The effects of fetal E2 exposure on prostate development do not follow a monotonic dose-response 
. Previous studies have shown that exposure of male mouse fetuses to a very small increase in serum E2 
, or to very low maternal doses of the estrogenic drugs DES and ethinylestradiol or the xenoestrogen BPA, lead to basal epithelial cell hyperplasia and to a permanent increase in prostate AR binding activity, resulting in an increase in prostate size in adulthood 
. Those findings showed that at low doses, estrogen has a stimulatory effect on the action of androgen in regulating prostate differentiation and subsequent prostate function, including development of early stage prostate cancer in adulthood 
. In contrast, opposite effects have been found at much higher doses of E2 and xenoestrogens. Prenatal or neonatal exposure of rats or mice to high doses of estrogens led to a decrease in prostate growth during the time of exposure in development, which led to reduced prostate size and androgen responsiveness in adulthood 
Non-monotonic dose responses were seen in our initial examination of the effects of estradiol and BPA on Ar and Esr1 expression in fetal mouse UGS mesenchyme 
, and dose-related variation in the pattern of gene expression was also observed for a large number of genes in human MCF-7 breast cancer cells, in response to doses of E2 between 10–100 pM 
. Because of these prior findings as well as different in vivo
effects of high and low doses of estrogen, we chose to examine the effects of E2 on gene expression in fetal prostate mesenchyme cells by microarray analysis, using two low doses (10 pM and 100 pM) as well as a high dose (100 nM) that had resulted in maximal Ar expression in our prior study with the same fetal mesenchyme cells 
. In laboratory rats and mice, the free serum concentration of E2 (unbound to plasma proteins and unconjugated) is about 2 pM during the initial period of prostate development 
although calculation of the actual biologically active fraction of E2 during sexual differentiation is complicated by uncertainty regarding the bioavailability of albumin-bound E2 and the capacity for the maternal-placental-fetal tissues to deconjugate sulfated estrogens 
. Total serum E2 during this period is in the range of 300 pM 
, and thus the low doses of E2 used in this study are physiologically relevant.
These microarray experiments were performed as a hypothesis generation step for a study of effects of estrogens on prostate development and differentiation, and the sample size is small. Because of this, the data must be seen as preliminary, but the results do indicate activation of different patterns of gene expression and dominance of different pathways at low, physiologically relevant, compared to high, pharmacological, doses of E2. Results from the lowest (10 pM and 100 pM) doses of E2 treatments indicate E2-inducible genes within pathways related to cell adhesion, actin cytoskeleton reorganization, EGF-like calcium binding, sterol biosynthesis and lipoprotein metabolism, and E2-suppressible genes within pathways related to growth factor signaling, tube development and additional effects on cell adhesion. At the high (100 nM) concentration, E2 induced genes enriched for steroid hormone signaling and metabolism, cytokines and their receptors, cell-to-cell communication, and TGF-β signaling (). Results from the 100 nM E2 treatment thus indicated effects on cell adhesion pathways, but also emphasized a stimulation of a positive feedback loop involving steroid hormone receptors and genes related to growth and metabolism that promote rather than inhibit cell growth. Taken together, these results suggest that fetal prostate mesenchymal cells may regulate epithelial cells through direct cell contacts when estrogen levels in mesenchyme are in the pM range, whereas growth factors might play significant roles when estrogen levels are higher in the nM range.
Importantly, an inverted U (non-monotonic) response was seen within the low-dose results, with enhancement of glycolysis observed at 10 pM E2 but significant suppression at 100 pM E2 (). The expression of these specific genes was not influenced by 100 nM E2, indicating that the stimulation of glycolysis is highly dependent on dose and only seen at low pM E2 concentrations. This is of particular interest given the Warburg effect, the observation that most cancer cells rely on glycolysis to generate the energy needed for cellular processes, in contrast to normal differentiated cells that use mitochondrial oxidative phosphorylation 
. The enhancement of glycolysis seen in our culture was only at the lowest dose tested here, 10 pM (2.72 pg/ml), and as such is intriguing because mice exposed prenatally to a very similar concentration of estradiol have enlarged prostates in adulthood 
relative to mice exposed to higher doses. It is interesting to speculate on whether there is a relationship between the enhancement of cell proliferation rate and glycolysis seen in cancer cells, and the enhancement of glycolysis in fetal prostate mesenchymal cells and increased prostate size due to hyperplasia seen in mice.
Only 29 genes out of those screened were influenced by all doses of estradiol examined (). For approximately half of these genes the dose-response relationship was monotonic, although some of these were maximally up- or down-regulated at the 100 pM dose. For the rest, the direction of the effect (stimulation or suppression of gene expression) was either strongly reversed at the highest (100 nM) E2 concentration (a non-monotonic response), or simply showed a suggestion of reversal at the highest dose. Of the monotonic profiles, two genes showed particularly strong linearity with dose: Angpt2 (angiopoetin 2) and Sprr1a (small proline-rich protein 1a). Angpt2 expression is strongly correlated with prostate cancer progression 
and is stimulated by growth factors, especially VEGF 
; Vegf expression is stimulated by androgen treatment in fetal prostate fibroblasts 
, but we did not observe an effect of estrogen on Vegf expression here. Expression of Sprr genes is typically restricted to cells committed to terminal differentiation 
. Although strong up-regulation of Sprr1a has been associated with abnormal cell differentiation in uterine tissue from neonatal CD-1 mice treated with diethylstilbestrol 
, effects in the developing prostate have not previously been reported.
Genes whose expression was significantly (P≤0.05) influenced by estradiol (E2) treatment at all doses tested.
Also of interest in this 29-gene subset are the clear inverse U effects on Perp and Gja1 expression. Perp is typically upregulated during apoptosis 
but is also important for promoting desmosomal cell-cell adhesion 
, and loss of Perp is associated with dysregulation of cell adhesion and promotion of tumor development and progression 
. Decreased expression of Gja1 (Cx43) is similarly consistent with loss or reduction of cell-cell communication. Only one gene in this 29-gene subset, Enpp2, showed a U-shaped response to increasing E2 concentrations; Enpp2 codes for autoaxin, an ecto-enzyme responsible for producing lysophophatidic acid (LPA), known to be a mitogen for both ovarian and prostate cancer cells, which stimulates cell proliferation, survival and migration (reviewed in 
). These non-monotonically expressed genes reinforce the general conclusion that pathways related to cell adhesion are influenced by estrogen treatment, but also suggest a different effect of the highest dose relative to the lower doses, with a progression toward increased cell proliferation and migration at increasing dose.
The Wnt signaling pathway was influenced at all E2 doses examined, but with an emphasis toward up-regulation of canonical Wnt/β-catenin stabilization signaling at the high dose, and non-canonical (PCP) signaling at lower doses. The high-dose effect may be mediated through the known association of β-catenin with AR and ER. Truica et al.
have shown that β-catenin significantly enhances androgen-stimulated transcriptional activation by the AR, and that β-catenin also increases AR transcriptional activation by E2 
. Although many Wnt genes are differentially expressed in the prostate according to age 
, their role in prostate development, and particularly their interactive and temporal roles, is only starting to be described.
At the high dose of E2 we observed changes in genes related to steroid hormone metabolism, and alterations in steroid hormone signaling that would lead in turn to disruption of the normal expression of other developmentally important genes. Of particular interest was the observed up-regulation of Cyp7b1, which catalyzes the metabolism of the DHT metabolites 3α-Adiol and 3β-Adiol, and is thought to control cellular levels of both androgens and estrogens 
. We verified by quantitative PCR (qPCR) that the up-regulation of Esr1 observed in these estrogen-treated cells was dose-dependent and consistent with our prior data (
, data not shown); up-regulation of Ar was seen by qPCR but did not reach statistical significance by microarray. Esr1 was stimulated across the entire E2 dose range in this study, and thus is a potential common mechanism for the initiation of consequent signaling events. Stimulation of Esr1 and Ar serve to amplify estrogen and androgen signaling respectively, and in the intact gland there would be further potential for signal amplification, with local conversion of testosterone not only via Srd5a1 to the more potent androgen DHT, but also via aromatization to E2.
It is important to note that the intracellular concentration of E2 within the urogenital sinus during development is still unknown. The dose of E2 that reaches ER in male mouse UGS mesenchyme cells would depend not only on E2 uptake from the blood but also on local aromatization of testosterone to E2. Because of this issue, we administered E2 over a wide dose range, but also ensured that the opportunity for aromatization was controlled by the use of DHT rather than testosterone in the culture medium. Total testosterone circulates in the range of 5–8 nM in the male rat and mouse fetus during prostate differentiation 
. Because there is no high-affinity testosterone binding protein in the blood at this time, and testosterone is only weakly bound to albumin, the result is that the percentage of total testosterone in blood that is bioactive is high, particularly compared to E2, which binds to the high-affinity plasma protein alphafetoprotein. Serum testosterone thus provides a substantial pool from which intracellular E2 can be formed by aromatization in fetal prostate mesenchyme cells 
. Arase and colleagues 
have measured E2 concentrations in fetal male mouse UGS tissue at GD17 and postnatal day (PD) 1, which approximated 10 and 25 pg/g, respectively. These concentrations are consistent with the low doses of E2 that we administered in this study, although again we do not know how much of this E2 reaches ER (the actual dose at target). Future work should address the dynamics of estrogen concentration and receptor activation both in vitro
and in vivo
The up-regulation of Pgr by all doses of E2 administered here to UGS mesenchyme cells is in general agreement with Risbridger et al., who reported up-regulation of progesterone receptors (PR) in the adult mouse prostate after estrogen treatment 
, and with data from Nishino et al. that showed enhancement of progesterone's proliferative effects on the adult rat prostate after co-treatment with E2 or DHT 
. The presence of PR may be more relevant during fetal life, when progesterone levels are higher, than in adulthood when progesterone levels are low. The issue of fetal responsiveness to progestins is complex in that there is evidence that progestins can have anti-androgenic influences on sexual differentiation, through inhibition of 5α-reductase 
. Up-regulation of Pgr is thus a potential mechanism for disruptive effects of estrogens on male accessory reproductive organ development, but its impact will require further study.
Neonatal estrogen treatment is known to affect the expression of several genes critical to prostate development. Notable examples are Hoxb13, Nkx3.1, Shh, Fgf10 and Bmp4 
. Some of the genes that responded to E2 treatment in our cells agree with the findings of others (Hoxb13, Bmp4), but several of the “candidate” genes were not affected at the doses we examined. There may be several reasons for this, but two are critical. First, we deliberately cultured only the mesenchyme cells, to specifically examine effects of E2 on gene expression in the cells that initiate early prostate differentiation. Without the two-way communication that occurs between epithelial and mesenchymal cells in the developing prostate the full range of gene expression will not be seen 
. For example, Nkx3.1 is expressed only in epithelial cells in regions of ductal growth, although its expression is dependent on the presence of UGS mesenchyme 
. Similarly, Ptc and Gli, components of the Shh signaling pathway that are important for directing ductal growth, are expressed in the mesenchyme but are regulated by Shh signaling from the epithelium 
. Additionally, in studies performed in vivo
, other factors provided via blood circulation (known or unknown), as well as shifts in hormone levels that occur during late fetal life, parturition and early postnatal life 
, will influence gene expression. Consequently, studies performed in whole tissues of intact animals are bound to yield different and more complex results.
Developmental estrogen exposure has the potential to acutely stimulate abnormal growth and induction of hyperplasia in the developing prostate 
, and this clearly establishes the potential for abnormal function in later life and a predisposition toward adult prostate disease 
. The growth of fetal prostate epithelial cells and duct formation are driven by signals from the UGS mesenchyme 
, and our results suggest that the developmental effects of estrogens or xenoestrogens on UGS differentiation may be mediated initially by enhanced mesenchymal cell responsiveness to sex steroid hormones through up-regulation of steroid hormone receptor concentrations, with subsequent effects on other genes that differed based on the dose of E2. The differing patterns of gene expression at low and high E2 concentrations and the presence of non-monotonic responses of some genes to the wide (10,000-fold) range of E2 concentrations studied are consistent with non-monotonic dose effects on prostate development in vivo