell cycle genes during pregnancy compared to virgin samples, normalised to HPRT expression. (B) Relative mRNA expression of progesterone target genes during pregnancy compared to virgin samples, normalised to HPRT expression. (C) Relative mRNA expression of secreted factors during pregnancy compared to virgin samples, normalised to HPRT expression. N.D.: Not Detected. Error bars indicate standard deviation for 3 individual mice.