Distinct Function of Estrogen Receptor α in Smooth Muscle and Fibroblast Cells in Prostate Development

Abstract

Estrogen signaling, through estrogen receptor (ER)α, has been shown to cause hypertrophy in the prostate. Our recent report has shown that epithelial ERα knockout (KO) will not affect the normal prostate development or homeostasis. However, it remains unclear whether ERα in different types of stromal cells has distinct roles in prostate development. This study proposed to elucidate how KO of ERα in the stromal smooth muscle or fibroblast cells may interrupt cross talk between prostate stromal and epithelial cells. Smooth muscle ERαKO (smERαKO) mice showed decreased glandular infolding with the proximal area exhibiting a significant decrease. Fibroblast ERαKO mouse prostates did not exhibit this phenotype but showed a decrease in the number of ductal tips. Additionally, the amount of collagen observed in the basement membrane was reduced in smERαKO prostates. Interestingly, these phenotypes were found to be mutually exclusive among smERαKO or fibroblast ERαKO mice. Compound KO of ERα in both fibroblast and smooth muscle showed combined phenotypes from each of the single KO. Further mechanistic studies showed that IGF-I and epidermal growth factor were down-regulated in prostate smooth muscle PS-1 cells lacking ERα. Together, our results indicate the distinct functions of fibroblast vs. smERα in prostate development.

Estrogen actions in the prostate can be mediated via two distinct receptors, estrogen receptor (ER)α or ERβ (1). These two receptors, although displaying similar structural homology and ligand binding specificity, control two distinct pathways in the prostate. ERα appears to be involved in cellular proliferation signaling, whereas ERβ is implicated in antigrowth, apoptotic, signaling (2). In addition to having different functions, these two receptors are localized to different prostate cell types. In mice, ERα is found in the mesenchymal cells, whereas ERβ is found predominantly in the epithelial cells (3–5). A growing body of evidence indicates that estrogen signaling is important in prostate ductal morphogenesis (6), extracellular matrix composition (7) and disorders, such as prostate cancer and benign prostatic hyperplasia (BPH) (8–12). Pregnant female mice treated with low dose, but above physiological, levels of estrogen gave birth to male offspring that displayed hyperplastic phenotypes at adulthood (13). This indicates that one or both of the ERs are critical for prostate development. Prins et al. (5) showed that the hyperplastic phenotypes described earlier were not observed in male ERα knockout (KO) mice from estrogenized mothers. In ERβKO mice, hyperplasia was still seen, confirming that ERα, but not ERβ, was the primary receptor involved in aberrant cellular growth. Interestingly, because hyperplasia was seen in the epithelial cell population, but ERα is expressed in the stromal compartment and a subset of basal cells, cross talk between the stromal and epithelial compartments occurs, and indeed, this phenomenon has been shown on numerous occasions (14–17). Disruption of this cross talk has been tested in tissue recombination models but has not been truly validated in an in vivo model that recapitulates the intact prostate microenvironment.

Two previous studies using total ERαKO mice have shown a significant 1.7- to 2.0-fold increase in testosterone levels at adulthood (6, 18). After castration, the prostate shrinks in size (19). This diminishment in prostate size is reversible after the addition of testosterone or dihydrotestosterone (20, 21). Combined, these data exhibit the potent mitogenic effects of testosterone in the prostate. Although testosterone levels were increased, our group was still able to observe defects in prostate development in ERαKO male mice, indicating a direct ERα role in prostate development.

Previous work in our lab has looked at the effect of total ERαKO using the ACTB promoter (6). These mice display distinct developmental defects that cannot be reproduced in epithelial ERαKO mice; thus, stromal ERα and not epithelial ERα must be responsible for the observed developmental phenotypes. Using fibroblast ERαKO (fERαKO) mice, we found that many of the reported phenotypes could be accounted for (22). However, estrogen and ERα have been reported to play varying roles in the smooth muscle of uterus (23), endothelium (24–26), and prostate (6, 27, 28) and to have a particular role in collagen deposition and formation in the prostate (7), yet little is known about the role of smooth muscle ERα (smERα) in prostate development. In the current study, we first looked at the effect of smERαKO on the ventral prostate (VP) development. We used Tgln cre (29), which was shown to have high efficiency in the VP (30), to create smERαKO mice. We compared this phenotype with the previously published fERαKO mice phenotype (22) as well as to mice with ERαKO in both fibroblasts and smooth muscle cells (double-cre ER α KO, dERαKO). We found that these dERαKO mice display an aggregate phenotype of both fibroblast and smooth muscle KO mice.

Previous studies testing the importance of ERα in the prostate development or pathogenesis used tissue recombination in a semi in vivo fashion. In these models, ERα is manipulated in the mesenchyme and then incubated with epithelial cells to see whether prostrate regrowth is possible (31). However, this system uses stromal cells from seminal vesicles, not from prostate, and the coinoculation of these stromal and epithelial cells cannot result in the formation of the normal, full structure of the prostate glands. In addition, these models do not take into account the full prostate microenvironment, which is composed of two main compartments, the epithelial and stromal. The epithelial compartment contains luminal, secretory, and basal cells, whereas the stroma is comprised of fibroblasts, smooth muscle cells, and, to a lesser extent, endothelial cells, and infiltrating immune cells (32–35). Tissue recombination distills this complex microenvironment system down to a system involving fewer cell types.

In our model, we demonstrate that different cell types invoke different ERα actions, providing a more powerful model to mimic the in vivo prostate microenvironment with which to further study the ERα role in prostate organogenesis, homeostasis, or pathogenesis.

Materials and Methods

Animals

Animals were housed in the S wing vivarium at the University of Rochester, School of Medicine. All protocols related to animals were overseen and approved by the animal care and use committee, and all animals were treated in accordance with National Institute of Health guidelines. The Tgln mice were created on an FVB background (29) and backcrossed more than 8 generations to a C57BL/6 background. FSP, and floxed ERα, mice were created on a C57BL/6 background. The genotypes of above mice were identified using PCR on DNA obtained from tails lysed in buffer (QIAGEN, Valencia, CA) with 0.5 mg/ml proteinase K (Invitrogen, Carlsbad, CA) overnight.

Mouse prostate dissection and histology analysis

Mice were euthanized, serum obtained via cardiac puncture, and the whole urogenital tract excised into PBS. Further dissection of specific lobes was achieved under an L2 illumination microscope (Leica, Heerbrugg, Switzerland). The ventral, anterior, and dorsal-lateral prostates were removed to 4% paraformaldehyde for 4–6 h depending on size. After fixation, lobes were processed through a gradient of alcohols to xylene and then embedded in paraffin. Sections of 5-μm thickness were affixed to slides and stained with hematoxylin and eosin. Slides were viewed using an eclipse E800 microscope (Nikon, Melville, NY), and images were taken using Spot Advanced camera software (Diagnostic Instruments, Inc., Sterling Heights, MI). ImageJ (National Institute of Health, Bethesda, MD) was used for image analysis.

Immunohistochemistry and immunofluorescence

Slides were dewaxed, rehydrated, and subjected to antigen retrieval in 10 mm sodium citrate buffer (pH 6.0) for 20 min at 96 C in a histowave oven (Thermo Shandon, Pittsburgh, PA). After antigen retrieval, slides were incubated in 3% H₂O₂ in methanol for 30 min to quench endogenous peroxidase, blocked in a 5% fetal bovine serum (FBS), 5% bovine serum albumin, and 5% nonfat milk solution for 1 h, and the primary antibody applied overnight. Antibodies used were, mouse ERα (MC-20, 1:400) and Ki67 (NCL-Ki-67p, 1:1000). The biotinylated secondary antibody (Vector Laboratories, Burlingame, CA) was applied for 1 h, and then slides were incubated with an Avidin-Biotin Complex (Vector Laboratories). Development was achieved through use of 3′-3′-diaminobenzidene (Vectastain), and then Mayer's hematoxylin was used to counterstain the slides. For immunofluorescence, after the first antibody, a fluorescein isothiocyanate-labeled secondary antibody was applied (Invitrogen) to tissue. Tissue slides were then counterstained with 4′,6-diamidino-2-phenylindole and mounted.

RNA extraction and real-time PCR

TRIzol (Invitrogen) was used per the manufacturer's instructions with the exception that the quantities were scaled down by half to account for the small size of the prostate. TRIzol in a volume of 500 μl was added to each prostate lobe and tissues homogenized using an electric homogenizer. RNA was extracted from the aquous phase, after incubation with phenol:choloroform, using isopropanol. RNA was then washed with 75% ethanol. Real-time PCR was performed using 1 μg of RNA, 4 μl of 5× iScript reaction mix, 1 μl of iScript reverse transcriptase (Bio-Rad Laboratories, Hercules, CA), and enough water to bring the total volume to 20 μl. Reverse transcription was completed using the program 95 C for 5 min, 50 C for 45 min, and 70 C for 15 min.

Quantitative-PCR (Q-PCR) of smERα-regulated gene profile

To each well of a 96-well plate, 4 μl of a 1:10 dilution of cDNA, 5 μl of iQ SYBR Green Supermix reagent (Bio-Rad Laboratories), and 1 μl of the appropriate primer were added. Q-PCR was run using an iCycler (Bio-Rad Laboratories). Data were normalized to β-actin using the ΔΔC_T method. Primers used to detect the gene profile of rat smooth muscle PS-1 cells are shown in Table 1.

Table 1.

Growth factor qPCR primer list

Primer target	Primer sequence
IGF	F: 5′-TGTGTACTGTCTTCTTCTGCCTGCA-3′
	R: 5′-AGCTCCCACAGAACCGCACA-3′
IGF binding protein 5	F: 5′-ATCTGACCAAGGCCCCTGCC-3′
	R: 5′-GCAGATGCCACGTTTGCGGC-3′
Platelet-derived growth factor α	F: 5′-CGACTCAGGTCCAGCGTGGT-3′
	R: 5′-ACAGAGGCACCCTCTCTTGGCC-3′
EGF	F: 5′-CCATCCTCAACTTTTCTGGGGCTCA-5′
	R: 5′-ACACGGGGAAGGCCAGAGAGC-3′
β-Actin	F: 5′-GCGTCCACCCGCGAGTACAA-3′
	R: 5′-TCCATGGCGAACTGGTGGCG-3′
Chemokine (C-C motif) ligand 2	F: 5′-ACGTGCTGTCTCAGCCAGATGC-3′
	R: 5′-GCTTCTTTGGGACACCTGCTGCT-3′
Fibroblast growth factor 10	F: 5′-CGGGGAGGCATGTGCGAAGC-3′
	R: 5′-GTCCCGCTGACCTTGCCGTT-3′
Fibroblast growth factor 7	F: 5′-ACACCCGGGGCACTGCTCTA-3′
	R: 5′-CAGTTCACGCTCGTGGCCGT-3′
Bone morphogenic protein 4	F: 5′-GCGGGACTTCGAGGCGACAC-3′
	R: 5′-ATCCGGGATGACGGCGCTCT-3′
Fibroblast growth factor 2	F: 5′-AACGGCGGCTTCTTCCTGCG-3′
	R: 5′-AGTTTGACGTGTGGGTCGCTCT-3′
Androgen receptor	F: 5′-GGCTACACTCGGCCCCCTCA-3′
	R: 5′-CTGTCCAAACGCATGTCCCCA-3′
G protein-coupled ER	F: 5′-CGCCGTGCTCTGCACCTTCA-3′
	R: 5′-GCTTTGGCCAGCGCCAGGTA-3′

F: Forward; R, reverse.

Branching morphogenesis

Prostates were removed en bloc from animals into PBS. After microdissection of individual prostate lobes, each lobe was incubated in 1% collagenase in PBS at 37 C for the following times: VP, 5 min; dorsal-lateral prostate, 10 min; and anterior prostate, 15 min. After collagenase incubation, tissues were removed to a PBS wash and then placed in fresh PBS for further dissection. Stroma was carefully peeled away on a compression slide using fine forceps and a 30-gauge needle tip. Pictures were taken using a Leica MZ1GF microscope with attached camera DFC480 and IM50 Image Manager imaging software (Leica). Ductal tips, branches, and branch points were counted and quantified.

Testosterone ELISA assay

Mouse blood was collected by cardiac puncture. The blood was collected into a serum separator tube, spun down, and serum was collected to detect testosterone using an ELISA kit following the manufacturer's instructions. Briefly, 50 μl of serum were added to the supplied microtitration strips, and 100 μl of testosterone and 100 μl of enzyme conjugate solution were added. Strips were shaken vigorously for 1 h and then washed five times with wash solution. The 3,3′,5,5′-tetramethylbenzidine chromagen solution was added, the strips shaken for 30 min, and then 100 μl of stop solution were added to stop the reaction. The absorbance was read at 450 nm with an ELISA plate reader (BioTek Instruments, Inc., Winooski, VT) and the data analyzed with SoftMax Pro 3.1.1 (Molecular Devices Corp., Sunnyvale, CA).

Cell studies

PS-1 and BPH-1 cells were maintained in normal media (DMEM) (Invitrogen) with 10% FBS. Cells were infected using a lentivirus pWPI-blasticidin vector containing flag-mERα or vector control plasmid and subsequently selected using 5 μg/μl blasticidin.

Conditioned media (CM) and growth of BPH-1

PS-1 ERα or vector lentivirus-infected cells were plated in media containing 5% charcoal-stripped FBS for 1 d in a 10-cm² plate. On the 2nd day, the media were changed, plates were treated with 10 nm 17β-estradiol (E₂), and CM was collected 24 h after treatment. BPH-1 cells were plated 5 × 10³ cells/well in a 24-well plate in 5% charcoal-stripped FBS media and allowed to attach overnight. The next day BPH-1 cells were subjected to CM as d 0. Fresh CM was added to cells every other day. Growth was measured by exposing cells to 5% 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide for 2 h and then dimethylsulfoxide for 1 h. Absorbance was read at 560 nm.

Masson Trichrome staining

Visual confirmation of collagen loss was obtained by use of the Accustain Trichrome kit (Sigma Diagnostics, St. Louis, MO). Slides were deparaffinized and then soaked in preheated 55 C Bouins solution for 15 min. Slides were allowed to cool in running tap water for 5 min and then stained in working Weigert's iron hematoxylin solution for 5 min, rinsed in deionized water, and stained in Biebrich scarlet acid fuschin for 2 min. Slides were rinsed again and then placed in working phosphotungstic/phosphomolibdic acid solution for 5 min, then placed directly into aniline blue for 30 min. Slides were washed in 1% acetic acid for 2 min and then dehydrated and mounted. ImageJ was used for separating out the blue color channel and then quantifying the amount of collagen per area of the basement membrane.

Picrosirius red staining

After rehydration, slides were stained in Weigert's hematoxylin for 8 min. Slides were rinsed and then placed serially through solutions A (2 min), B (60 min), and C (2 min) from the picrosirius red kit (Polysciences, Inc., Warrington, PA), with a wash in between each solution. Slides were dehydrated through xylene and mounted. Red collagen color was quantified using ImageProPlus (Media Cybernetics, San Diego, CA). For each field (total of 15 or more high-powered fields/mouse), the basement membrane was isolated as a region of interest using ImageProPlus (Media Cybernetics). The total area of red collagen fibrils was divided by the area of the basement membrane in that field. The relative amount of collagen was then averaged among the field and this average compared between wild-type (Wt) and smERαKO mice.

Statistical analysis

Student's two tailed t test with Welsh correction was used to determine significance between mice branching morphogenesis, testosterone levels, and collagen fraction. One way ANOVA was used to analyze growth data, and two way ANOVA with Bonferroni post hoc test was used to analyze the IGF-I/epidermal growth factor (EGF) time course and prostate glandular infolding. GraphPad Prism 5 (GraphPad Software, San Diego, CA) and Microsoft Office 2007 (Microsoft, Redmond, WA) software was used in deriving significance.

Results

Generation and characterization of smooth muscle-specific ERαKO mice

Floxed ERα mice were created using the cre/lox method as described previously (6, 27). Tgln is encoded by the SM22α gene in smooth muscle tissue (36). Combining flox ERα mice with cre recombinase linked to the Tgln promoter, we were able to create smERαKO mice (Fig. 1A). For experimental purposes, ERα F/F mice or ERα F/+ littermates were used as Wt controls. Genotyping results from a smERαKO mouse and Wt littermate are demonstrated in Fig. 1B. To control for any aggression-related increase in testosterone levels, mice were separated for 1 wk before serum was collected and analyzed for circulating testosterone levels. These levels were not significantly altered between smERαKO mice and Wt controls (Fig. 1C), indicating that any change in phenotype is not related to variation in testosterone levels. Further comparison of other organs, including bladder, intestine, stomach, aorta, and testes, and heart function revealed no significant difference (Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org).

Fig. 1.

Breeding and characterization of smERαKO mice. A, Tgln Cre male mice were mated with floxed ERα mice. The resultant Tgln heterozygous floxed ERα mice were then mated again to floxed ERα mice to produce smERαKO mice. B, Genotyping of smERαKO. Lane 1 demonstrated the Tgln cre band at 233 bp and the homozygous flox band at 881 bp. Wt mice (lane 2) showed either a single floxed ERα band or one floxed ERα band and one Wt ERα band at 741 bp (not shown). C, Testosterone levels of Wt and smERαKO male mice. The sera were collected from smERαKO and age-matched Wt littermates of 12 wk old, and testosterone levels were detected using ELISA. Testosterone levels remain unchanged between Wt and smERαKO mice.

smERα plays a key role in regulating prostate glandular infolding

The prostate is a complex structure that, in the mouse, matures and becomes quiescent by sexual maturity (37). To examine the phenotype changes after complete prostate maturation, mice were euthanized at 12 wk of age. Along with proximal to distal growth, the prostate epithelial cells also form infolding within each duct, thereby increasing surface area (38). If prostate growth is affected by stromal ERα, then it may also stand to reason that epithelial infolding is affected through stromal-epithelial interactions. To evaluate the effect of ERαKO in smooth muscle and fibroblast cells on prostate development, prostate histololgy was evaluated. In smERαKO mice, prostate morphology is changed in a spatial manner with the proximal area of the gland exhibiting less infolding compared with Wt mice (Fig. 2, A and A-1). This decrease was quantified as the number of folds per square millimeter of epithelial circumference (representative fields used for counting are displayed in Supplemental Fig. 2). This change was found to be significant. fERαKO mice do not display this phenotype and appear to have glandular structure similar to that of Wt mice (Fig. 2A). The prostate develops in a proximal to distal fashion, and it is possible that during the initial growth phase, smERα regulates infolding. However, in later growth phases, the distal area of the smERαKO prostate is able to catch up to its Wt counterpart due to loss of ERα and thinning of the smooth muscle cells toward the distal end of the prostate during this time.

Fig. 2.

Phenotypic changes of smERαKO mice. A, Histological comparison of smERαKO, fERαKO, and Wt mouse prostates of 12 wk. Histological changes were present between the groups, hemotoxylin and eosin-stained smERαKO, fERαKO, and Wt mouse prostates at 12 wk of age were compared. Proximal (Prox) and distal (Dist) areas of the prostates were compared. The proximal area of the smERαKO VP has a reduction of infolding as compared with the VPs of Wt and fERαKO (arrows signify areas of magnification). A-1, Quantification data reveals a significant decrease in smERαKO mouse VP infolding (P < 0.01). No significant change was seen in fERαKO mice. B and B-1, Collagen deposition was measured using picrosirius red staining. smERαKO mice show significantly less collagen fibrils than Wt mice or fERαKO mice (P < 0.05). C and C-1, VPs were digested using collagenase, and the tips were counted and quantified. smERαKO mice show no difference in the amount of tips compared with Wt, yet fERαKO mice do (P < 0.05) (n = 3).

Collagen and basement membrane show disregulation in smERαKO mouse prostates

Because evidence suggests that glandular development can be influenced via the extracellular matrix and basement membrane (7), we decided to stain mouse tissues with picrosirius red to identify collagen deposition. The basement membrane is composed of a host of proteins, one of the major ones being collagen type IV, although other collagens are present, such as III and VII (39). In the smERαKO mouse model, the VP shows a significant reduction in basement membrane thickness, quantified as the total amount of collagen present over the basement membrane thickness per field of picrosirius red-stained mouse VPs (Fig. 2, B and B-1). Interestingly, the fERαKO mice do not display this phenotype and show no significant difference when collagen amount is compared with Wt mice. Along with our previous results, this indicates that ERα exhibits its influences via a spatial and cell type-specific manner.

Prostate bud formation is reduced in fERαKO mice but not in smERαKO mice

The prostate growth is a tightly regulated process that involves proximal to distal growth and secondary and tertiary branching of the prostate ducts (38). To determine whether stromal ERα was involved in this prostate branching morphogenesis, mouse prostates were collected and microdissected. After microdissection, prostate tips, branches, and branch points were counted using a dissection microscope. The smERαKO mice prostates do not show a significantly reduced number of prostate tips in the VP (Fig. 2, C and C-1). However, in agreement with previous reports from our lab, fERαKO mice do show a decrease in the number of VP tips. Thus, we can conclude that fERα plays a role in branching morphogenesis but not in the prostate infolding or collagen deposition.

ERα is able to regulate growth and collagen through IGF/EGF and matrix metalloproteinase 9 (MMP9) pathways

To determine the mechanism through which ERα exerted its influence, rat PS-1 cells were used as an in vitro model. PS-1 cells have been characterized previously as being a prostate smooth muscle cell type (40). First, cells that expressed ERα were created and verified using immunofluorescent staining and Western blotting (Fig. 3, A and C). When ERα was introduced into PS-1 cells, these cells grew significantly faster than vector-infected cells (Supplemental Fig. 3). Working under the assumption that ERα was influencing epithelial cells in a paracrine mechanism, CM from PS-1 cells with or without ERα was collected. This CM was used to treat BPH-1 epithelial cells and the growth effect observed (Fig. 3B). BPH-1 cells exposed to CM from PS-1 ERα-infected cells grew significantly better than cells treated with PS-1 vector-infected CM. Several growth factors that are involved in prostate development were assayed, and Q-PCR results showed a significant increase of both EGF and IGF-I in ERα+ cells (Fig. 3D). This increase was not due to differences in cell passage number between parental PS-1 cells and infected PS-1 cells (Supplemental Fig. 3). Previous studies have shown that E₂ can up-regulate IGF-I in smooth muscle cells and that this up-regulation can increase these cells' proliferation (41). In addition to IGF, we have also observed that ERα can regulate EGF in other cell lines (Yeh, S., unpublished data). EGF has previously been linked to prostate cell differentiation through an androgen receptor (AR)-dependent axis (42). To determine whether IGF-I and EGF are direct target genes of ERα, we performed a time course with 0, 6, and 24 h of 10 nm E₂ treatments in PS-1 ERα+ cells. At 6 h, no difference in EGF/IGF-I signal is seen, indicating that these genes are most likely not direct targets of ERα. At 24 h, the levels of these two genes are significantly changed between vector and ERα+ PS-1 cells, (Fig. 3E), further suggesting an indirect regulation. ERα has been linked to IGF-I expression in smooth muscle in a previous report (41) that showed decreased IGF-I function in smooth muscle cells isolated from Balb/C Wt mice, thus it remains unclear whether this regulation is direct. To investigate further, we performed promoter analysis of IGF-I using Dragon ERE software version 3 (http://datam.i2r.a-star.edu.sg/ereV3/index.html). This analysis did not identify any perfect ERE within 6 kb upstream of the IGF-I promoter, although one noncanonical site was flagged (data not shown), which may account for the slight, but insignificant, increase in IGF-I expression seen at 6 h. Together, these data suggest that IGF-I and EGF are regulated in a secondary indirect fashion. It is important to note that in addition to IGF/EGF regulation, (C-C motif) ligand 2 (CCL2) is found to be down-regulated in ERα+ PS-1 cells, whereas bone morphogenic protein 4 (BMP4) and AR show no significant change (Fig. 3D). Because a literature search provided no direct link between ERα and MMP9, it is possible that this regulation is through CCL2, which has been shown to up-regulate MMP9 in prostate cells (43). AR levels remain unchanged, so we can conclude that the growth phenotype observed was not caused via increased AR signaling. Similarly, ERβ protein amounts were not dramatically changed between vector and ERα-infected PS-1 cells (Supplemental Fig. 3). Our previous reports with fERαKO mice showed that BMP4 was a target of ERα responsible for branching morphogenesis (22). Here, we show that ERα is unable to change the levels of BMP4 in smooth muscle cells. These data are consistent with our mouse findings, where branching morphogenesis was unchanged in smERαKO mice. MMP9 is a matrix metalloproteinase known to degrade components of the basement membrane (44, 45). To see whether MMP9 is involved in the ERα-mediated collagen decrease, protein extracts from PS-1 cells with or without ERα were tested for MMP9 protein levels via Western blotting. Results show that ERα is able to inhibit MMP9 at the protein level (Fig. 3C). It is possible that through a pathway involving EGF/IGF, ERα is able to maintain normal prostate growth and also to maintain the basement membrane integrity through down-regulation of MMP9, possibly through a CCL2-related mechanism.

Fig. 3.

Validation of ERα roles and mechanism in rat smooth muscle PS-1 cells. A, PS-1 cells containing ERα were created and verified using immunofluorescence and Q-PCR. B, BPH-1 epithelial cells were treated with CM collected from PS-1+/− ERα cells. When ERα is not present, the CM failed to stimulate prostate epithelial cell growth. C, MMP9 was increased in smooth muscle PS-1 cells lacking ERα. Total MMP9 protein was detected using Western blotting and rabbit polyclonal MMP9 antibody (Abcam, Cambridge, MA). D, Q-PCR analysis of several growth factors. mRNA from PS-1 cells with or without ERα was collected 24 hours after treatment with 10nM E₂. cDNA was obtained via reverse transcription from mRNA and qPCR analysis was performed using a panel of growth factors. E, PS-1 cells with or without ERα were treated with 10 nm E₂, and RNAs from cell lysates were collected at various time points. Results are representative of three independent experiments. DAPI, 4′,6-Diamidino-2-phenylindole; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (*, P < 0.05; **, P < 0.01).

Mice lacking ERα in both smooth muscle and fibroblasts develop an aggregate prostate phenotype

Mice lacking ERα in both smooth muscle and the fibroblasts were created (dERαKO) by mating Tgln cre with FSP cre and the resultant offspring with ERα F/F mice to obtain offspring mice displaying all three genotypes (Supplemental Fig. 4). In general, the dERαKO mice displayed a compound phenotype of both the fERαKO and smERαKO mice. dERαKO mice VPs displayed a decrease in the number of ventral tips, which is found to be significant when compared with Wt and is similar to the decrease seen in fERαKO mice (Fig. 4, A and C). When gross histology was looked at to determine infolding amounts, the same phenotype as the smERαKO mice was observed in the dERαKO mice. dERαKO mice show a significant decrease in the proximal epithelial infolding compared with Wt mice (Fig. 4, B and D). Finally, collagen deposition was also significantly decreased in dERαKO mice compared with Wt mice, and this was consistent with smERαKO mice (Fig. 4, E and F).

Fig. 4.

dERαKO mouse prostates display an aggregate phenotype of fERαKO and smERαKO mice. A and C, dERαKO mouse VPs display reduced branching morphogenesis similar to fERαKO mice when compared with prostates of age-matched Wt littermates. (P < 0.01, n = 4). B and D, dERαKO mouse prostates have significantly decreased glandular infolding compared with that of age-matched Wt littermates (P < 0.01). E and F, Collagen deposition is also decreased in prostates of dERαKO mice, similar to smERαKO mice. The decrease was found to be significant compared with Wt mice prostates (P < 0.05) (n = 6). Prox, Proximal; Dist, distal.

Discussion

The present study shows that ERα acts through both smooth muscle and fibroblast cells to influence the prostate epithelial cells. It has been shown that smooth muscle cells can have an influence on epithelial cells in the context of development (46). However, it remains unclear in what manner ERα may affect this cross talk. Studies using ubiquitous ERαKO mice have shown that ERα plays a role in the prostate organogenesis (27), and here, we show how, specifically, two stromal cell types differentially influence the prostate organ homeostasis. Additionally, epithelial ERαKO mice via prostate epithelial-specific probasin-cre show no branching morphogenesis abnormalities (47). Specifically, our smERαKO model demonstrates that ERα plays a supporting role in the maintenance of the basement membrane, of glandular infolding, and in gross prostate morphology in a cell type-specific manner.

Originally, it was expected that ERαKO in both smooth muscle and fibroblast cells would show an even more severe decrease in the VP branching morphogenesis than was seen in the fERαKO alone. However, instead of showing a more severe singular phenotype, dERαKO mice showed an overall more severe defect of prostate development. This dual ERα role led to prostates of dERαKO mice demonstrating a combination of the phenotypes seen from fERαKO and smERαKO mice.

Smooth muscle cells are two to four layers thick around the proximal area and thinner toward the distal area with a thin layer of periductal fibroblasts between them and the epithelial cells (33). Our data suggest that ERα in the smooth muscle cells plays an important role in epithelial cell differentiation in a spatial-dependent manner. In the proximal area, where smooth muscle cells are prevalent, ERα is able to easily exert its influence to assist in normal epithelial cell differentiation, but in the distal area, smooth muscle cells are not as abundant, so ERα may not play as important of a role. From this result, we hypothesize that potential growth factors from the fibroblasts, and not smooth muscle cells, which are not regulated by ERα, could help to maintain the normal epithelial differentiation pattern. In normal prostate development, ERα and smooth muscle IGF/EGF may cooperate to influence the glandular development (Fig. 5). Our group has previously shown that fERαKO mice show a decrease in branching morphogenesis as adults as well as a host of other urogenital phenotypes. However, it would seem that although fERα plays an important role in branching morphogenesis, smERα is more important for maintenance of the glandular structure and basement membrane thickness. Of note is that estrogenization of neonatal male mice results in an increase of periductal fibroblasts with the end result being a decrease in branching morphogenesis. The purported mechanism for this phenotype is through constraint of epithelial cells (7). In our fERαKO mice, a different mechanism through BMP4 regulation and stromal apoptosis is able to achieve a similar phenotype (22). Interestingly, when we assayed this gene, it was unchanged in the smERαKO mice (Supplemental Fig. 5). If a gene is up-regulated in fERαKO mice and down-regulated or unchanged in smERαKO mice, the net sum of the expression would display in the dERαKO mice. This effect may mask the importance of ERα in prostate gene expression. For this reason, it is important to separate out the fibroblast and smooth muscle cell ERα function.

Fig. 5.

ERα in different stromal cell types is able to exert differential effects on prostate epithelial cells. The figure depicts that ERα in smooth muscle is able to up-regulate IGF-I and EGF levels to influence epithelial cell growth, whereas down-regulation of MMP9 maintains basement membrane integrity. Stromal fERα through inhibition of BMP4 is able to decrease prostate branch morphogenesis.

By separating out the roles of the smooth muscle and fERα, we have exhibited the power of the cre/lox system. Previous research into the role of ERα in prostate development has used tissue recombination using tissues derived from ERα total KO or Wt mice. Our findings demonstrate some similar phenotypes, because results from these studies have shown that ERα in the stromal compartment is important for prostate development (48). However, these models use systems where ER is removed from every cell type in the stroma, including endothelial cells, neuroendocrine cells, and, importantly, both smooth muscle and fibroblast cells, which are important for two different sets of phenotypes. Recent evidence has implicated the endothelial cells as an important factor for promoting prostate tumor metastasis (49). Using an endothelial cell-specific promoter, we could create a mouse model where only endothelial cell ERα is lost to discover the potential role of ERα in modulating endothelial cell-mediated prostate cancer invasion. A major drawback of the cre/lox mouse model is that complete removal of the target gene is often impossible. In our model, we see that there is 50% cre activity in the VP (Supplemental Fig. 6). This can pose a problem when interpreting data, because a partial KO may have a different phenotype than that in either a complete KO or Wt mouse. One good example of this is the TGFβ receptor II (TGFβR2). KO of TGFβR2, using loxP sites at introns 1 and 2 of Tgfβr2 and cre recombinase driven by the FSP promoter, resulted in prostatic intraepithelial neoplasia (PIN) lesions in the prostate. From this, Bhowmick et al. (50) concluded that the loss of TGFβ signaling via KO of TGFβR2 resulted in PIN lesions. In a tissue recombination model, Franco et al. (51) created a mixture of 50% normal prostate stromal cells and 50% normal cells where the TGFβR2 had been knocked out. When these cells were combined with normal epithelial cells, the resultant tissue recombinant was able to create adenocarcinoma (51). This finding was later expanded and explained in a similar system when TGFβR2 was completely knocked out, left intact, or modulated to 50% KO. In the intact group, benign prostate growth was increased, and in the total KO group, PIN lesions were observed (similar to FSP-TGFβR2 mice). In the 50% KO group, adenocarcinoma was observed (52). These data clearly present one of the shortcomings of transgenic mouse models, incomplete KO may lead to different phenotypes.

The role of ERα and ERβ in the initiation and progression of prostate cancer has been studied (reviewed in Refs. 28, 53), but only recently has evidence begun to come to light about whether stromal or epithelial ERα is important in this transition (54). Epithelial ERα has been shown to have no effect on prostate development at 12 wk but plays an important role in the development of squamous metaplasia (47). KO of ERα in the stromal could lead to a decrease in cancer growth/initiation. Perhaps more interesting would be the effect of smERαKO on prostate metastasis due to ERα's role in basement membrane integrity. Stromal ERα is already implicated in BPH initiation (10). This is especially poignant, because BPH is characterized by an increase in smooth muscle cells (55).

The IGF pathway has been studied extensively in normal and malignant prostate. IGF is important for normal cell proliferation and when up-regulated can lead to malignant transformation (56). Here, we show that ERα is able to regulate this protein's expression in a rat smooth muscle cell line and also at a gene level in a transgenic mouse model. Disruption of the IGF pathway during development could help to explain some of the phenotypes seen in our mouse model, such as the decreased infolding and decreased branching morphogenesis (27). EGF has been shown to act as a stimulator of cell growth through the MAPK/ERK pathway (57). EGF has been shown to mediate the ERα pathway in ovary (58), but it is unclear whether this interaction is found in the prostate. Further studies will reveal the intricate mechanisms behind EGF/ERα-mediated cell growth.

The dERαKO mice show an aggregate of the phenotypes observed in the smooth muscle and fibroblast cell types, indicating a distinct role for ERα in two separate stromal cell types. Our in vitro and in vivo data confirm the ability of smooth muscle ERα to modulate IGF-I and EGF expression, which has been shown to modulate stromal cell proliferation, potentially leading to the phenotype changes observed in this article.