Oncogene (2002) 21, 4099 ± 4107 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Stabilization of b-catenin induces lesions reminiscent of prostatic intraepithelial neoplasia, but terminal squamous transdi€erentiation of other secretory epithelia

Fotini Gounari1, Sabina Signoretti3,4, Roderick Bronson2, Ludger Klein1, William R Sellers3,4, Jennifer Kum3,4, Anja Siermann1, Makoto M Taketo5, Harald von Boehmer1 and Khashayarsha Khazaie*,1

1Department of Cancer Immunology and Aids, Dana Farber Cancer Institute/Harvard Medical School, Boston, Massachusetts, MA 02115, USA; 2Histo-Pathology Core, Dana Farber Cancer Institute/Harvard Medical School, Boston, Massachusetts, MA 02115, USA; 3Department of Adult Oncology, Dana Farber Cancer Institute/Harvard Medical School, Boston, Massachusetts, MA 02115, USA; 4Department of Pathology, Dana Farber Cancer Institute/Harvard Medical School, Boston, Massachusetts, MA 02115, USA; 5Department of Pharmacology, Kyoto University Graduate School of Medicine, Sakyo, Kyoto 606-8501, Japan

The present study documents that stabilization of b- Introduction is sucient to induce lesions reminiscent of prostate intraepithelial neoplasia (PIN). Such lesions Stabilization b-catenin is associated with the genesis of were present in all compound mutant mice and all colorectal neoplasms (Harada et al., 1999; Morin et al., prostate acini expressing stabilized b-catenin. High 1997; Sparks et al., 1998), hair follicle tumors (Gat et grade PIN-like lesions resembling early human prostate al., 1998), and aberrant di€erentiation of T lympho- cancer were detected as early as 10 weeks of age. cytes (Gounari et al., 2001). Furthermore, activation of Surprisingly, stabilization of b-catenin in other secretory Wnt is associated with mammary carcinoma in epithelia including salivary, preputial, harderian, and mice (for review see Li et al., 2000). Wnt signaling mammary glands induced extensive squamous metaplasia ultimately leads to the stabilization of b-catenin, by and keratinization associated with terminal di€erentia- preventing its phosphorylation at critical exon 3 tion of the target cells, but failed to cause neoplastic encoded serine and threonine residues and its subse- transformation. Epidermal hyperplasia, hair follicle quent degradation (for reviews see Bienz and Clevers, cysts, and odontomas were also observed. The prostatic 2000; Lau and Clevers, 2001). It is possible that lesions exhibited upregulation of c-myc, increased rate of aberrant b-catenin signaling may contribute to cancer cellular proliferation, loss of the Na-K-Cl co-transporter in a wide range of tissues (Reya et al., 2001). NKCC1, and expression of androgen . Basal cell There is little published data implicating the b- markers such as p63 and 5 were not expressed by catenin pathway in the etiology of prostate cancer. the masses of PIN-like lesions, but were present in small Circumstantial evidence from studies on subsets of foci of proliferating b-catenin expressing basal cells. Our human prostate cancers documents the overexpression observations indicate that b-catenin stabilization is a of the Wnt-5A (Iozzo et al., 1995), APC crucial event for the initiation of PIN-like lesions, but (Brewster et al., 1994; Phillips et al., 1994; induces squamous metaplasia rather than tumorigenesis Suzuki et al., 1994; Watanabe et al., 1996), and in secretory epithelia other than the prostate. mutations a€ecting the stability of b-catenin (Chesire Oncogene (2002) 21, 4099 ± 4107. doi:10.1038/sj.onc. et al., 2000; Voeller et al., 1998) in tumors. of 1205562 critical sequences of the b-catenin third exon was reported in each of nine separate metastases examined Keywords: b-catenin; prostate; neoplasia; metaplasia from a single patient, indicating that this may represent an early and perhaps clonal event (Chesire et al., 2000). While these observations point to the involvement of b-catenin, their almost anecdotal frequency and their retrospective nature leave the signi®cance of the genetic alterations and their relevance to prostate cancer unclear. To investigate the potential of b-catenin to cause epithelial tumors, we have taken advantage of *Correspondence: K Khazaie, Department of Cancer Immunology Catnb+/l8x(ex3) (Harada et al., 1999) and MMTV-Cre and AIDS, Dana Farber Cancer Institute/Harvard Medical School, (Wagner et al., 1997) mice. Catnb+/l8x(ex3) mice have the 44 Binney Street, Boston, MA 02115, USA; 3rd exon of the b-catenin gene ¯anked by loxP E-mail: [email protected] Received 21 January 2002; revised 13 March 2002; accepted 11 sequences, and upon Cre mediated recombination April 2002 express a dominant stable form of b-catenin. In the Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4100 compound mutant progeny, targeted recombination was promoted in a range of tissues, including the prostate. Our observations demonstrate that the recombination of a single b-catenin allele results in sucient stabilization of the to induce prostate intraepithelial neoplasia (PIN)-like lesions. In agree- ment with an earlier report focusing on the mammary gland (Miyoshi et al., 2001), in other tissues this event induced almost exclusively squamous metaplasia and keratinization rather than neoplastic lesions. These observations are in line with a tissue speci®c mode of action of b-catenin, highlighting the prostate as a target of neoplastic transformation by this signaling pathway.

Results

Targeted excision of b-catenin exon-3 in the mouse prostate To examine the tissue distribution of Cre, adult (MMTV ± Cre) mice (Wagner et al., 1997) were crossed to (R26R) lacZ reporter mice (Soriano, 1999). Cryosections from individual organs of 3-month-old compound mutant male (MMTV ± Cre6R26R) mice were stained for b-galactosidase enzymatic activity. Speci®c mosaic staining of lacZ was detected in the skin, salivary, preputial, harderian, prostate and mammary glands (Figure 1A, a ± f). Other organs such as the intestine, stomach, and liver did not show speci®c lacZ staining. Gut speci®c expression of a dominant stable b- catenin has been reported to cause rapid intestinal adenomatous polyps (Harada et al., 1999). To promote expression of this stabilized form of b-catenin in various tissues, Catnb+/l8x(ex3) mice were crossed with MMTV ± Cre transgenic mice. The (MMTV ± Cre6 Catnb+/l8x(ex3)) compound mutant progeny was easy to distinguish because of the smaller size of the newborn pups, cachectic posture, loss of hair, and thickened skin of the adults. The 12% yield of compound mutants (10 male and 11 female/175 total) was signi®cantly below the 25% expected from crossings of simple MMTV ± Cre with simple Catnb+/l8x(ex3) mice. Excision of b-catenin exon 3 in (MMTV ± Cre6 Catnb+/l8x(ex3)) mice was followed by Western-blot analysis using b-catenin speci®c mAbs. Detection of a Figure 1 (A) Detection of Cre activity in compound mutant new b-catenin protein species (*85 kDa) con®rmed MMTV ± Cre6R26R mice by staining of frozen tissue sections for b-galactosidase. Inserts show parallel stainings of control Cre mediated recombination at the targeted . tissues from simple mutant R26R mice. Images were taken Mutant b-catenin was detected in the skin, salivary, through 680 lens. (B) Analysis of protein extracts from prostate preputial, harderian, and mammary glands (Figure of Catnb+/lox(ex3) (cont) littermates or of various tissues of 1B), as well as the prostate (Figure 1C, a). The +/lox(ex3) MMTV ± Cre6Catnb with b-catenin speci®c mAb; (sal) stabilized form of b-catenin was absent in prostate salivary, (prep) preputial, (hard) harderian, (mam) mammary. (C) +/l x(ex3) Analysis of protein extracts from prostate of a 4 month old male protein extracts from control Catnb 8 (Figure 1B, MMTV ± Cre (control) and compound mutant (MMTV ± Cre x cont) or MMTV-Cre mice (Figure 1C, a). C-myc and Catnb+/lox(ex3)) litter mates. (a) Extracts from two representative cyclin D1 are transcriptional targets of b-catenin in animals (9 and 14) with b-catenin mAb, diamond indicates the wt colon carcinoma cells (He et al., 1998; Tetsu and and arrowhead points to the exon 3 excised b-catenin, (b) myc,(c) diamonds indicate cyclin D1 and arrowhead points to cyclin D2, McCormick, 1999). Prostates from two mice of 16 and (d) diamond indicates E-cadherin (E-Cad120) and arrowhead 20 weeks of age were used for Western-blot analysis. 97 points to the Mr 97 000 fragment (E-Cad )(e) p27 and (f) b- Expression of c-myc was clearly upregulated in the

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4101 level of E-cadherin (Figure 1C, d). Failure to detect changes in the levels of these may in part be due to the fact that the organs contained both normal and abnormal cells.

Stabilization of b-catenin is sufficient for induction of PIN-like lesions and hyperplasia Without exception, the prostates of all examined male (MMTV ± Cre6Catnb+/l8x(ex3)) mice, ranging from 10 to 21 weeks of age (one mouse of 10 weeks, two of 11 weeks, one of 13 weeks, three of 20 weeks, and one of 21 weeks) contained lesions histologically characteristic of high grade PIN. The lumen of acini were ®lled with atypical cells in a cribriform or solid pattern. These lesions were present in all prostatic lobes (Figure 2, compare a and b), namely the ventral (Figure 2, compare c and d), dorsal (Figure 2, compare e and f), and anterior (Figure 2, compare g and h) prostate. The prostatic appeared multilayered (Figure 2, i; arrowhead) and contained cytologically atypical lumi- nal cells (Figure 2, j arrowhead). Lesions frequently were neighbored by a monolayer of di€erentiated columnar epithelial cells, morphologically similar to the epithelium of healthy control mice (Figure 2i, arrow, and Figure 2j insert, arrow). The characteristic PIN histology, the extent (a€ecting all lobes) and the penetrance (a€ecting all examined male mice) of the lesions, suggest that b-catenin was initiating carcino- genesis. Absence of obvious invasion or metastases in the older mice suggests that additional genetic changes may be required for malignant progression of the lesions.

Accumulation of b-catenin, early loss of epithelial basolateral marker (NKCC1), and expression of androgen receptor (AR) in PIN-like lesions +/l x(ex3) Figure 2 Prostate pathology of MMTV ± Cre6Catnb 8 b-catenin in healthy secretory cells of the prostate of mice. Histological analysis with HE staining of 6 mm sections from representative prostate tissues of 4-month-old male control compound mutant mice was localized at the basolateral (MMTV ± Cre) (a, c, e, and g) and compound mutant (MMTV ± cellular membrane (Figure 3, arrow). Dysplastic cells Cre6Catnb+/lox(ex3)) littermate mice (i, b, d, f, h and j). (a and b) within the lesions were characterized by intracellular 65 magni®cation of the entire prostates. Thick arrows show the accumulation of b-catenin, as revealed by intense urethra, while the boxes show ®elds that are magni®ed in the ®gures below. Blue boxes (c and d, 650) mark ®elds in the immuno-¯uorescence staining (Figure 3a,c, arrow ventral prostate, red boxes (e and f, 650) in the dorsal, and black heads), con®rming the Western blot results and boxes (g and h, 650) show ®elds in the coagulating gland suggesting stabilization of b-catenin in these cells. This (anterior prostate). (i) 6100 and (j) 6150 magni®cation of allowed for the analysis of epithelial markers in the representative lesions to highlight the overlapping aberrant nuclei exact cells with stabilized b-catenin. One property of and shape of the tumor cells in comparison with that of the neighboring healthy secretory cells. Arrowheads point to lesions, early neoplastic epithelial cells is the loss of polariza- while arrows show healthy cells. Note that the dysplastic cells in tion, which is revealed by the loss or reorganization of PIN-like lesions show nuclear polymorphism, hyperchromia, and basolateral or apical markers. Two distinct Na-K-Cl prominent nucleoli cotransporter isoforms have been identi®ed. The NKCC1 (CCC1, BSC2) isoform is present in a wide variety of secretory epithelia and is localized to the prostate from compound mutant but not from control basolateral membrane, while NKCC2 (CCC2, BSC1) is (MMTV ± Cre) mice (Figure 1C, b). C-myc upregula- found only in the kidney (Haas and Forbush, 1998). tion is indicative of active b-catenin signaling, and Expression of the Na-K-Cl co-transporter (NKCC1) suggests stimulation of cell proliferation. Little change was restricted to the basolateral region of prostate was detected in the level of cyclin D1 or D2 (Figure secretory epithelial cells (Figure 3b, arrow). Stabiliza- 1C, c). No changes were seen in the levels of cKIP/p27 tion of b-catenin in the prostate epithelium led to the (Figure 1C, e). No signi®cant change was seen in the loss of NKCC1 immuno¯uorescent detection in the

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4102

Figure 3 Double immuno¯uorescence and immunoperoxidase analysis of expression of b-catenin, NKCC1 and AR by dysplastic cells of the PIN-like lesions. (a, b, c) b-catenin and NKCC1 staining. (f, g, h) b-catenin and AR staining, (d and e) AR staining. (a) Stabilization of b-catenin in the lesion is evident at the single cell level (arrow head) and is indicated by intense green staining, as compared to the weak basolateral staining of nearby normal cells (arrow). (b) NKCC1 is visualized by red staining; extinction of NKCC1 at the single cell level (arrow head) coincides with the accumulation of b-catenin, while basolateral distribution of NKCC1 is evident from intense red staining in neighboring healthy cells (arrow). (c) The exclusive expression of stabilized b-catenin or of NKCC1, revealed by exclusive green (arrow head) or red (arrow) stain and absence of orange (overlap). (f) Stabilization of b- catenin in the lesion is indicated by intense green staining, as compared to the weak staining of nearby normal cells. (g) Expression of AR is indicated by red staining; note weaker staining of the PIN-like lesion as compared with that of the neighboring healthy monolayer of secretroy cells. (h) Co-expression in healthy cells of b-catenin and AR (orange); note strong b-catenin but weak AR expression in the PIN-like lesion. Arrowheads point to lesion and arrows to healthy epithelium. (d) Immunoperoxidase staining of PIN-like lesion for AR. Arrow shows healthy secretory epithelium surrounding the PIN like lesion, indicated by arrow head. (e) Control secondary antibody staining of a serial section of the same lesion as in d. Arrowheads point to the dysplastic cells of the PIN-like lesion, and arrows to the healthy cells on the rim of the lesion

Figure 4 Immunohistochemical analysis of the expression of E-cadherin, NKCC1 and keratin 5, and immunoperoxidase staining of p63 and BrdU. (a, b, c) E-cadherin and NKC1 staining. (d) b-catenin and keratin 5 staining. (e and f) p63 staining. (g) Brdu staining. (a) Normal punctate membrane localization of E-cadherin in healthy epithelium (arrow) as well as PIN-like lesions (arrow head) indicated by green staining. (b) Expression of NKCC1 by healthy secretory epithelium is indicated by intense red staining (arrow), while its absence in the PIN-like lesions is evident from absence of staining (arrow head). (c) Overlap of E-cadherin and NKCC1 in healthy cells (arrow) is indicated by orange color, while exclusive expression of b-catenin in the lesion (arrow head) is indicated by green color. (d) Stabilization of b-catenin in dysplastic cells of the lesion is seen as intense green staining (arrow head), as compared with the weak staining of neighboring healthy cells (arrow). Expression of keratin 5 by basal cells is indicated by red staining; note its absence in the lesion. (e and f) Immunoperoxidase staining of basal cells at the rim of prostate acini for p63. Section shown in e contains a PIN-like lesion (arrowhead) while f is a healthy gland. Note that expression of p63 (arrows) is restricted to basal cells and does not appear in the PIN-like lesion. (g) Immunoperoxidase staining of BrdU in the nuclei of dysplastic cells. Insert shows lack of BrdU staining in a section of healthy prostate. Note the large, overlapping and atypical nuclei of cells in the lesion shown in g as compared to those of the healthy prostate shown in the insert

very early lesions, detectable at the single cell level compare d to e, arrowheads). However, the level of AR (Figure 3b,c, arrowhead) and in the more advanced expression was obviously reduced in the majority of the high grade PIN-like lesions (not shown). This loss is dysplastic cells (Figure 3f ± h, arrowheads) as compared suggestive of b-catenin mediated loss of epithelial to the normal secretory epithelium (Figure 3d, arrow). polarization. Downregulation of AR in the prostate of compound PIN-like lesions stained positive for androgen mutant mice was also observed by Western blot receptor (AR) by (Figure 3, analysis using AR speci®c antibody (not shown).

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4103 Expression of AR in the b-catenin induced PIN-like Luminal characteristic and increased mitotic activity of lesions indicates that the tumors had characteristics of the dysplastic cells secretory luminal epithelium, as is typical of human PIN and prostate carcinoma (Stanbrough et al., 2001). Prostate basal cells typically express 5 and 14 Downregulation of AR in the b-catenin induced lesions (Hayward et al., 1996) as well as p63 (Signoretti et al., is in agreement with reports suggesting that the level of 2000). Basal cell markers, keratin 5 (Figure 4d, arrow) AR expression in human PINs is downregulated and p63 (Figure 4e,f, arrows) stained a rim of residual (Magi-Galluzzi et al., 1997). basal cells in the periphery of acini, but were not expressed in the dysplastic epithelium of the PIN-like lesions (Figure 4d,e, arrowheads). These observations E-cadherin expression, size and subcellular localization indicate that the lesions either did not have a basal was not altered origin or rapidly lost basal cell markers upon stabiliza- Aberrations of E-cadherin expression and function tion of b-catenin. BrdU staining revealed increased have been noted in malignant prostate epithelium cellular proliferation among the dysplastic cells (Figure (Rashid et al., 2001). These have been attributed to 4g) (39 dividing cells amongst 540 cells, in four ®elds), transcriptional repression of the gene and/or post while no staining was detected in the prostate of control +/lox(ex3) translational cleavage of the native Mr 120 000 E- (Catnb ) mice (no dividing cells detected among cadherin (E-Cad120), to a novel cleaved membrane- over 400 cells in six ®elds). These observations correlate 97 bound Mr 97 000 fragment (E-Cad ) that lacks the b- b-catenin stabilization with increased epithelial prolif- catenin binding domain (Rashid et al., 2001). Western eration and formation of PIN-like lesions. blot analysis of prostate extracts revealed a prominent Interestingly, foci of basal cell hyperplasia consisting E-Cad120 and a weaker E-Cad97 protein species in both of cells expressing both activated b-catenin and keratin 5 healthy control as well as in diseased compound were detected. These foci were dicult to score due to mutant mice, with no signi®cant di€erence between their small size, appearing as aggregates of less than the mice (Figure 1C, d). Double immuno¯uorescence 20 cells in the plane of the section. Keratin 5 was analysis for NKCC1 and E-cadherin con®rmed normal downregulated in the foci (Figure 5b, arrowhead) as localization of E-cadherin at the cell membrane of compared to the adjacent basal cells (Figure 5b, arrow), dysplastic cells within the lesions (Figure 4a,c, arrow- and was extinguished in cells protruding in the lumen. heads) that no longer stained for NKCC1 (Figure 4b,c, Basal cells neighboring the foci stained for keratin 5 arrowheads). These results suggest that stabilization of (Figure 5b, arrow) but were negative for b-catenin b-catenin does not lead to severe disruption of (Figure 5a, arrow), leaving the possibility open that these homotypic cell ± cell adhesion, and are in agreement may be common progenitors of both basal and secretory with b-catenin acting at the initiation of the PIN. cells in transition to a secretory , with the cells losing basal markers and gaining luminal markers as they enter dysplasia. Another explanation is that the PIN-like lesions originated from the secretory cells. A relatively infrequent observation was that of squamous keratinized cells and/or keratin debri in the lumen of the acini of compound mutant mice, that stained for keratin 5 (Figure 5c, arrowhead) and/or (Figure 5d,e, arrowheads; see Figure 5f for HE staining). Squamous metaplasias have been associated with typical of the prostate in cases where the patient has undergone extensive treatment, and may be a rare event in prostate carcinogenesis.

b-catenin stabilization causes epithelial transdifferentiation in multiple organs

Figure 5 Immuno¯uorescence analysis of b-catenin, keratin 5, Histological examination revealed multiple lesions in and keratin 1 in foci of proliferating basal cells, and in keratinized other organs that had shown Cre activity as revealed prostate acini of compound mutant mice. Accumulation of b- by lacZ staining in MMTV ± Cre6R26R mice. The catenin (a) by a focus of keratin 5 expressing cells (b)ina predominant e€ect was squamous metaplasia with hyperproliferating basal focus. Arrowheads point to the lesion, and arrows to healthy neighboring basal cells. Expression of b- extensive keratinization. Generalized thickening of the catenin (c, green) in PIN-like lesion and keratin 5 deposits in the skin, including the paws, ears, lips and eyelids, and lumen (c, red, arrowhead), and basal cell layer (c, red, arrow). patchy loss of hair, gave the mice their characteristic Keratin 1 is expressed in squamous cells and deposits in the appearance. In agreement with earlier reports (Gat et lumen of the gland (d, red, arrowhead), and is absent in the al., 1998), we observed thick ridges in skin and hair glandular epithelium (d, arrow) or the PIN (arrowhead). The outline of a keratin 1 positive intraluminal `ghost cell' (e, coat, with marked thickening of the paws and ears. arrowhead). HE staining of keratinized ghost cell (f, arrowhead) Histologically, the skin showed hyperplasia and

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4104

Figure 6 HE stained paran-embedded tissue sections from organs of 5-month-old (MMTV ± Cre6Catnb+/lox(ex3)) mice. (a) Skin, shows follicular epithelioid cysts (black arrowhead), trichofolliculomas (black arrows) containing also sebaceous glands (Sg), epidermal hyperplasia and aberrant invaginations (yellow arrowheads). (b) Salivary gland, (c) preputial gland, (d) harderian gland, are characterized by squamous metaplasia with keratinization of the glandular epithelium (black arrowheads). Inserts in a ± d show healthy control tissue from simple mutant MMTV ± Cre sibling. (e) Mammary gland, from a female compound mutant mouse 2 months after delivery with no lactation. Sections show di€use squamous metaplasia involving both ductal and lobular epithelium, with orthokeratinosis and extensive deposition of keratin ®bers (black arrowhead), and intense in¯ammatory in®ltrate with multi nucleated giant cells. Insert shows healthy mammary gland of MMTV-Cre sibling. (f), Lower jaw, showing inscissor tooth (black arrow) surrounded by odontoma (black arrowhead). Images were taken through a 620 lens

increased epidermal invaginations which developed cause widespread epithelial tumors. On the other hand, into aberrant hair follicle cysts and characteristic the stabilization of b-catenin as a single dominant tricofolliculomas (Figure 6). Surprisingly, similar le- genetic event is sucient to induce lesions character- sions were seen in the salivary gland (Figure 6b, black istic of prostate intraepithelial neoplasia. Such lesions arrowhead), preputial gland (Figure 6c, black arrow- were present in all compound mutant mice and all head), harderian gland (Figure 6d, black arrowhead), prostate acini expressing stabilized b-catenin. Hetero- and mammary gland (Figure 6e, arrowheads). We geneous and multifocal lesions resembling high grade con®rmed earlier observations (Miyoshi et al., 2001) human PIN were detected as early as 10 weeks of age. that the mammary glands were severely a€ected Nevertheless, and contrary to recent speculations (Figure 6c,e). The inscissor tooth (Figure 6f, arrow) (Yang et al., 2002), there was no convincing evidence was surrounded by numerous lobules of dentin, for invasion or by the lesions in mice reminiscent of odontoma (Figure 6f, arrowhead). reaching up to 5 months of age. Taken together, our observations indicate that Little is known about the mechanisms of initiation of stabilization of b-catenin may not be sucient to prostate cancer. Histopathological studies of prostate cause widespread epithelial tumors. The hyperplasia cancers have led to the identi®cation of a speci®c type and dysplasia seen in the prostate seems to be an of lesion that represents the primary precursor of exception rather than the rule. Stabilization of b- human prostate cancer, known as prostate intra- catenin is also etiologically linked with colon cancer epithelial neoplasia (PIN) (McNeal and Bostwick, (Harada et al., 1999; Kinzler and Vogelstein, 1996). 1986). PIN is recognized as a continuum between However, since in our compound mutant mice, b- low-grade and high-grade forms, with high-grade PIN catenin stabilization did not occur in the gut, most likely representing the immediate precursor of adenomatous polyps were not observed. Like prostate early invasive carcinoma. Characteristic architectural cancer, squamous metaplasia is not a common feature and cytological features are shared between PIN of colon cancer. These observations suggest that the lesions and early invasive carcinomas including multi- stabilization of b-catenin may have di€erent conse- focal nature of the lesions, and common chromosomal quences, depending on the type of epithelial cell abnormalities (reviewed in Abate-Shen and Shen, targeted, and perhaps also on the immediate environ- 2000). While the observed b-catenin associated lesions ment of target cells. reported here have the characteristic appearance of PIN, the association with progression has not been demonstrated. This may be considered as a caveat but Discussion we think that it is more likely to re¯ect the speci®c role of b-catenin in the initiation of PIN-like lesions. A Taken together, our observations indicate that contrary similar role for b-catenin has been established in the to expectations b-catenin may not have the potential to initiation of intestinal adenomatous polyps, which are

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4105 considered to be the immediate precursors of invasive actin , and contributes to tumor suppres- carcinoma and colon cancer (Harada et al., 1999; sion by promoting cell adhesion and by competing for Kinzler and Vogelstein, 1996). Extensive animal the free pool of b-catenin (Gottardi and Gumbiner, modeling has shown that either the homozygous 2001; Gottardi et al., 2001). Adhesion could contribute inactivation of the adenomatous polyposis coli (APC) to tumor suppression by physically joining cells, by gene (Fodde et al., 1994; Luongo et al., 1994; Shibata facilitating other juxtacrine signaling events, or by et al., 1997; Takaku et al., 1998) or the stabilization of binding and antagonizing the nuclear signaling func- the b-catenin gene product (Harada et al., 1999) are tion of b-catenin (Gottardi and Gumbiner, 2001; necessary and sucient to initiate these lesions. A Gottardi et al., 2001). Loss of E-cadherin is associated recent re-evaluation of the accumulated histological with poor prognosis in human prostate cancer, and data has con®rmed that these lesions are not invasive, thus may be linked to the progression rather than the suggesting that additional genetic changes are required initiation of the cancer (Bussemakers et al., 2000; for progression from adenoma to carcinoma (Sund- Davies et al., 2000). In our model, the normal berg, 2000). localization and strong expression of E-cadherin in Since in our compound mutant mice, b-catenin the PIN-like lesions correlates with the absence of stabilization did not occur in the gut, adenomatous invasion or metastasis, and with the notion that the polyps were not observed. Using mice that express an action of b-catenin is restricted to the initiation of the inducible Cre speci®cally in the gut, we have PIN lesions. Additional genetic changes are likely to be reproduced results like earlier reports (Harada et al., required to promote the progression of these lesions. 1999) that activation of b-catenin is sucient for the It has been proposed that androgen speci®c signaling initiation of adenomatous polyps throughout the via the epithelial (Stanbrough et al., 2001) and stromal intestine (our unpublished results). We did not observe (Kurita et al., 2001) AR regulates the growth and any squamous metaplasia with these intestinal lesions. apoptosis of secretory epithelia in healthy prostate. In These observations suggest that the stabilization of b- response to exogenous androgen, b-catenin binds AR, catenin may have di€erent consequences, depending on shuttles to the nucleus (Mulholland et al., 2002), and the type of epithelial cell targeted, and perhaps also on stimulates androgen dependent transcriptional activa- the immediate environment of target cells. tion (Truica et al., 2000). Prostate cancer is an Over expression of Wnt genes is causally linked with androgen dependent disease that can be treated by mammary carcinogenesis (Li et al., 2000), as well as androgen ablation therapy. Expression of AR in the b- some degree of squamous metaplasia in mice (Harold catenin induced PIN-like lesions indicates that these Varmus, personal communication). We propose that had characteristics of secretory luminal epithelium, and induction of mammary in Wnt over- of human PIN. There is some controversy as to expressing mice requires the engagement of pathways whether the level of AR expression in PINs is in addition to b-catenin. Our results may appear downregulated (Magi-Galluzzi et al., 1997) or en- contradictory to the reported induction of mammary hanced (Truica et al., 2000). In our study, immuno- adenocarcinomas in mice over expressing an amino- histochemical and Western-blot analysis (not shown) terminal truncated b-catenin (Imbert et al., 2001). revealed lower steady-state levels of AR in PIN lesions. However, that mouse model di€ers from ours in the The presented information together with earlier extent of truncation of b-catenin and in its hormone studies indicate that the b-catenin pathway is a key dependent (MMTV-LTR driven) overexpression. regulator of cell fate determination in epithelial cells in One of the earliest consequences of activation of b- general and in a limited number of tissues may be the catenin in the prostate epithelium was the apparent loss initiating signal for neoplasia. The interesting question of NKCC1 expression as revealed by immuno¯uores- now is to address the tissue speci®city of this choice, cent staining. Two distinct Na-K-Cl cotransporter namely the genetic factors, signaling, and tissue isoforms have been identi®ed by cDNA cloning and environmental factors that determine the choice expression. The NKCC1 (CCC1, BSC2) isoform is between transdi€erentiation and pre-neoplasia in present in a wide variety of secretory epithelia and is di€erent tissues. localized to the basolateral membrane, while NKCC2 (CCC2, BSC1) is found only in the kidney (Haas and Forbush, 1998). We document for the ®rst time the Materials and methods basolateral expression of NKCC1 in the secretory epithelium of healthy mouse prostate, and the rapid Mice loss of detection of this molecule upon stabilization of Catnb+/7l8x(ex3) (Harada et al., 1999), MMTV ± Cre b-catenin. This loss may re¯ect the loss of epithelial (Wagner et al., 1997), R26R (Soriano, 1999) have been polarity and may be a useful marker for identifying described previously. early human prostate cancer lesions. Stabilization of b-catenin did not alter the membra- nous localization of E-cadherin. E-cadherin is a Histology transmembrane protein engaged in homotypic cell-cell Mice were perfused with 4% paraformaldehyde, and isolated adhesion (Vasioukhin et al., 2001). Its cytoplasmic tissues were ®xed in 4% paraformaldehyde for 4 ± 6 h and domain interacts with b-catenin, a-catenin, and the then transferred to 70% ethanol. Fixed samples were

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4106 embedded in paran, sectioned at 6 mm, and stained with Kit, HRP/DAB-BioGenex, QD200-OX). b-galactosidase hematoxylin and eosin. activity was revealed on 6 mm cryosections as described (Khazaie et al., 1994). Antibodies Western blotting Primary antibodies: Rabbit polyclonal antibodies against keratin 1 and keratin 5 (BAbco; PRB-165P and PRB-160P); Protein lysates were prepared in RIPA bu€er as described in c-myc (Upstate biotechnology, 06-340); E-cadherin (BD (Gounari et al., 2001), electrophoresed on 10% SDS ± PAGE Transduction Laboratories, C20820); cyclin D1 (Santa Cruz, and blotted onto nitrocellulose membranes (DCL, Schleicher sc-593), NKCC1 was a kind gift from RJ Turner (Moore- and Schuel, Dassel, Germany). Primary antibodies were Hoon and Turner, 1998). Mouse monoclonal antibodies incubated with the membranes for 1 ± 3 h at room tempera- against b-catenin, p27, and anti-BrdU (BD Transduction ture, and then incubated with an HRP-linked anti-mouse or Laboratories, C19220, K25020, and 347580), AR (Upstate anti-rabbit secondary antibody for 30 min at room tempera- Biotechnology, PG-21 06-680), b-actin (Chemicon Internat- ture. The signal was developed using the ECL detection ional, Inc., MAB1501). FITC-conjugated anti-mouse (Alexa- system (Amersham, Little Chalfont, UK). ¯uor 488) and Texas Red-conjugated anti-rabbit (Alexa¯uor 594) secondary antibodies (Molecular Probes) were used for immunohistochemistry and HRP-conjugated anti-mouse and Acknowledgments anti-rabbit secondary antibodies were used for Western K Khazaie was supported by the National Colorectal blotting (Transduction Labs). Further antibodies, used for Research Alliance, Dana Farber Cancer Institute, and the immunoperoxidase staining, were goat anti-Mouse polyvalent Harvard Cancer Center. M Taketo was supported by the IgG Biotin Conjugate (Super Sensitive MultiLink-BioGenex, Organization for Pharmaceutical Safety and Research HK340-9K) and Peroxidase Conjugated Streptavidin (Bio- (OPSR) and the Ministry of Education, Science, Sports Genex, HK330-9K). and Culture (MESSC) of Japan. We thank N Harada for the construction of the Catnb +/lox(ex3) mice. K Khazaie is obliged to Robert Mayer, Glenn Drano€, and Ramesh Immunohistology and enzymatic assay and staining for Shivdasani for their support and encouragement. This b-galactosidase work would have not been possible without the advice of Immuno¯uorescence staining was as described before Massimo Loda, Lothar Hennighausen, Keiko Miyoshi and (Miyoshi et al., 2001). Images were visualized by a Nikon Jonathan M Shillingford. Li Zhang and James Dilks are E300 ¯uorescent microscope and recorded with a Spot CCD acknowledged for excellent histological service. Eric Smith camera. Immunoperoxidase staining was according to is acknowledged for expert assistance with recording and instructions of the kit manufacturer (SS MultiLink Detection presentation of images.

References

Abate-Shen C and Shen MM. (2000). Genes Dev., 14, 2410 ± Hayward SW, Baskin LS, Haughney PC, Cunha AR, Foster 2434. BA, Dahiya R, Prins GS and Cunha GR. (1996). Acta Bienz M and Clevers H. (2000). Cell, 103, 311 ± 320. Anat., 155, 81 ± 93. Brewster SF, Browne S and Brown KW. (1994). J. Urol., 151, He TC, Sparks AB, Rago C, Hermeking H, Zawel L, da 1073 ± 1077. Costa LT, Morin PJ, Vogelstein B and Kinzler KW. Bussemakers MJ, Van Bokhoven A, Tomita K, Jansen CF (1998). Science, 281, 1509 ± 1512. and Schalken JA. (2000). Int. J. Cancer, 85, 446 ± 450. Imbert A, Eelkema R, Jordan S, Feiner H and Cowin P. Chesire DR, Ewing CM, Sauvageot J, Bova GS and Isaacs (2001). J. Cell. Biol., 153, 555 ± 568. WB. (2000). Prostate, 45, 323 ± 334. Iozzo RV, Eichstetter I and Danielson KG. (1995). Cancer Davies G, Jiang WG and Mason MD. (2000). J. Urol., 163, Res., 55, 3495 ± 3499. 985 ± 992. Khazaie K, Prifti S, Beckhove P, Griesbach A, Russell S, FoddeR,EdelmannW,YangK,vanLeeuwenC,CarlsonC, Collins M and Schirrmacher V. (1994). Proc. Natl. Acad. Renault B, Breukel C, Alt E, Lipkin M, Khan PM and Sci. USA, 91, 7430 ± 7434. Kucherlapati R. (1994). Proc. Natl. Acad. Sci. USA, 91, Kinzler KW and Vogelstein B. (1996). Cell, 87, 159 ± 170. 8969 ± 8973. Kurita T, Wang YZ, Donjacour AA, Zhao C, Lydon JP, Gat U, DasGupta R, Degenstein L and Fuchs E. (1998). Cell, O'Malley BW, Isaacs JT, Dahiya R and Cunha GR. 95, 605 ± 614. (2001). Cell Death Di€er., 8, 192 ± 200. Gottardi CJ and Gumbiner BM. (2001). Current Biol., 11, Lau W and Clevers H. (2001). Nat. Genet., 28, 3±4. R792 ± R794. Li Y, Hively WP and Varmus HE. (2000). Oncogene, 19, Gottardi CJ, Wong E and Gumbiner BM. (2001). J. Cell. 1002 ± 1009. Biol., 153, 1049 ± 1060. Luongo C, Moser AR, Gledhill S and Dove WF. (1994). Gounari F, Aifantis I, Khazaie K, Hoe¯inger S, Harada N, Cancer Res., 54, 5947 ± 5952. Taketo MM and von Boehmer H. (2001). Nat. Immunol., Magi-Galluzzi C, Xu X, Hlatky L, Hahnfeldt P, Kaplan I, 2, 863 ± 869. Hsiao P, Chang C and Loda M. (1997). Mod. Pathol., 10, Haas M and Forbush III B. (1998). J. Bioenerg. Biomembr., 839 ± 845. 30, 161 ± 172. McNeal JE and Bostwick DG. (1986). Hum. Pathol., 17, 64 ± Harada N, Tamai Y, Ishikawa T, Sauer B, Takaku K, 71. Oshima M and Taketo MM. (1999). EMBO J., 18, 5931 ± 5942.

Oncogene Tissue dependent transdifferentiation or metaplasia by b-catenin F Gounari et al 4107 MiyoshiK,CuiY,RiedlingerG,LehoczkyJ,ZonL,OkaT, Sparks AB, Morin PJ, Vogelstein B and Kinzler KW. (1998). Dewar K and Hennighausen L. (2001). Genomics, 71, Cancer Res., 58, 1130 ± 1134. 150 ± 155. Stanbrough M, Leav I, Kwan PW, Bubley GJ and Balk SP. Moore-Hoon ML and Turner RJ. (1998). Biochim. Biophys. (2001). Proc. Natl. Acad. Sci. USA, 98, 10823 ± 10828. Acta., 1373, 261 ± 269. Sundberg JP. (2000). Colon Cancer Study Section. Sundberg Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, JP (ed). The Jackson Laboratory. Vogelstein B and Kinzler KW. (1997). Science, 275, 1787 ± Suzuki H, Aida S, Akimoto S, Igarashi T, Yatani R and 1790. Shimazaki J. (1994). Jpn. J. Cancer Res., 85, 847 ± 852. Mulholland DJ, Cheng H, Reid K, Rennie PS and Nelson TakakuK,OshimaM,MiyoshiH,MatsuiM,SeldinMFand CC. (2002). J. Biol. Chem., 20, 20. Taketo MM. (1998). Cell, 92, 645 ± 656. Phillips SM, Morton DG, Lee SJ, Wallace DM and Tetsu O and McCormick F. (1999). Nature, 398, 422 ± 426. Neoptolemos JP. (1994). Br. J. Urol., 73, 390 ± 395. Truica CI, Byers S and Gelmann EP. (2000). Cancer Res., 60, Rashid MG, Sanda MG, Vallorosi CJ, Rios-Doria J, Rubin 4709 ± 4713. MA and Day ML. (2001). Cancer Res., 61, 489 ± 492. Vasioukhin V, Bauer C, Degenstein L, Wise B and Fuchs E. Reya T, Morrison SJ, Clarke MF and Weissman IL. (2001). (2001). Cell, 104, 605 ± 617. Nature, 414, 105 ± 111. Voeller HJ, Truica CI and Gelmann EP. (1998). Cancer Res., ShibataH,ToyamaK,ShioyaH,ItoM,HirotaM, 58, 2520 ± 2523. Hasegawa S, Matsumoto H, Takano H, Akiyama T, WagnerKU,WallRJ,St-OngeL,GrussP,Wynshaw-Boris Toyoshima K, Kanamaru R, Kanegae Y, Saito I, A, Garrett L, Li M, Furth PA and Hennighausen L. Nakamura Y, Shiba K and Noda T. (1997). Science, (1997). Nucleic Acids Res., 25, 4323 ± 4330. 278, 120 ± 123. Watanabe M, Kakiuchi H, Kato H, Shiraishi T, Yatani R, Signoretti S, Waltregny D, Dilks J, Isaac B, Lin D, Garraway Sugimura T and Nagao M. (1996). Jpn. J. Clin. Oncol., 26, L, Yang A, Montironi R, McKeon F and Loda M. (2000). 77 ± 81. Am.J.Pathol.,157, 1769 ± 1775. Yang F, Li X, Sharma M, Sasaki CY, Longo DL, Lim B and Soriano P. (1999). Nat. Genet., 21, 70 ± 71. Sun Z. (2002). J. Biol. Chem., 15, 15.

Oncogene