Role of epithelial cell fibroblast growth factor receptor substrate 2α in prostate development, regeneration and tumorigenesis

Prostatic tumors are the most frequently diagnosed tumors in American males. Yet, the mechanisms underlying prostate tumor initiation, progression and escape from regulation by androgen are not well understood. Since many developmental events are recapitulated during tumorigenesis, understanding prostate development will aid comprehension of prostate tumorigenesis. Prostate development starts at embryonic day 17 (E17), when a group of urogenital epithelial cells in the hindgut endoderm grow out into the surrounding mesenchyme and form the anterior, ventral, dorsal and lateral prostate buds (Donjacour and Cunha,1988; Sugimura et al.,1986). The prostatic buds then undergo extensive branching morphogenesis prior to puberty. At pubertal stages, the prostate rudiments grow rapidly; the ducts elongate from distal points and exhibit complex infolding of the intraductal mucosum. It has been shown that several growth factor families, including the fibroblast growth factor (FGF) family, play important roles in regulating prostate morphogenesis(Donjacour et al., 2003; Huang et al., 2005; Lin et al., 2007a; Thomson and Cunha, 1999).

The FGF family consists of 22 gene products controlling a wide spectrum of cellular processes. The FGF elicits regulatory signals by activating FGF receptor (FGFR) transmembrane tyrosine kinases encoded in isoforms of four genes (McKeehan et al., 1998; Powers et al., 2000; Wang and McKeehan, 2003). In the prostate, members of the FGF and FGFR families are partitioned between the epithelial and mesenchymal compartments and underlie directionally specific and reciprocal communication between the two compartments. Ablation of the FGF signaling disrupts prostate development(Donjacour et al., 2003; Huang et al., 2005; Lin et al., 2007a). Aberrant FGF signaling, in particular the reduction in the resident epithelial FGFR2IIIb, the ectopic appearance of FGFR1, and the overexpression of FGF9 in epithelial cells, is associated with prostate tumor progression(Giri et al., 1999; Jin et al., 2003b; Kwabi-Addo et al., 2001; Lu et al., 1999; McKeehan et al., 1998; Polnaszek et al., 2004; Ropiquet et al., 1999).

Recently, we reported that ablation of FGFR2 in prostatic epithelial precursor cells in early prostate development inhibits prostatic ductal branching morphogenesis and the acquisition of androgen dependency during development (Lin et al.,2007a). Similarly, ablation of Fgf10, a ligand for the FGFR2IIIb isoform, also disrupts prostate development(Donjacour et al., 2003; Huang et al., 2005). The detailed mechanism underlying how FGFR2 elicits intracellular regulatory signals in the prostate remains elusive. FRS2α is an adaptor protein that links FGFR kinases to multiple downstream signaling pathways. Activation of the MAP kinase and phosphatidylinositol 3 (PI3) kinase pathways by FGFR1 is primarily mediated via FRS2α(Kouhara et al., 1997; Lin et al., 1998; Ong et al., 1996; Rabin et al., 1993). Whether the FRS2α-mediated signals are important for FGFR2 in control of prostate development and adult tissue homeostasis remains to be determined.

FRS2α is generally expressed in fetal and adult tissues(McDougall et al., 2001). Disruption of Frs2α alleles in early embryogenesis is lethal(Hadari et al., 2001). To circumvent this limitation and enable studies on later stages of development,we employed the loxP-Cre recombination system to inactivate Frs2α alleles in the prostate epithelium by crossing mice carrying loxP-flanked Frs2α(Frs2α^flox)(Lin et al., 2007b) and Nkx3.1^Cre knock-in alleles(Lin et al., 2007a). Similar to Fgfr2 ablation, disruption of Frs2α in the prostate epithelium reduced epithelial cell proliferation, impaired branching morphogenesis during prostate development, and impaired post-puberty androgen-dependent prostate regeneration. Unlike Fgfr2^cnprostates, Frs2α^cn prostates remained strictly dependent on androgen with respect to tissue homeostasis. Ablation of Frs2α significantly inhibited the initiation and progression of autochthonous mouse prostate tumors. This indicated that FRS2α-mediated mitogenic signals are important for prostatic development and regeneration,and suggest a role of aberrant FRS2α-mediated signaling in prostate tumorigenesis.

Mice carrying loxP-flanked Frs2α alleles, the R26R-lacZ reporter, and the Nkx3.1^Creknock-in alleles were bred and genotyped as described(Jin et al., 2003b; Lin et al., 2007b; Soriano, 1999). Orchiectomy and androgen therapy were carried out as described previously(Lin et al., 2007a). All animals were housed in the Program for Animal Resources of the Institute of Biosciences and Technology, and were handled in accordance with the principles and procedures of the Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the Institutional Animal Care and Use Committee.

TRAMP-C2 cells were maintained in RD (50% RPMI, 50% DMEM) medium with 5%FBS as described (Foster et al.,1997). For protein stability analyses, the cells(1×10⁶ cells in 10-cm dishes) were treated with 10 μg/ml cycloheximide for the indicated times. For FGF2 response analyses, the cells(1×10⁵ in 6-well plates) were serum starved for 24 hours prior to being treated with 10 ng/ml FGF2 for 10 minutes.

Prostate tissues were dissected at postnatal day 0.5 as described(Lamm et al., 2001) and transferred to 0.4 μm Millicell-CM filters (Becton Dickinson Labware Europe, Meylan, France) inside 24-well tissue culture plates. Each well contained 0.5 ml of serum-free DMEM:Ham's F-12 (1:1) supplemented with 2% ITS(12.5 μg/ml insulin, 12.5 μg/ml transferrin, 12.5 ng/ml selenious acid),2.5 μg/ml linoleic acid-albumin (Sigma, St Louis, MO), 25 μg/ml gentamycin, 0.25 μg/ml amphotericin B and 10 nM testosterone. PI3 kinase inhibitor (LY294002) and ERK1/2 inhibitor (Calbiochem, Darmstadt, Germany)were added to the medium at a final concentration of 10 μM where indicated. Cultures were maintained for 3 days at 37°C prior to harvest for analysis.

Prostates and prostate ducts were dissected and sectioned for histological analyses as previously described (Lin et al., 2007a). Main ducts and distal ductal tips were quantified from at least three animals. Data are presented as mean±s.d.

Hematoxylin and Eosin staining was performed on 5-μm sections;immunohistochemical analyses and in situ hybridization were performed on 7-μm sections mounted on Superfrost/Plus slides (Fisher Scientific,Pittsburgh, PA). The antigens were retrieved by autoclaving samples in 10 mM Tris-HCl buffer (pH 10.0) for 5-10 minutes or as suggested by manufacturers of the antibodies. The source and dilutions of primary antibodies are: mouse anti-cytokeratin 8 (1:15; Fitzgerald, Concord, MA); mouse anti-smooth muscleα-actin (1:1) and mouse anti-PCNA (1:1000) from Sigma (St Louis, MO);mouse anti-P63 (1:150), rabbit anti-FRS2α (1/500) and rabbit anti-androgen receptor (1:150) from Santa Cruz (Santa Cruz, CA); mouse anti-T antigens (1:500) from BD Pharmingen (Frankin Lakes, NJ); rabbit anti-phosphoAKT (1:500) and anti-phosphoERK1/2 (1:500) from Cell Signaling Technology (Danvers, MA); rabbit anti-probasin (1:3000) from the Greenberg laboratory (Fred Hutchinson Cancer Research Center, Seattle, WA); and rabbit anti-PSP94 (1:2000) from the Xuan laboratory (University of Western Ontario,London, Canada). The TSA Plus Fluorescein System from PerkinElmer (Shelton,CT) was used to visualize anti-FRS2α, anti-phosphoAKT and anti-phosphoERK1/2 antibodies and the ExtraAvidin Peroxidase System from Sigma to visualize all other antibodies. For whole-mount lacZ staining, the prostate tissues were lightly fixed with 0.2% glutaraldehyde for 30 minutes. lacZ staining was carried out by rocking in 1 mg/ml X-Gal at room temperature overnight as described (Liu et al., 2005).

For proliferation analyses in 1- and 2-week-old prostates, the percentage of PCNA-positive cells in the growing duct tips was determined, as most proliferating cells were located at the most-distal part of ducts at this stage. For 4-week-old and post-pubertal regenerating prostates, the percentage of PCNA-positive cells in the whole gland was determined, as the proliferating cells were randomly distributed in the whole gland. The data were collected from at least five slides per prostate, and three prostates per genotype, and are presented as mean±s.d.

Dissected prostate tissues or TRAMP-C2 cells were homogenized in RIPA buffer (50 mM Tris-HCl buffer pH 7.4, 1% NP40, 150 mM NaCl, 0.25%Na-deoxycholate, 1 mM EGTA, 1 mM PMSF), and extracted soluble proteins were harvested by centrifugation. Samples containing 50 μg protein were separated by SDS-PAGE and electroblotted onto nylon membranes for western analyses with the indicated antibodies. The dilutions of the antibodies are:anti-phosphoFRS2α, 1:1000; anti-phosphoERK1/2, 1:1000; anti-phosphoAKT,1:1000; anti-AKT, 1:1000; anti-FRS2α, 1:1000; anti-ERK1/2, 1:3000; and anti-FGFR2, 1:1000. For immunoprecipitation, soluble tissue lysates containing 500 μg protein were incubated with 2 μg of the indicated antibodies overnight, and then the immunocomplexes associated with 10 μl protein A-Sepharose beads were collected by centrifugation and extracted with SDS sample buffer for western analyses. The specific bands were visualized using the ECL-Plus chemoluminescent reagents. The films were scanned with a densitometer and the bands quantified using Image J software (NIH).

Secretory proteins were extracted as described(Lin et al., 2007a). The protein concentrations were determined and adjusted to a final concentration of 1 mg/ml with PBS. Samples containing 25 μg protein were separated by 5-20% gradient SDS-PAGE and visualized by Coomassie Blue staining. Samples of 50 μg protein were used for western blotting analyses with anti-probasin or anti-PSP94 as indicated.

For in situ hybridizations, paraffin-embedded tissue sections were rehydrated, followed by 20 μg/ml protease K digestion for 7 minutes at room temperature. After prehybridization at 65°C for 2 hours, the hybridization was carried out by overnight incubation at 65°C with 0.5 μg/ml digoxigenin-labeled RNA probes for the indicated genes. Non-specifically bound probes were removed by washing four times with 0.1×DIG washing buffer at 60°C for 30 minutes. Specifically bound probes were detected using alkaline phosphatase-conjugated anti-digoxigenin antibody (Roche,Indianapolis, IN).

For real-time RT-PCR analyses, total RNA was extracted from prostates using the RiboPure Kit (Ambion, Austin, TX). The first-strand cDNAs were reverse transcribed from the RNA template using SuperScript II reverse transcriptase(Invitrogen, Carlsbad, CA) and random primers according to manufacturer's protocols. Real-time PCR analyses were carried out using the SYBR Green JumpStart Taq Ready Mix (Sigma) as instructed by the manufacturer. Relative abundances of mRNA were calculated using the comparative threshold (CT) cycle method and were normalized to β-actin or 18S rRNA as an internal control. The mean and s.d. among at least three individual experiments are shown. RT-PCR products were analyzed on 2% agarose gels for validation of products by size.

Previously, we reported that FGFR2 signaling is important for prostate development and acquisition of androgen responsiveness. To determine whether Frs2α, a receptor-proximal adaptor in FGF signaling, was essential for prostatic development, the loxP-Cre recombination system was used to conditionally inactivate Frs2α alleles in prostatic epithelial cell precursors (Fig. 1A). Mice carrying homozygous Frs2α^cn alleles were viable and fertile. Disruption of the Frs2α in prostates was confirmed by PCR(Fig. 1B). Real-time RT-PCR analyses revealed that expression of Frs2α in 1-week-old Frs2α^cn prostates was significantly reduced compared with the control prostates (Fig. 1C).

Frs2α expression was further characterized by in situ hybridization with Frs2α-specific probes. The results showed that although Frs2α expression was constant in the stromal compartment, it exhibited a spatial and temporal expression pattern in the epithelium (Fig. 1D). In 1-week-old prostates, Frs2α was expressed relatively uniformly at high levels in the epithelium. In 2-week-old prostates, the expression became restricted to cells located at distal tips of the ductal network, where the cells were actively engaged in proliferation. In the prostates of mice at 4 weeks or older, expression of Frs2α was limited to basal epithelial and stromal cells. No Frs2α expression was detected in luminal epithelial cells. Ablation of Frs2α with Nkx3.1^Cre diminished Frs2α expression in the epithelial compartment of prostates, including basal and luminal epithelial cells (Fig. 1D),whereas Frs2α expression in the stromal cells remained intact. Thus, the basal level of Frs2α expression indicated by RT-PCR in prostates of Frs2α^cn and 4-week-old control mice was a constitutive property of stromal and basal cells.

Unlike Fgfr2^cn prostates that only have dorsal and lateral lobes, Frs2α^cn prostates were relatively normal and consisted of anterior, dorsal, lateral and ventral prostate (AP, DP, LP and VP, respectively) lobes, although these were smaller and more transparent than wild-type prostates (see Fig. S1A in the supplementary material). The prostates of genotype Frs2α^flox/flox,Frs2α^flox/WT,Frs2α^flox/cn, Frs2α^WT/WT or Nkx3.1^Cre, had no noticeable differences in organ morphology and histological structures, and were all considered as control prostates.

To improve direct visualization of branching morphogenesis, the R26R-lacZ reporter allele was introduced into the Frs2α^cn mice by crossing with mice bearing a lacZ reporter silenced by a loxP-flanked sequence(Soriano, 1999). Activation of the R26R-lacZ reporter by the Cre recombinase occurred concurrently with disruption of the Frs2α^flox alleles in the same cells. Whole-mount staining with X-Gal showed that similar to Fgfr2^cn embryos, prostatic buds formed at E17.5 in Frs2α^cn embryos with no apparent difference to controls. Although all mutant buds formed AP, DP, LP and VP lobes, fewer ductal branches were observed in the Frs2α^cn prostates than in controls(Fig. 2A). We then dissected each prostatic lobe from 1-week-old mice to examine the Frs2α^cn prostates in more detail. The X-Gal stain confirmed that that the complexity of the ductal network in the Frs2α^cn prostates was reduced. Quantitative analyses of ducts microdissected out of 4-week-old prostates revealed a reduction in the number of ductal tips in every lobe of the Frs2α^cn prostates(Fig. 2B). This indicated that the extent of branching morphogenesis, but not budding, was compromised in Frs2α^cn prostates.

To determine whether FRS2α protein was carried over into mutant prostatic buds after ablation of expression by the Nkx3.1^Cre-mediated recombination, the prostate rudiments at different stages were assessed with an anti-FRS2α antibody. FRS2α was strongly expressed in epithelial cells of control prostate rudiments and the expression gradually declined with development of the prostate. Although Nkx3.1^Cre is expressed prior to E17.5,FRS2α protein remained evident in mutant epithelial cells at E18.5 and then diminished at the neonatal stage (Fig. 3A). The number of cells positive for P63 (also known as Trp63 -Mouse Genome Informatics) remained relatively constant over the same period with a slight decrease in Frs2α^cn prostates at later stages (Fig. 3B). Most epithelial cells in normal prostate rudiments expressed P63(Pu et al., 2007). This indicated a persistence of FRS2α protein beyond ablation of expression. Separate experiments in vitro indicated that the half-life of FRS2α was longer than that of FGFR2 in prostate epithelial tumor cells(Fig. 3C,D). These results indicated that the carryover of FRS2α after ablation by recombination might be sufficient to participate in signals that support prostatic bud formation in early prostate morphogenesis.

Presence of the proliferating cell nuclear antigen (PCNA) showed that the number of proliferating cells in Frs2α^cndistal tips was reduced compared with controls at 1-2 weeks of age(Fig. 4A,B). At 3-4 weeks, the prostates were undergoing rapid growth associated with puberty. The proliferating cells were widely distributed across the prostates. Proliferation in whole dorsolateral prostate (DLP) and AP lobes in Frs2α^cn prostates was also lower than in the controls (Fig. 4A,B). Post-puberty cell proliferation at 6 weeks in both Frs2α^cn and control prostates was dramatically reduced. No difference was observed between Frs2α^cn and control prostates at this stage(data not shown). Together, the results demonstrate that FRS2α-mediated signals are important for prostate cell proliferation during development. Since these results mirror the Fgfr2 ablation, they suggest a role of FRS2α in mediating FGFR2 mitogenic signals in the epithelial cells during prostate development.

The expression of 22 key regulatory genes was compared between control and Frs2α^cn 1-week-old prostates by real-time RT-PCR. The Frs2α^cn prostates exhibited reduced expression of BMP7, HOXb13, HOXd13 and NKX3.1 and increased expression of FGFR4. No significant difference was observed between Frs2α^cn and control prostates in the expression of the other genes tested (Fig. 4C). HOXb13, HOXd13 and NKX3.1 are highly expressed in luminal epithelial cells and play a role in differentiation of luminal epithelial cells (Economides and Capecchi,2003). The results suggest that in addition to its effects on proliferation and branching morphogenesis, ablation of FRS2α might also compromise luminal epithelial cell differentiation.

As an adaptor, FRS2α has multiple binding sites for downstream substrate that are required for FGFR to activate the MAP kinase and AKT pathways (Gotoh et al., 2004; Ong et al., 2000; Xu and Goldfarb, 2001). To determine whether FRS2α-mediated signals activate the MAP kinase and AKT pathways, prostates were harvested from mice at 1, 2 and 4 weeks of age and subjected to immunoblot analysis of activated FRS2α, ERK1/2 (also known as MAPK3/1 - Mouse Genome Informatics) and AKT (also known as AKT1 - Mouse Genome Informatics). FRS2α was strongly expressed and phosphorylated in 1-week-old normal prostates relative to diminished levels in 4-week-old prostates (Fig. 5A). Coincident with changes in FRS2α, phosphorylation of ERK1/2 in the control prostates was strong in early development and diminished at later stages. By contrast, phosphorylation of AKT was relatively constant. Ablation of Frs2α reduced the phosphorylation of ERK, but not AKT(Fig. 5A,B). Immunohistochemical analysis of tissues using anti-phosphorylated ERK1/2(Fig. 5C) and anti-phosphorylated AKT antibodies (Fig. 5D) confirmed the immunoblot results. In contrast to Frs2α ablation, ablation of Fgfr2 reduced both ERK1/2 and AKT phosphorylation in addition to the expected reduction in phosphorylation of FRS2α (Fig. 5A,C,D). These results suggest that FGFR2 is a major source of FRS2α activation in prostate epithelial cells during development, and that activation of the MAP kinase pathway, but not the AKT pathway, by FGFR2 is likely to be mediated by FRS2α in prostates.

ERK1/2 inhibitors in organ culture of newborn prostates were employed to test whether inhibition of the MAP kinase pathway mimicked the FRS2αdeficiency with respect to prostatic branching morphogenesis. Similar to tissue from Frs2α^cn prostates, normal prostate tissue treated with the inhibitors exhibited decreased ductal complexity indicative of compromised branching morphogenesis of prostatic buds(Fig. 5E). However, inhibition of ERK1/2 did not further decrease the ductal complexity of Frs2α^cn prostates, implying that the FRS2α-mediated pathway is the major activator of the MAP kinase pathway during prostate branching morphogenesis. PI3 kinase inhibitors also significantly reduced the extent of ductal differentiation, similar to the effect of MAP kinase inhibitors. This suggests that both MAP and PI3 kinase pathways are important for prostate branching morphogenesis, but that FRS2α is primarily an effector of FGFR2 kinases for activating the MAP kinase pathway without compromise of the PI3 kinase pathway in prostate epithelial cells.

Although Frs2α^cn prostates were smaller overall, they exhibited a similar epithelial infolding to normal wild-type prostates (see Fig. S1 in the supplementary material). No significant difference was observed in the display of luminal cell cytokeratin 8 (also known as keratin 8 - Mouse Genome Informatics), or α-actin in stromal cells, or androgen receptor (AR) in both luminal epithelial and stromal cells,between Frs2α^cn and control prostates (see Fig. S2A in the supplementary material). Frs2α^cn prostates exhibited near-normal levels of secretory proteins, although some reduction in the production of probasin and PSP94 (also known as MSMB - Mouse Genome Informatics) was noted in the VP lobe (see Fig. S2B,C in the supplementary material).

Previously, we reported that ablation of the Fgfr2 alleles interfered with the strict androgen-dependent properties of prostate with respect to tissue homeostasis. To test whether ablation of Frs2α exhibited a similar effect, 8-week-old mice were deprived androgens by orchiectomy. Similar to the effect on control prostates, the androgen deprivation induced apoptosis in Frs2α^cn prostates within 2 days (data not shown) and the tissues atrophied in 2 weeks (see Fig. S3 in the supplementary material). This is in marked contrast to Fgfr2^cnprostates, which remained largely unresponsive to the same androgen deprivation after 2 weeks (Lin et al.,2007a). This suggested that resident epithelial cell FGFR2 instructed the prostate to acquire strict dependence on androgen via pathways other than those mediated by FRS2α.

During regeneration induced by restoration of androgen, expression of FRS2α was apparent in the restored prostate luminal epithelial cells(Fig. 6A). Real-time RT-PCR analyses confirmed that Frs2α expression was significantly increased in regenerating prostates as compared with the castrated remnants(Fig. 6B). This suggested that FRS2α plays a role in prostate regeneration. To test this, the rate of epithelial cell proliferation was compared among normal, FRS2α and FGFR2-deficient regenerating prostates induced by androgen. PCNA immunostaining of proliferating cells revealed that, similar to the effect of Fgfr2 ablation, disruption of Frs2α significantly reduced proliferation activity in the prostate epithelium(Fig. 6). Taken together, the results indicate that FRS2α potentially mediates mitogenic signals of FGFR2 during androgen-induced proliferation, but FGFR2 support of androgen-dependent gene expression in the prostate is mediated by FRS2α-independent pathways.

Although FRS2α was not expressed in luminal epithelial cells of mature normal prostates, it appeared prominently in luminal epithelial cells in prostate tumors in the TRAMP mouse (Fig. 7A,B), the tumors of which were induced by oncogenic T antigens targeted to prostate epithelial cells(Kaplan-Lefko et al., 2003). This was coincident with the ectopic appearance of FGFR1 in foci of high-grade PIN and overt tumor cells in the TRAMP prostates(Fig. 7A). These results are consistent with the previous finding that the ectopic appearance of normally stromal resident FGFR1 is often associated with prostate tumor progression(Jin et al., 2003a; Kwabi-Addo et al., 2001). The significantly increased phosphorylation of FRS2α, ERK1/2 and AKT in TRAMP-C2 epithelial cell lines derived from the TRAMP tumors induced by FGF2,indicated the presence of active ectopic FGFR1 signaling in the tumor cells(Fig. 7C). TRAMP-C2 cells are characterized by expression of ectopic FGFR1(Foster et al., 1999). FGF2 is specifically recognized by ectopic FGFR1, but not by resident FGFR2IIIb, in prostate epithelial cells (McKeehan et al., 1998). Thus, it is likely that FRS2α plays a role in mediation of ectopic FGFR1 in prostate tumors.

To determine whether FRS2α had any impact on tumorigenesis in the TRAMP mice, the Frs2α^cn and TRAMP mice were crossed. Histological analysis showed that by 10 weeks, the majority of TRAMP mice carrying the floxed Frs2α alleles developed the expected low-to-high-grade PIN lesions. PIN foci were apparent across the whole prostate with an estimated 80% of the prostate exhibiting various degrees and grades of PIN lesions (Fig. 7D,E). Although the TRAMP transgenic T antigens were highly expressed in the Frs2α^cn prostates, PIN lesions across the Frs2α^cn prostates were fewer in number and the majority of the lesions were lower grade than in control TRAMP mice (Fig. 7D,E). By 24 weeks, most control TRAMP mice developed advanced, poorly differentiated tumors, whereas tumors in the Frs2α^cn/TRAMP hybrids were well-differentiated (Fig. 7D). Application of a Kaplan-Meier survival analysis showed that ablation of Frs2α alleles significantly increased the life span of TRAMP mice (Fig. 7F). These results strongly suggest that ablation of FRS2α-mediated signaling pathways in prostate epithelial cells inhibits prostate tumor initiation and progression in the TRAMP mice. and this mostly likely occurs through the adaptor function of FRS2α with ectopic FGFR1.

Members of the FGF family play cell-specific roles in prostatic development, function, compartmental homeostasis and tumorigenesis(Lin et al., 2007a; McKeehan et al., 1998). Here we report that the expression of FRS2α, a major adaptor protein in the FGF signaling cascade, is spatiotemporally expressed and is closely associated with prostatic cell proliferation. Although diminished in the adult prostatic epithelium, expression of FRS2α was activated in epithelial cells of regenerating and tumor prostates. Tissue-specific ablation of Frs2α in prostatic epithelial precursor cells compromised prostatic branching morphogenesis and growth. Furthermore, ablation of Frs2α alleles retarded prostatic tumor formation and progression in the autochthonous TRAMP mouse model of prostate cancer. The data demonstrate that FRS2α is important for prostatic development and growth, and imply that aberrant activation of FRS2α-mediated signaling might contribute to prostatic tumorigenesis.

During prostatic organogenesis, FGF7 and FGF10 are expressed in the mesenchyme, and FGFR2IIIb, the cognate receptor for FGF7/10, is expressed in the epithelium of the distal parts of elongating and branching ducts in prostatic rudiments (Huang et al.,2005; Thomson and Cunha,1999). Ablation of FGFR2 in prostate epithelial cell precursors inhibits cell proliferation and cellularity throughout prostatic development(Lin et al., 2007a). Because of its role as a general adaptor bridging activated FGFR kinases and the MAP kinase, ablation of Frs2α potentially compromises activity of the FGF7/10-FGFR2IIIB axis. Together with the data showing that ablation of Fgfr2 concurrently reduced activation of FRS2α and the MAP kinase pathway, the results suggest that FGFR2 signals for activating the MAP kinase pathway and regulating cell proliferation in prostate epithelial cells were mediated by FRS2α. By contrast, phosphorylated AKT was reduced in Fgfr2^cn, but not in Frs2α^cn, prostates(Fig. 5). This indicates that FGFR2 activates the AKT pathway via an FRS2α-independent mechanism. Unlike Fgfr2 ablation, which abrogates anterior and ventral prostatic bud formation, ablation of FRS2α did not disrupt development of prostatic buds. It is not uncommon that the prostate has a lobe-specific response to morphoregulatory genes(Economides and Capecchi, 2003; Podlasek et al., 1997; Pu et al., 2007). Because the apparent half-life of FRS2α is longer than that of FGFR2, our results could not resolve whether FGFR2 signals in prostatic bud formation were mediated by FRS2α-independent pathways or by the carryover of FRS2α. Further experiments are needed to clarify this issue.

Several signaling pathways, including those of SHH, Notch, BMP and FGF,have been implicated in prostate development. Ablation of Frs2αor Fgfr2 alleles reduced expression of HOXd13. In addition, ablation of Frs2α reduced expression of BMP7 and HOXb13 and increased expression of FGFR4, FGF7, FGF10 and BMP4. Of this set of regulators, ablation of Fgfr2 reduces expression of BMP4(Lin et al., 2007a). This difference suggested that FGFR2 signaling in prostate epithelial cells is at least in part mediated by FRS2α, but that FGFR2 is not the only upstream activator of FRS2α in the prostate. Notably, HOXb13 is only expressed in ventral prostates. Concurrent inactivation mutations of both Hoxb13and Hoxd13 cause severe hypoplasia in ventral prostate morphogenesis(Economides and Capecchi,2003). Our results suggest that both Hoxb13 and Hoxd13 might be downstream targets of FRS2-mediated signaling cascades in the prostate, and that FRS2α-mediated signals are crucial for ventral epithelial cell differentiation. The expression of PSP94 and probasin, which are secretory proteins characteristic of mature luminal epithelial cells, was consistently and significantly reduced in Frs2α^cn ventral prostate (see Fig. S2B in the supplementary material). These results suggest that ablation of Frs2α compromises luminal epithelial cell differentiation in ventral prostate. Further experiments will be carried out to test this possibility.

Knock-in of the Cre cDNA to the Nkx3.1 locus disrupts one Nkx3.1 allele. Consistent with an earlier report that verified reduced expression of Nkx3.1 in heterozygous Nkx3.1-null prostates (Bhatia-Gaur et al.,1999), our results showed a reduction in expression of Nkx3.1 in Frs2α^cn prostates. By contrast, expression of Nkx3.1 in Fgfr2^cnprostates that also carry only one Nkx3.1 allele remained similar to that in controls (Lin et al.,2007a). Others reported that Nkx3.1 expression in prostate is regulated both by androgen and FGF signaling pathways(Pu et al., 2007). Thus, it is possible that expression of Nkx3.1 is repressed by FGFR2 via FRS2α-independent pathways and that the repression is alleviated upon ablation of FGFR2 and to a degree sufficient to result in a net expression of Nkx3.1 equal to that of wild-type prostates. It is also possible that loss of FGFR2 activates the non-canonical expression of Nkx3.1.

Ablation of Fgfr2 alleles compromises the acquisition of dependence of the prostate on androgen with respect to tissue homeostasis. By contrast, FRS2α-deficient prostates remained strictly androgen-dependent. Similar to controls, Frs2α^cn prostates responded to androgen deprivation within 24 hours (see Fig. S3 in the supplementary material and data not shown). This and the fact that FRS2α was very low or absent in luminal epithelial cells of mature prostates, suggest that FRS2α is not essential to FGFR2 signaling with regard to its effects on the androgen-dependence of adult prostates.

The prostate is one of the few organs that have regenerative capacity in mice. Deprivation of androgen by orchiectomy and subsequent restoration induces rapid regression and regeneration, respectively. Although expression of FRS2α was very low or absent in the mature resting prostate epithelium, its expression was significantly increased in the regenerating epithelial cells induced by androgen. Although, overall, the androgen-induced regeneration of the prostate was not blocked by the absence of FRS2α,the rate and extent of proliferation during the process were compromised(Fig. 6). Thus, the elevated expression of FRS2α in regenerating epithelial cells contributes to the rate of androgen-induced prostate regeneration, but is not essential for it.

FRS2α has four tyrosine phosphorylation sites for GRB2 binding that link the FGFR kinase to the PI3 kinase/AKT pathway, and two for SHP2 (also known as PTPN11 - Mouse Genome Informatics) binding that link it to the MAP kinase pathway and are important for mediating mitogenic signals of FGFR(Gotoh et al., 2004; Kouhara et al., 1997; Ong et al., 1996; Ong et al., 1997; Xu and Goldfarb, 2001). The four GRB2-binding sites are required for mediating signals to activate the FiRE enhancer element of the mouse syndecan 1 gene(Zhang et al., 2007). We previously reported that FRS2α phosphorylation by FGFR kinases in rat prostate tumor cells is FGFR isotype-specific and associated with the promotion of cell proliferation by the presence of ectopic FGFR1(Wang et al., 2002). Here we showed that from the early branching morphogenesis to mature stages, the expression of FRS2α in luminal epithelial cells of mouse prostates was proportional to the number of cells actively engaged in proliferation. Expression of FRS2α was transiently activated in the luminal epithelial cells during regeneration and appeared in PIN lesions and tumors where the cells were also actively proliferating. Our unpublished results using in situ hybridization have confirmed that that Fgfr1 is not expressed in prostate epithelial cells during development, or during androgen-induced regeneration, or in resting mature prostates. Moreover, prostate epithelial cell-specific ablation of Fgfr1 using Nkx3.1^Creinduced no detectable abnormalities in prostate development or regeneration of mature prostates. However, FGRF1 was apparently expressed in the epithelial cells of lesion foci in the TRAMP prostate(Fig. 7A). Thus, the coincidence of the ectopic appearance of FGFR1 coupled with the constitutive expression of FRS2α might support and drive the neoplastic properties of the tumors, particularly with respect to mitogenesis. Prevention of expression of ectopic FGFR1 in tumor epithelial cells by a cross between ablation of the Fgfr1 alleles specifically in prostate epithelial cells and the TRAMP mice should shed light on this subject. Together, the results reveal that the ectopic FGFR1/FRS2α signaling axis is a feature of the TRAMP tumor epithelial cells. Disruption of the axis by ablation of the Frs2α alleles is likely to interfere with tumor initiation and progression in the TRAMP model.

In summary, we report that FRS2α is spatiotemporally expressed in the prostatic epithelium and plays a role in prostate development, androgen-driven regeneration and tumorigenesis. The major role appears to be in mediation of prostatic epithelial cell proliferation in these three processes, rather than a direct effect on androgen responsiveness and differentiated function. The prostate epithelial cell-specific Frs2α^cnmice provide an additional useful model for scrutinizing not only the molecular mechanisms underlying prostate growth and development, but also how prostate cancer cells escape from strict controls on tissue homeostasis to propagate autonomously.

We thank Mary Cole for critical reading of the manuscript. The work was supported by AHA0655077Y from the America Heart Association, DAMD17-03-0014(F.W.) from the US Department of Defense, and CA96824 (F.W.), CA115985(M.M.S.), DK076602 (M.M.S.), CA59971 (W.L.M.) and CA84296 (N.M.G.) from the NIH.