Notch1-expressing cells are indispensable for prostatic branching morphogenesis during development and re-growth following castration and androgen replacement.

Notch expression is frequently associated with progenitor cells, and its function is crucial for development. Our recent work showing that Notch1 is selectively expressed in basal epithelial cells of the prostate and higher Notch1 expression during development suggests that Notch1-expressing cells may define progenitor cells in the prostate. To test this hypothesis, we have generated a transgenic mouse line in which the Notch1-expressing cells can be ablated in a controlled manner. Specific targeting was achieved by expressing the bacterial nitroreductase, an enzyme that catalyzes its substrate into a cytotoxin capable of inducing apoptosis, under the Notch1 promoter. Cell death in transgenic prostate was confirmed by histological analyses including terminal dUTP nick-end labeling and caspase 3 immunocytochemical staining. We evaluated the consequences of ablation of Notch1-expressing cells in two systems, organ culture of early postnatal prostates and re-growth of prostate in castrated mice triggered by hormone replacement. Our data show that elimination of Notch1-expressing cells inhibited the branching morphogenesis, growth, and differentiation of early postnatal prostate in culture and impaired prostate re-growth triggered by hormone replacement in castrated mice. Furthermore, we found that Notch1 expression following castration and hormone replacement was concomitant with known basal cell markers p63 and cytokeratin 14 and was high in the proliferative human prostate epithelial cells. Taken together, these data suggest that Notch1-expressing cells define the progenitor cells in the prostatic epithelial cell lineage, which are indispensable for prostatic development and re-growth.

The prostate undergoes proliferation and differentiation not only during development but also following castration and androgen replacement. In rodents, the prostate is formed by budding of the urogenital sinus at late embryonic stages under the influence of androgen and epithelial cell interactions with stromal cells (1). At the cellular level, the glandular structure includes epithelium and stroma. The epithelium is composed of essentially two major types of cells, basal and luminal epithe-lial cells that can be in part distinguished by their expression of different markers. Whereas basal cells are positive for cytokeratin (CK) 1 5, CK14, and p63, luminal cells express CK8, CK18, and Nkx3.1 (2,3). It has been well documented that androgen is required for prostatic morphogenesis and maintenance. Without androgen, the prostate cannot be generated during embryogenesis. Upon castration or androgen deprivation, significant apoptosis occurs in the mature epithelium, and the prostate shrinks in size (4). However, upon androgen replacement, the prostate re-grows to the original size. Defining the nature of the progenitor cells during prostatic development and re-growth will enhance our understanding of normal and pathologic processes in the prostate.
Recently, we have reported dynamic expression patterns of Notch1 during prostatic development and tumorigenesis (5). Immunostaining of prostates from a Notch1-GFP transgenic mouse line, in which expression of GFP is under control of the Notch1 promoter (6), revealed that Notch1-expressing cells localize within the basal cell compartment where progenitor cells are believed to reside. In addition, Notch1 expression is downregulated in mature prostates but becomes up-regulated in certain primary and metastatic tumor cells in the TRAMP (transgenic adenocarcinoma of the mouse prostate) (7). These results suggest a possible role for Notch1 signaling during prostatic development and tumorigenesis.
Notch proteins are membrane-bound receptors. In mammals, four members including Notch1, -2, -3, and -4 have been identified, and all of them share substantial homology with their original ortholog Notch in Drosophila (8). Notch expression is associated with progenitor cell types and is developmentally regulated in a variety of tissues/organs, including the brain (9), thymus (10), blood vessel (11), skin (12), eye (13), and ear (6,14). Notch-mediated cell-cell interactions are well documented to be responsible for cell fate specification during embryogenesis (15). Aberrant activation or mutation of the Notch pathway not only leads to developmental disorders, including Alagille syndrome (16), but also neoplasia such as T-cell leukemia (17,18) or mammary (19) and salivary adenocarcinomas (20). Of the four family members, Notch1 is by far the most extensively studied.
In the present experiments, we hypothesize that Notch1expressing cells define the progenitor cells in the prostatic epithelium, and elimination of them would significantly impair prostatic branching morphogenesis and re-growth. To address this issue, we employed a transgenic approach to eliminate selectively the Notch1-expressing cells by targeted expression of a bacterial nitroreductase, which can convert 5-aziridinyl-2,4-dinitrobenzamide (prodrug CB1954) into a cytotoxin (21,22), under the control of the Notch1 promoter. We then evaluated the consequences of such selective cell ablation during prostate re-growth following castration and hormonal replacement in adult mice as well as prostatic epithelial differentiation and branching morphogenesis during early postnatal development. We found that ablation of Notch1-expressing cells profoundly prevented epithelial ductal outgrowth in the developing prostate and impaired prostatic re-growth in adults. In addition, we found that Notch1 expression was elevated following castration but returned to normal levels after androgen replacement. Consistent with this finding, Notch1 expression level in PrEC, the commonly studied human prostatic proliferating basal cells (23)(24)(25), was much higher than in benign prostatic hyperplasia 1 (BPH-1) cells (26) and prostate epithelial organoids that are predominantly composed of luminal epithelial cells (27). Therefore, these results strongly suggest that Notch1-expressing cells define prostatic epithelial progenitor cells during prostatic epithelial development and re-growth.

EXPERIMENTAL PROCEDURES
Vector Construction, Generation, and Genotyping of Transgenic Mice-The NotI/AgeI fragment containing the GFP gene was excised from the original pNotch1-GFP plasmid used for generation of Notch1-GFP transgenic mice (6) and replaced with the 2.3 kb NotI fragment of NTR-IRES-EGFP using blunt end ligation. For preparation of the NTR-IRES-EGFP fragment, the NTR gene and the IRES-EGFP fragment were ligated and inserted into the pCMV vector devoid of the NTR gene to generate pCMV-NTR-IRES-EGFP vector. Transgenic mice were generated by microinjecting the 14.7 kb fragment of pNotch1-NTR-IRES-EGFP plasmid linearized with XhoI and SapI into fertilized eggs of FVB mice. Two founder strains were obtained, and as expected from our previous work (5,6), they showed specific green fluorescence in the epithelial bud of freshly dissected postnatal prostates as well as in the expected region of the ear. The line that showed stronger fluorescence was chosen for further studies. We maintained these mouse lines by backcrossing transgenic mice with FVB. Mice were genotyped using PCR with primers specific to the EGFP sequence. The sequences of the primers are 5Ј-ATGGTGAGCAAGGGCGAGGA-3Ј and 5Ј-ACGAACTC-CAGCAGGACCA-3Ј. All animal experiments were approved by the Animal Care Committee at Genentech.
Conventional RT-PCR-Total RNA was prepared from prostates of mice using Qiagen RNeasy mini kit, and the RT-PCR was performed using Qiagen SensiScript reverse transcriptase and PerkinElmer Life Sciences AmpliTaq DNA polymerase. The reverse transcription was primed with random hexamers. The primers for PCR amplification of nitroreductase were 5Ј-ACGCTACTGCAATACAGCCCAT-3Ј and 5Ј-AACCAGCTTCAGCCAGACATC-3Ј, and the primers for GAPDH were 5Ј-CCCACTAACATCAAATGGGG-3Ј and 5Ј-CCTTCCACAATGCCAA-AGTT-3Ј.
TaqMan Real Time Quantitative RT-PCR-Panels of normalized first strand cDNA synthesized from poly(A) ϩ RNA (Clontech, Palo Alto, CA) were used to measure quantitatively the expression of the Notch1 gene in a variety of mouse tissues. Notch1-specific probes and primers were used (probe, 5Ј-CTTGGCTGCCCGATACTCTCGTTCAGA-3Ј; forward primer, 5Ј-ACCAGACAGACCGCACCG-3Ј; reverse primer, 5Ј-ACGGAGTACGGCCCATGTT-3Ј). Housekeeping gene GAPDH was used for internal control. The sequences of the probe and two primers are as follows: 5Ј-TTCCTACCCCCAATGTGTCCGTCGT-3Ј, 5Ј-ACTG-GCATGGCCTTCCG-3Ј, and 5Ј-CAGGCGGCACGTCAGATC-3Ј, respectively. For evaluation of Notch1 expression in human samples, total RNA was prepared from tissue or cells using RNeasy Kit (Qiagen, Valencia, CA), and the reagents used were described previously (5). The PCR products were checked on agarose gel to ensure specific amplification. The Notch1 expression level in each sample was normalized to the GAPDH level in the same sample using the ⌬Ct method. Data in triplicate experiments were calculated and presented as mean Ϯ S.D.
Prodrug Preparation and Injection-The prodrug CB1954, i.e. 5-aziridinyl-2,4-dinitrobenzamide, was synthesized in the Bioorganic Chemistry Department at Genentech, according to published protocols (28). The prodrug was dissolved in dimethyl sulfoxide as stock solution at 0.2 M. Phosphate-buffered saline (PBS) was used to dilute the stock solution to desired concentrations. For evaluation of apoptosis in various organs, prodrug at 50 mg/kg body weight was injected intraperitoneally daily for 5 days into 12-14-week-old mice. On the 6th day, the animals were euthanized with carbon dioxide, and various organs were harvested, and portions of these organs were fixed in 10% formalin for hematoxylin and eosin (H&E) staining or 4% paraformaldehyde for immunohistochemistry analyses.
Dissection and Culture of Prostate Tissue-The prostate whole mounts consisting of all lobes were dissected from postnatal mice and cut in two along the middle line, and each piece was placed individually onto cell culture inserts (8-m pore size; BD Biosciences) in serum-free medium as described (29). Serum-free medium was Dulbecco's modified Eagle's medium/F-12 with serum-free supplement (I-1884; Sigma), 2 mM glutamine, 5 mg/ml glucose, 100 units/ml penicillin, 100 mg/ml streptomycin, 1 ϫ 10 Ϫ8 M testosterone (Sigma), and 1:1000 epidermal growth factor (Clonetics). Unless otherwise noted, the medium was changed every other day; the prodrug concentration for treatment was 62.5 M, and BrdUrd was always introduced into the medium 16 h prior to tissue fixation. Images of the cultures were taken under a Nikon TE300 inverted microscope using Compix imaging systems with a cooled RGB CCD camera and analyzed with Photoshop 7.0. For BrdUrd pulse-labeling experiments, the prostate cultures (P9) were treated with 62.5 M prodrug for 24 h first and then refed with medium containing BrdUrd. Two hours later, the cultures were rinsed in serumfree medium for 30 min and then either fixed or maintained for an additional 2 days before fixation with 4% paraformaldehyde.
Prostate Re-growth Following Hormone Replacement-Transgenic male mice as well as their wild type male littermates at the age of 12-14 weeks old were first castrated. Starting on day 9 after castration, all mice were injected with 50 mg/kg prodrug daily for 5 consecutive days. On day 14 after castration, some mice in each group were sacrificed, and prostates were harvested for evaluation. Other mice were implanted with testosterone pellets (15 mg/pellet/mouse, Innovative Research, Sarasota, FL). On day 17, i.e. 3 days after hormone replacement, mice were sacrificed, and their prostates were harvested. For this time point, 16 h before harvesting, three injections of BrdUrd at 50 g/g body weight were administered with 3 h intervals inbetween. The weight of whole prostates harvested at each time point was measured with saline solutions completely absorbed by Kimwipes. Prostates were further analyzed by immunohistochemistry.
Histology, Immunohistochemistry, Image Acquisition, and Cell Counting-For general histologic analysis, tissues were collected and fixed in 10% neutral buffered formalin, then embedded in paraffin, sectioned at 4 m, and stained with hematoxylin and eosin. For histochemistry, prostate whole mount tissues freshly dissected from mice or grown in culture were fixed with 4% paraformaldehyde in 0.1 M sodium phosphate, pH 7.4, at room temperature for 30 -60 min. The tissues were then rinsed with PBS and stored in 25% sucrose in PBS at 4°C. Cryosections of 5-20 m thickness were cut using OCT as the embedding reagent. Immunostaining with anti-EGFP (chicken, 1:100; Chemicon, Temecula, CA), anti-cytokeratin 14 (rabbit, 1:10,000; Babco, Berkeley, CA), anti-cytokeratin 8 (sheep, 1:150; Pickcell, Amsterdam, Netherlands), anti-caspase 3 (rabbit, 1:1000; R&D Systems, Minneapolis, MN), or anti-BrdUrd (mouse, 1:30; BD Biosciences, Palo Alto, CA) were detected with Texas Red-and/or fluorescein isothiocyanate-conjugated secondary antibodies. For BrdUrd and cytokeratin 14 double staining, sections were pre-treated with 2 N HCl for 40 min at room temperature before incubation with the primary antibodies. An in situ Cell Death Detection kit (TUNEL) (Roche Applied Science) was also used to detect apoptosis. For double staining of TUNEL and cytokeratin 14, TUNEL was performed after the immunostaining, including the washing of secondary antibody, was finished. Staining of whole mount tissue was performed similarly except that primary antibodies were incubated for 2 days at 4°C with shaking. Slides were mounted in Fluoromount-G (Southern Biotechnology) supplemented with counter-staining dye 4,6-diamidino-2phenylindole (DAPI) (Sigma) and observed under a Zeiss Axiophot epifluorescence microscope. Unless otherwise noted, images were captured with Compix imaging systems using a cooled RGB CCD camera and analyzed using Adobe PhotoShop 7.0.
Cell counting was performed on the digital images captured from the slides. DAPI staining was used to obtain the total number of cells in a given field. Cytokeratin 14-positive cells were counted as basal cells, and the rest of cells in the epithelium were luminal cells. At least five randomly selected regions such as epithelial ducts from different sections were counted for each group, and a two-way, unpaired t test was used for statistical analyses where necessary.

Establishment of Notch1-NTR-IRES-EGFP Transgenic
Mouse Line-To eliminate selectively Notch1-expressing cells, we expressed the bacterial nitroreductase gene (NTR) (28,30) under the control of the Notch1 promoter. This promoter had been used for generation of Notch1-GFP transgenic mice and had been shown to be able to drive GFP expression faithfully as demonstrated by Notch1 in situ hybridization and GFP labeling in parallel E14 sections (6). Cells that express nitroreductase can be selectively induced to undergo apoptosis in the presence of the prodrug CB1954, i.e. 5-aziridinyl-2,4-dinitrobenzamide (22). We also inserted the IRES-EGFP cassette following the NTR gene to facilitate visualization of Notch1expressing cells that are undetectable due to lack of suitable anti-Notch1 antibodies and/or relatively low expression levels of Notch1 protein in the prostate. A schematic map of the construct is shown in Fig. 1A. Routine genotyping of mice was performed using the EGFP sequence as an amplicon in PCR (Fig. 1B). The results of using EGFP or NTR sequences for PCR genotyping are 100% concordant (data not shown). Of the 228 mice genotyped, 110 (48.2%) were positive for EGFP. We further verified the expression of both NTR and EGFP genes in transgenic animals. As shown in Fig. 1C, NTR mRNA was detected in the prostates of transgenic mice but not in those of wild type mice by RT-PCR. To verify the expression of the transgene EGFP, we dissected prostates from male neonates (P2-P5) and examined a piece of the prostate tissue mounted onto slides under a fluorescence microscope. Green fluorescence could be selectively detected in the epithelial region of the freshly dissected, live prostate tissue but not in the stromal area of the transgenic mice tissue or in wild type tissue (data not shown), as demonstrated in the Notch1-GFP line (5). Green fluorescence became undetectable following fixation. To verify the EGFP expression pattern at the cellular level, we used an anti-EGFP antibody to enhance the signal (see "Experimental Procedures"). As shown in Fig. 1D1, all the epithelial cells in the epithelial bud but not stromal cells of P3 prostate were labeled by the EGFP antibody, and at this stage, they all expressed CK14 (Fig. 1D2). At stage P10, when luminal cell differentiation is prominent (31), EGFP expression became restricted to cells in the basal cell compartment, which are CK14positive ( Fig. 1, D3 and D4). This EGFP expression pattern is consistent with the Notch1 expression pattern, which has been determined to be associated with CK14-positive cells (5).
Specific Cell Targeting in Notch1-NTR-IRES-EGFP Transgenic Mice Upon Prodrug Treatment-To determine whether the expression of bacterial nitroreductase could cause destruction of cells by treatment with prodrug, we injected the prodrug into adult transgenic and wild type animals. Following 5 days of daily injections, individual apoptotic cells, as revealed by the condensation and degradation of the nuclei, were evident in the epithelial lining of the prostatic epithelium in the transgenic mice (Fig. 2B, arrow) but not in the corresponding regions of wild type mice ( Fig. 2A). To verify that apoptosis occurred in the targeted compartment, i.e. the basal cell compartment, we performed double staining of TUNEL and the basal cell marker cytokeratin 14. As shown in Fig. 2D1, the majority of apoptosis occurred in CK14-positive cells. Cell counts from randomly selected areas revealed that ϳ11% of basal cells were TUNELpositive. In sharp contrast, wild type mice treated with the prodrug had no apparent cytotoxicity (Fig. 2C1), and the prostates from these mice were histologically normal (Fig. 2, A and C2). The apoptosis in the transgenic prostatic epithelium was also confirmed with immunostaining for the active form of caspase 3, an effector caspase for apoptosis (data not shown).
To provide direct evidence that apoptosis occurs in Notch1expressing cells (EGFP-positive cells), we performed EGFP/ caspase-3 double immunostaining with cultured prostates prepared from postnatal day 3 (P3) and P8 mice where the EGFP expression is still detectable. Whereas P3 represents a predifferentiation stage of the prostatic epithelium where no luminal cells are formed, P8 prostatic epithelium is composed of some luminal cells (31). In P3 prostate, apoptosis was specifically observed in EGFP-positive cells, and localized to the epithelium but not the stroma (Fig. 3, A1-3). At P8, apoptosis was mainly observed in EGFP-positive cells that are in the basal layer (arrows, Fig. 3, B1-3) or in the progenitor cells in the distal duct tip where the epithelial lumen has not been formed (arrowheads, Fig. 3B1-3). Thus, the treatment with prodrug selectively kills Notch1-expressing cells in the basal cell compartment of the prostate epithelium in the transgenic mice.
The Extent of Prodrug-induced Apoptosis in Various Organs Correlates with Notch1 Expression Levels-Since Notch1 is also expressed in other tissues, we examined whether the prodrug treatment also leads to ablation of cells in these tissues. We collected and analyzed the thymus, brain, heart, kidney, liver, mammary gland, skin, small intestine, large intestine, and urinary bladder from both wild type and transgenic mice following prodrug injection. H&E staining revealed that although there was substantial apoptosis in the thymus, the effect on other organs was less evident. As shown in Fig. 4, B1 and B2, the thymus of transgenic mice showed diffuse atrophy with widespread apoptosis and necrosis of lymphocytes. In contrast, the kidney showed no obvious lesion (Fig. 4D). At high magnification, cells in the medulla of the thymus, predominantly comprised of progenitor thymocytes, showed more profound apoptosis (Fig. 4B2) than those in the cortical area where cells are more differentiated (Fig. 4B1). No apparent lesion was observed in the thymus and kidney from wild type mice treated with the prodrug (Fig. 4, A1 and A2, and C).
To examine whether the level of apoptosis seen in H&E staining correlates with the expression level of Notch1 in each of the organs examined, we performed a quantitative Taqman real time RT-PCR to compare the relative levels of Notch1 transcript in various organs prepared from adult mice. As shown in Fig. 4E, of all organs examined, the thymus had the highest expression level, and prostate exhibited a relatively high level. The expression level of Notch1 in the kidney was among the lowest. The extent of apoptosis correlated with the level of Notch1 expression across the tissues examined. The expression level in testis was relatively high as well, but because it was not included among the organs for histological examination, the level of apoptosis in it was not clear. It is noted that although expression of Notch1 in developing brain is high (32), Notch1 expression is down-regulated in the mature brain. Similar down-regulation of Notch1 expression in adult tissue is also reported for the prostate (5).
Notch1-expressing Cells Are Necessary for Prostatic Epithelial Development-In rodents, the prostate is mainly formed during early postnatal stage via epithelial budding and expansion. Initially, all epithelial cells are proliferative. Only around postnatal day 5, some epithelial cells become post-mitotic and terminally differentiate into luminal cells, but the remaining basal epithelial cells retain their progenitor properties (31). Notch1 is expressed in all epithelial cells in the prostatic epithelium before P5 but becomes associated with only the basal cells and not the luminal cells at later stages of development (5). We hypothesize that Notch1 expression defines the epithe- lial progenitor cells, and ablation of Notch1-expressing cells before P5 would block prostate branching morphogenesis, and destruction of Notch1 expression cells at stages after P5 would impair differentiation of new luminal cells.
In the first group of experiments, prostate whole mounts were dissected from P4 mice and plated in culture in the presence or absence of the prodrug. As shown in Fig. 5A, incubation with the prodrug (62.5 M) completely blocked branching morphogenesis of transgenic tissue but did not cause any detectable effect in wild type tissue or transgenic tissue maintained in the absence of the prodrug. Staining of the cultured whole mount prostates with a combination of CK8 and CK14 antibodies after prodrug treatment for 2 and 5 days revealed a severe impairment of epithelial budding process in transgenic tissues (Fig. 5B), suggesting that treatment of prodrug at an early stage damaged the majority of undifferentiated progenitor cells and prevented the formation of epithelial buds. Consistent with this notion, TUNEL assay revealed significant apoptosis after 1 day of prodrug treatment in transgenic tissue (Fig. 6, A1 and A2) but not in wild type or mocktreated transgenic controls (data not shown). The apoptosis observed was mainly in CK14-positive epithelial population (Fig. 6A2). A small fraction of apoptotic cells were observed in adjacent areas, which might be the result of the alterations in their interaction with the ablated cells. In agreement with these observations, examination of cell proliferation using Br-dUrd immunocytochemistry showed that while wild type tissue displayed robust BrdUrd incorporation, the transgenic tissue showed minimal incorporation. The majority of proliferation occurred in epithelium (see also Ref. 29). Whereas the wild type tissue showed 49 Ϯ 8% (n ϭ 5) of epithelial cells incorporated BrdUrd, few cells in the transgenic epithelium were BrdUrdpositive (Fig. 6, B and C).
BrdUrd pulse-chase labeling was used to evaluate the role of Notch1-expressing cells in prostate growth and differentiation. For this purpose, P9 mouse prostate tissue was used because at this stage some luminal cells have been formed and new luminal cells continue to differentiate from progenitor cells. As expected, treatment of transgenic tissue with the prodrug caused specific killing of cells in the basal layer, but not preexisting luminal cells as indicated by the greatly reduced number of CK14-positive cells and intact CK8-positive cells compared with wild type control tissue (Fig. 7, B and A). BrdUrd pulse-labeling coupled with CK14 immunocytochemistry revealed that the majority of proliferating cells are in the basal layer (Fig. 7C), and elimination of these cells in transgenic tissue abolished incorporation of BrdUrd (Fig. 7D), suggesting that Notch1-expressing basal cells are the proliferative cells in the prostate epithelium. When the wild type cultures were maintained for an additional 2 days after BrdUrd pulse labeling, virtually all of the BrdUrd-positive cells lost expression of CK14 and migrated to the luminal region of the epithelium (Fig. 7E), indicating formation of newly differentiated luminal cells. In contrast, transgenic tissue treated with prodrug showed minimal BrdUrd incorporation and lacked differentiation of new luminal cells (Fig. 7F). Therefore, ablation of Notch1-expressing cells impairs prostatic differentiation.
Ablation of Notch1-expressing Cells Impairs Prostatic Regrowth in Castrated Mice Following Testosterone Replacement-An important feature of the prostate is that androgen is required for the maintenance of prostatic structure and function. Traumatic damage to the testis or castration causes prostatic involution and a decrease in its secretion function. However, subsequent administration of exogenous testosterone induces re-growth, differentiation, and resumption of the structure and function to its original size or level (33). Based on these findings, it is assumed that there are progenitor cells residing in the epithelium, which retain the capacity to repopulate the lost epithelial cells. However, the nature of these progenitor cells is uncharacterized. We hypothesized that Notch1-expressing cells define this progenitor population that plays an important role in the prostatic re-growth process.
To test this hypothesis, we followed the prostatic re-growth process of transgenic mice that had been castrated and treated with prodrug, and we compared it with that of wild type mice undergoing the same procedure. In agreement with the literature (33), 14 days after castration, wild type mice showed dramatic shrinkage of the prostate as compared with uncastrated normal mice of the same age. Transgenic mice also showed prostatic shrinkage following castration. Grossly, the wet weight of prostate in both groups was similar, averaging ϳ34 mg. TUNEL assay and immunostaining with anti-active caspase 3 in both wild type and transgenic prostate revealed prominent apoptosis in the luminal cell compartment and showed only small numbers of apoptotic cells in the basal cell compartment and the stroma (data not shown). The ratio of luminal cells and basal cells at this time point was about 1:1 (data not shown). Consistent with the results shown in Fig. 2, apoptosis in the basal cell compartment of transgenic tissue was elevated as compared with that in wild type tissue (data not shown). Following hormone replacement, transgenic mice

Notch1-expressing Cells in Prostate Development and Re-growth
exhibited impaired prostate re-growth. On day 17, i.e. 3 days after replacement, the mean prostate weight of transgenic mice (55.9 Ϯ 3.3 mg, n ϭ 3) was significantly lower than that of the control group (91.6 Ϯ 6.1 mg, n ϭ 3). Immunostaining for CK14 in prostatic sections showed that while in wild type tissue, basal cells were separated by 1-2 luminal cells (Fig. 8, A1 and  A2), the basal cells in the transgenic tissue frequently remained contiguous or much more reminiscent of the histology before hormone replacement (Fig. 8, A3 and A4). The ratio of luminal cells to basal cells in transgenic prostate epithelium (1.8 Ϯ 0.2:1, n ϭ 8) was Ͼ2-fold less than that in wild type tissue (4.3 Ϯ 0.7:1, n ϭ 8) (Fig. 8A5). BrdUrd incorporation was also compared at this time point as an index for the number of progenitor cells that were undergoing proliferation. As shown in Fig. 8B3, remarkably lower incorporation of BrdUrd was detected in the transgenic prostate epithelium than in the wild type tissue (Fig. 8B1). In contrast to the high percentage of cells (25.9 Ϯ 5.1%, n ϭ 8) incorporating BrdUrd in the wild type tissue, only 11.5 Ϯ 3.3% (n ϭ 8) of cells in the transgenic tissue were proliferating (Fig. 8B5). Based on their laminar position and morphology, the majority of cells that were BrdUrd positive did not appear to be luminal cells, in agreement with the

Notch1-expressing Cells in Prostate Development and Re-growth
concept that luminal cells are terminally differentiated and non-dividing cells. In addition, the intensity of BrdUrd staining in transgenic prostate was much weaker than that in wild type (Fig. 8, B1 and B3).
Notch1 Expression Is Elevated Following Castration and High in PrEC Cells-Based on the above findings, we predicted that Notch1 expression pattern would match those of other basal cell markers that define the compartment of progenitor cells. In a separate study in which we followed prostate gene expression profiles of normal adult mice during the course of castration and hormonal replacement processes by using the microarray technique (Fig. 9A), 2 we found indeed that the expression pattern of Notch1 highly correlated with those of p63 and cytokeratin 14, two basal cell markers (see review Ref. 1). During the entire castration and hormonal replacement course, there was a peak elevation at 14 days following castration and a resumption to nearly normal levels after 3 days of hormone replacement. In contrast, expression levels of Nkx3.1, a luminal cell marker, showed an opposite pattern (Fig. 9A).

FIG. 5. Blockade of branching morphogenesis of cultured P4 transgenic prostates in the presence of prodrug.
A, representative images of cultured prostate at different time points from triplicate experiments. Although either the wild type (WT) tissue maintained in the presence of prodrug (left column) or the mock-treated transgenic (TG) prostate tissues (right column) shows normal branching morphogenesis, transgenic tissue treated with prodrug showed a completely blocked branching morphogenesis (middle column). The final concentration of prodrug was 62.5 M, and prodrug was added on day 0 (d0) when the prostates were freshly dissected and plated in culture dishes. The cultures were refed with medium containing prodrug on day 3 (d3). Experiments with higher concentrations of the prodrug (125, 250 M) yielded similar results without showing toxicity in prodrug-treated wild type tissue (not shown). B, epithelial staining of whole mount prostate after prodrug treatment using a combination of CK8 and CK14 antibodies. Note that while wild type tissue showed normal branching morphogenesis, the epithelial budding in transgenic tissue was blocked. Scale bar, 500 m.

Notch1-expressing Cells in Prostate Development and Re-growth
Although the change in Nkx3.1 expression may also be attributed to possible influence of androgen on Nkx3.1 transcription (34) in addition to the alteration of luminal cell population, the drastic difference between Notch1 and Nkx3.1 expression strengthens the notion that Notch1 is associated with basal cells but not luminal cells.
Similarly, we also inferred that Notch1 expression would be higher in prostate epithelial cells that have progenitor phenotypes. Because mouse epithelial cells with this property were not available, we determined Notch1 expression levels in human prostate epithelial cells (PrEC) (23,24), and we compared it with the levels in human prostate epithelial organoid, which is composed of mainly luminal epithelial cells and some parenchymal cells (27), and in the BPH-1 cells (26) using quantitative Taqman real time RT-PCR. PrEC have been extensively used in prostate research and are generally believed to represent proliferating basal cell populations of human adult prostate, as these cells are derived from mature prostatic epithelium and are maintained in the absence of androgen, which favors the selection of proliferating basal cells (23,35). As shown in Fig. 9B, the Notch1 expression level in PrEC cells was 5.3-and 4.1-fold higher than the epithelial organoid and BPH-1 cells, respectively. This result agrees with a recent report (25) showing expression of Notch1 in a less differentiated state and loss of Notch1 expression in a more differentiated state of PrEC cells. DISCUSSION By taking advantage of the nitroreductase activity, we have successfully established a transgenic mouse line in which Notch1-expressing cells can be selectively ablated. By using this strategy we found that Notch1-expressing cells may define the progenitor cells in the prostatic epithelium. These cells are necessary for the epithelial branching morphogenesis during prostatic development and are important contributors for prostatic re-growth following castration and androgen replacement.
Cell ablation is an important technique that has been elegantly used in cell lineage and cell-cell interaction studies in the developing nervous system. Initially, cell ablation was achieved via laser beam, cytotoxin injection, or mechanical surgery in the invertebrate nervous system. These early studies clearly demonstrated how the presence or absence of a given cell affects cell fate and growth patterns of adjacent cells (36 -38). More recently, cell death genes such as ICE and ced-3 have been employed under the control of specific promoters to effectively and specifically kill target cells but not adjacent cells (39). To achieve cell ablation at specific time points, sophisticated systems such as herpes simplex virus 1-thymidine kinase (40,41), diphtheria toxin, and its receptor approaches (42,43) have been designed such that cell ablation can happen only when the substrates or ligands are administered. However, these systems exhibited certain levels of non-specificity or could potentially induce secondary effects. The approach we used in the present experiments is a combination of the bacterial nitroreductase and a specific Notch1 promoter. With this approach, we are not only able to target a cytotoxin to a specific cell population but also able to eliminate cells at various developmental time points by using an enzymatic activity that is exogenous to the host. Interestingly, this tissue specificity and the ability to control timing of administration have resulted in consideration of the prodrug strategy for gene therapy (44,45). A1-A3, TUNEL/CK14/DAPI triple staining of a cultured P4 transgenic (TG) prostate treated with the prodrug for 1 day. Note that apoptosis revealed by TUNEL (A1) mainly occurred in CK14-positive cells (A2) and that apoptosis in wild type tissue was negligible (not shown). A3, shown is the counterstaining with DAPI of the slide shown in A1 and A2. B and C, BrdUrd/CK14 double immunostaining of wild type (B1 and B2) and transgenic (C1 and C2) prostate, respectively. The tissue was treated with prodrug for 1 day, rinsed with PBS, and exposed to the medium containing BrdUrd for an additional 16 h before fixation. Note that while robust epithelial proliferation, which was colocalized with CK14 staining, was revealed by BrdUrd immunostaining in wild type tissue (B1 and B2), minimal proliferation was seen in transgenic tissues (C1 and C2). Scale bar for A1-A3 shown in A3, 10 m; for B and C shown in C2, 20 m.
In addition, we demonstrated that specific organs such as prostate can be dissected and placed in culture in the presence of the prodrug to study the role of Notch1-expressing cells during prostatic development. This model may also facilitate studies on other Notch1-expressing organs such as the thymus and developing nervous system.
The present experiments reinforce the idea that Notch-expressing cells are important for cell growth and differentiation. First, at an early developmental stage, such as P3, Notch1 expression is associated with all prostatic epithelial cells that are proliferating or act as progenitor cells. Elimination of the Notch1-expressing cells at this stage completely blocks the epithelial branching morphogenesis. Second, at later stages when cell differentiation becomes prominent (e.g. P10), Notch1 expression is localized to basal cell compartment where prostatic progenitor cells are believed to reside. Our data showed that when Notch1-expressing cells are destroyed at this later stage, the number of cells in the basal layer decreased, and differentiation of new luminal epithelial cells was greatly inhibited as indicated by the BrdUrd pulse labeling experiments. Third, Notch1 expression is down-regulated in mature prostates, and this corresponds to the decreased proliferation in the epithelium at this stage.
Although previous studies have documented the prostatic re-growth process, the progenitor cells within the epithelium responsible for the re-growth are uncharacterized. During prostate regression following castration, massive apoptosis occurred in the luminal layer of the epithelium, resulting in a dramatic decrease in the ratio of luminal cells to the basal cells from ϳ10:1 to about 1:1 at the time point of 2 weeks after hormone deprivation (33). Following hormone replacement, robust cell proliferation happens in the epithelium, and the ratio of luminal cells to basal cells resumes. Our work that prodrug treatment significantly affected the prostate re-growth suggests that Notch1 expression may define the progenitor cells that undergo proliferation in this process. This model is supported by the correlation of Notch1 gene expression patterns and the dynamics of the basal cell population in the entire process following hormone ablation and replacement. Our microarray experiments showed that castration leads to an increased level of Notch1 in the prostate, and androgen replacement reversed the change. The pattern of Notch1 expression matched very well with other known markers of progenitor cells, i.e. p63 and CK14. In addition, the data that PrEC cells, which represent the proliferating basal cells of mature prostate epithelium, expressed much higher levels of Notch1 than prostatic samples that are mainly composed of luminal cells, provided additional support for the notion that Notch1 expression is associated with the progenitor cells that contribute to the re-growth process.
It should be pointed out that although our study links Notch1 expression with progenitor cells, it does not differentiate the so-called stem cell and transit-amplifying cell populations in the prostate (46,47). Recent work suggests that there are possibly two types of proliferative cells in the prostatic epithelium: the strict-sense stem cells that can give rise to all cell types in the epithelium and so-called intermediate or transit-amplifying cells that give rise to specific types of cells (reviewed in Ref. 48). The idea that there are probably stem cells in the prostate has been mainly based on the observation that the prostate has the capacity of self-renewal as it can undergo regression and re-growth process caused by castration and hormonal replacement for as many as 30 cycles (4,49). However, the nature of this population in vivo still remains elusive. On one hand, the stem cell population is generally believed to localize in the basal cell layer (48). On the other hand, recent studies suggest that, in addition to expression of basal cell markers cytokeratin 5, 14, and p63, these cells may also express luminal cell markers cytokeratin 8 and 18, as wells as other markers such as GST, ␣ 2 ␤ 1 integrin (50), or prostate stem cell antigen (25,51). Moreover, it has been proposed that prostate stem cells are not equally distributed in different regions of the prostate. The proximal region of prostate ducts is reported to contain higher density of cells that resemble stem Notch1-expressing Cells in Prostate Development and Re-growth cell features than the distal regions (46). Addition of more complexity to this issue is that more cell proliferation is observed in the distal region than the proximal region of developing prostatic budding epithelium (52). Nevertheless, progress has been made toward maintenance and propagation of prostatic proliferative epithelial cells in vitro (23,53), which leads to development of PrEC that possesses progenitor cell phenotypes and is available commercially. Additional experiments suggest that primary prostate epithelial cells grown in culture are able to form two types of colonies, which may very likely represent two stages of epithelial cells during differentiation (53). Using a label-retaining strategy, Tsujimura et al. (46) were able to differentiate slow and fast cycling cells in the prostate, which may represent the stem cell and transit-amplifying cell populations. Our data showing the important role of Notch1-expressing cells during prostate growth and re-growth and the association of Notch1 expression with the progenitor cell population in vivo and with PrEC in vitro suggest that Notch1 may be used as a marker for progenitor cells in the prostate, which mimics Notch1 expression in neuronal progenitor cells in the developing neuroepithelium (15,54). Generation of high affinity anti-Notch1 antibodies that can detect low levels of Notch1 protein in tissues could facilitate research of prostatic progenitor cells and their relationship with so-called prostatic stem cells. In addition, generation of a prostate-specific Notch1 gene inactivation model will help in understanding the exact role that Notch1 plays during prostatic development.
There is increasing evidence that alterations in the function of the Notch signaling pathway in vertebrates can lead to cancer and developmental abnormalities (for reviews see Refs. 55 and 56). Our previous work showed that Notch1 expression is up-regulated in certain types of prostatic malignant cells (5), but whether Notch signaling is directly involved in prostatic tumorigenesis remains to be determined. Several mouse models for prostate cancer have been established (57), such as the transgenic adenocarcinoma of the mouse prostate (7), PTEN/ Nkx3.1 compound mutant mouse (58,59), and those reported more recently, including prostate-specific PTEN knock-out model (60) and Myc-driven transgenic model (61). For further determination on how Notch1-expressing cells play a role during prostate tumorigenesis, these prostatic cancer animal models can be crossed with our transgenic line to generate a compound model, and the prodrug can be injected into the compound model either before tumor formation to determine the importance of progenitor cells during tumor initiation or during tumor growth to determine the consequence of elimination of Notch1-expressing cells in tumor progression. Grafting of transgenic/knockout prostate tissue into the renal capsule of wild type animals may also be considered for use to minimize the complication of wide expression of Notch1. Future studies in this direction would advance our understanding on the role of Notch1-expressing cells in prostatic tumorigenesis. FIG. 9. High expression level of Notch1 in proliferating prostate epithelial cells. A, Notch1 expression profile data obtained from a separate study using microarray to understand the process following castration and hormone replacement in mice. N, normal adult mice. C3, 3 days after castration. C14, 14 days after castration. C14ϩT3, 3 days after hormone replacement performed on the 14th day after castration. The intensity of fluorescence on microarray from 5 samples in each group was used to calculate the relative expression of each gene. The data have been normalized to the levels in normal controls for comparison. B, Taqman real time RT-PCR is performed for comparison of Notch1 mRNA level in human samples. BPH-1, an immortalized cell line derived from benign prostate hyperplasia. Note that expression of Notch1 in mice tightly correlates with basal cell markers p63 and CK14 but not luminal cell marker Nkx3.1 in the process following castration and hormonal replacement (A) and that PrEC human progenitor epithelial cells express a much higher level of Notch1 than more differentiated prostate samples (B).