Expression and androgen regulation of C-CAM cell adhesion ... - Nature

ã

Oncogene (1999) 18, 3252 ± 3260 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

Expression and androgen regulation of C-CAM cell adhesion molecule isoforms in rat dorsal and ventral prostate Andrew N Makarovskiy3, Yeong-Shiau Pu1, Piao Lo1, Karen Earley1, Michael Paglia3, Douglas C Hixson2 and Sue-Hwa Lin*,1 1

Department of Molecular Pathology, Box 89, The University of Texas, MD Anderson Cancer Center, 1515 Holcombe Blvd., Houston, Texas TX, 77030, USA; 2Department of Medical Oncology, Rhode Island Hospital, Brown University, Providence, Rhode Island 02903, RI, USA; 3Department of Urology, Rhode Island Hospital, Brown University, Providence, Rhode Island, RI 02903, USA

C-CAM is an epithelial cell adhesion molecule with two major splice variants that dier in the length of the cytoplasmic domain. C-CAM1 (long (L)-form) strongly suppresses the tumorigenicity of human prostate carcinoma cells. In contrast, C-CAM2 (short (S)-form) does not exhibit tumor-suppressive activity. In the present study we have investigated the functional signi®cance of L-form and S-form C-CAM in rat prostate by examining their expression and distribution in dierent prostate lobes and their response to androgen deprivation. RNase protection assays with a probe for both CCAM isoforms detected high levels of C-CAM messages in the rat dorso-lateral prostate (DLP). L- and S-form proteins, localized by indirect immuno¯uorescence using isoform-speci®c antipeptide antibodies, were co-expressed on the apical surface of prostate epithelial cells in normal DLP. Androgen depletion did not signi®cantly change the steady state levels of C-CAM message and protein expression in the DLP, although there was a change in the pattern of protein expression in these lobes. In contrast, C-CAM isoform messages and proteins were undetectable in normal ventral prostate (VP) but increased markedly in this lobe in response to castration, producing isoform ratios similar to those in DLP. These results demonstrate that coordinate expression of CCAM isoforms is maintained in the VP following androgen depletion and suggest that androgen suppresses C-CAM expression in VP but not in DLP. These results suggest that balanced expression of L- and S-form CCAM is important for normal prostate growth and dierentiation. Keywords: C-CAM; androgen regulation; prostate

Introduction Cell adhesion molecules that mediate cell-cell and cellmatrix interactions play critical roles in the development of multicellular organisms. Pathological states such as tumor progression and metastasis have been shown to be associated with changes in expression of these molecules. This is exempli®ed by the signi®cant decrease of E-cadherin, a widely expressed Ca2+dependent cell adhesion molecule, during progression of rat prostate cancer (Bussemakers et al., 1992). In human prostate cancer, decreased E-cadherin expres*Correspondence: S-H Lin Received 6 August 1998; revised 12 November 1998; accepted 5 January 1999

sion correlates with the loss of tumor dierentiation and is thought to contribute to the invasive phenotype of prostate tumors (Umbas et al., 1992). Recent studies show that other cell-adhesion molecules, such as CCAM (Hsieh et al., 1995) and the integrins (Cress et al., 1995), also have functions related to the suppression of cancer progression. Understanding how these cell adhesion molecules are regulated in both normal and pathological states will allow us to exploit these molecules to treat cancer. C-CAMs are cell adhesion molecules of immunoglobulin supergene family and structurally dierent from cadherins and integrins (Lin et al., 1991). L- and Sform C-CAM, the predominant isoforms in the rat, are generated from a single gene by alternative splicing (Cheung et al., 1993a; Najjar et al., 1993). L-form CCAM has a 71-amino acid cytoplasmic domain. Due to deletion of exon 7 (53 bp), S-form C-CAM has a cytoplasmic domain of 10 amino acids. Decreased C-CAM expression is a common feature of carcinomas in a variety of tissues including liver (Hixson et al., 1985), colon (Neumaier et al., 1993), and prostate (Kleinerman et al., 1995). Immunohistologic studies of human prostate cancer specimens showed that C-CAM is undetectable in areas of abnormal proliferative growth, suggesting that loss of C-CAM plays a central role in cancer development (Kleinerman et al., 1995). This conclusion is supported by several lines of evidence demonstrating that C-CAM has a role in preventing tumorigenesis and can function as a tumor suppressor in prostate cancer (Hsieh et al., 1995). We have shown, for example, that constitutive expression of L-form C-CAM potently suppressed the tumorigenicity of PC-3, a C-CAM negative cell line (Hsieh et al., 1995). Conversely, expression of C-CAM antisense RNA downregulated C-CAM expression and resulted in the neoplastic conversion of a nontumorigenic prostate epithelial cell line (Hsieh et al., 1995). To identify the sequence critical for L-form CCAM's tumor-suppressive eect, we have performed structural analysis and found that the cytoplasmic domain is critical for its tumor suppressor activity (Luo et al., 1997). Consistently, lack of tumor suppression in tumor cells expressing S-form C-CAM has also been reported (Kunath et al., 1995; Luo et al., 1997). These observations suggest that signal transduction by the L-form cytoplasmic domain is important for L-form C-CAM-mediated tumor suppression. Because L-form C-CAM plays a critical role in prostate tumorigenesis, there is a need to determine how S-form C-CAM is involved in this process. In the prostate, there are two major types of epithelial cell,

Androgen regulation of C-CAM isoforms in rat prostate AN Makarovskiy et al

i.e., basal and luminal cells. These two cell types are functionally distinct and dier in their potential for neoplastic progression. It is not clear whether L- and S-form C-CAM are expressed in dierent populations of prostate epithelial cells or are co-expressed in the same cells. The former would indicate that L-form and S-form have dierent functions in dierent cell populations, whereas the latter suggests that a balance between L- and S-form C-CAM level may be critical for normal growth and dierentiation. Furthermore, because androgen regulates prostate growth and dierentiation and androgen ablation is one of the major therapy strategies for prostate cancer, the possibility that L- and S-form C-CAM are differentially regulated by androgen needs to be investigated. In the present study, we used RNase protection assays and isoform-speci®c antibodies to examine the lobular distribution and androgen regulation of CCAM isoforms in rat prostate.

a

b

Results L-form and S-form C-CAM messages in VP and DLP To assess the steady-state levels of L-form and S-form messages in the VP and DLP, RNase protection analyses were performed with antisense probes that can detect L-form and S-form C-CAM simultaneously (Figure 1a). The L-form and S-form messages were detected only in DLP and not in VP (Figure 1B). Semiquantitative data were obtained by densitometric analyses; the ratio of L-form to S-form C-CAM message, after adjustment for the dierence in the sizes of the protected fragments, was about 1 : 5. This result showed that both the L-form and the S-form CCAM were mainly expressed in the DLP with abundant prevalence of the S-form over the L-form. Localization of C-CAM isoforms in DLP The prostate gland has two types of epithelial cells: luminal epithelial cells and basal epithelial cells. Although there is still some controversy, it is commonly believed that the luminal epithelial cells are terminally dierentiated cells with secretory function, whereas the basal epithelial cells harbor the stem-cell population for prostate regeneration. To determine the spatial distribution of C-CAM expression in these epithelial subpopulations, frozen prostate sections were double immunostained with Ab669, which recognizes both C-CAM isoforms, and Mab K903 speci®c to the high-molecular-weight cytokeratins present in basal cells. Reactivity of the isoform speci®c antibodies was tested similarly. As shown in Figure 2, both DP and LP tissues reacted speci®cally and uniformly with C-CAM antibodies. The reactions were con®ned to the apical surface of the luminal cells, and no staining were detected on the basal cells. These analyses also showed that C-CAM was uniformly expressed on all glands and ducts examined. These observations suggest that in DLP, CCAMs are mainly expressed in well-dierentiated epithelial cells. While messages for both L-form and S-form CCAM were detected in the DLP and the presence of C-

Figure 1 Expression of L- and S-form C-CAM transcripts in VP and DLP. (a) The splicing of L-form and S-form C-CAM deduced from comparison of genomic and cDNA sequences (Cheung et al., 1993a) and the region used in the preparation of the antisense RNA probe. The restriction map and exon arrangement of the C-CAM1 genomic clone are shown. R, EcoRI; B, BamHI; H, HindIII. (b) RNase protection assay. Aliquots of total RNA (10 mg) obtained from rat VP and DLP were hybridized with a 32P-labeled antisense RNA probe speci®c for the C-terminal cytoplasmic domain of L-form C-CAM. tRNA (10 mg) was used as a control

3253


3254

CAM proteins were identi®ed by Ab669 (Figure 2), it was not clear whether the two isoforms were expressed dierentially or co-localized in the same epithelial cells. To determine the relative distribution of the C-CAM isoforms in the prostate gland, we used isoform-speci®c antipeptide antibodies. The homogenous and strong staining pattern produced by the S-form speci®c antibody, anti-CS, was con®ned to the apical surfaces of the luminal epithelial cells and was superimposable with Mab5.4 and Ab669 in double labeled sections (Figure 3). Similar labeling patterns were also observed with L-form speci®c antibody, anti-CL, which produced only weak staining, suggesting that the L-form C-CAM protein was present in DLP at a lower level than the S-form (Figure 3). Results from immuno-

chemical analyses of C-CAM isoforms were thus in close agreement with RNase protection assays showing higher levels of S-form message (Figure 1). Taken together, these data suggest that C-CAM isoform expression is transcriptionally regulated and that the two isoforms are co-expressed in luminal epithelial cells. Immunostaining of LP with these antibodies produced essentially identical results (data not shown). Eect of androgen on C-CAM isoform expression in DLP To investigate whether the levels of C-CAM isoform expression in the DLP were regulated by androgen, we compared the levels of C-CAM isoform transcripts in

Figure 2 Localization of C-CAM proteins in luminal epithelial cells. Frozen sections of DP (a ± c) and LP (d ± f) were double immunostained with Ab669 and Mab K903 and photographed with FITC (green) (a and d) or Texas Red (b and e) ®ltration. Panels c and f are the same sections shown in a, b and d, e, respectively, photographed using a dual ®lter cube that shows both FITC and Texas Red ¯uorescence. Arrows indicate apical surface of the luminal cells and arrowheads indicate basal membrane. (6006)


castrated animals with those of intact controls. As shown in Figure 4, the levels of C-CAM transcripts in DLP were not aected by castration, suggesting that the expression of C-CAM isoforms in DLP was not regulated by androgen. Prostate tissues collected 3 days after castration were analysed by immuno¯uorescence for changes in CCAM isoform expression associated with androgen deprivation. No signi®cant change in overall staining intensity was detected in either LP or DP with isoformspeci®c antibodies, and the relative abundance of the L- and S-form C-CAM were indistinguishable from those of intact animals. However, marked changes in the distribution of C-CAM on cell surface were observed (Figure 5); staining was no longer con®ned

to the apical surface of the luminal cells but was also prominent on the lateral surfaces (Figure 2 vs Figure 5). Actively degenerating luminal cells in the acinar portions of the gland as well as cuboidal epithelium of the prostatic ducts displayed a prominent shift of CCAM to the lateral surfaces. Unlike basal cells in normal prostate, which were negative for C-CAM, basal cells in castrated rats showed a heterogenous CCAM staining revealed by double labeling immuno¯uorescence using C-CAM antibodies together with Mab K903 speci®c for high-molecular-weight cytokeratins of the basal cells (Figure 5a ± c). These observations showed that although the steady-state level of C-CAM expression in DLP was not changed by androgen deprivation, castration did alter the

Figure 3 Co-localization of L-form and S-form C-CAM proteins in DP. Panels a and b, c and d, and e and f show pairs of micrographs from sections double immunostained with Mab 5.4 (b, d and f) and Ab669 (a), anti-CS (c) or anti-CL (e) antisera. Arrows indicate apical surface of the luminal cells and arrowheads indicate basal membrane. (6006)

3255


3256

pattern of C-CAM expression by redistributing CCAM to the lateral surfaces of the luminal cells and shifting the expression to the unpolarized basal cells. Therefore, the spatial distribution of C-CAM was altered by androgen deprivation. This suggested that androgen was required to maintain the polarized expression of C-CAM in luminal epithelial cells of DLP. Eects of androgen on C-CAM isoform expression in VP We surmised that the inability to detect C-CAM in ventral prostate was attributed to androgen-mediated suppression. To investigate this possibility, we examined the eect of androgen ablation on C-CAM isoform expression by RNase protection analysis. As shown in Figure 6, signi®cant increase of both the Land the S-form C-CAM messages was detected in RNA samples prepared from the VPs of castrated rats as compared with intact controls. Moreover, the levels of the L- and S-form C-CAM messages were increased in a coordinated fashion such that their relative abundance (S-form4L-form) was similar to those observed in DLP. The castration-induced increase in C-CAM levels was suppressed by injecting castrated rats with testosterone (Figure 6), suggesting that this increase was a response to androgen depletion.

VP collected from the control and the castrated rats were also analysed by immuno¯uorescence staining to determine whether androgen deprivation increased CCAM protein expression in this region. With the exception of endothelial cells and a few positive staining areas (less than 1% of the sections) (Figure 7), VPs from the control rats were generally unreactive with Ab669 in immuno¯uorescence assays. However, 3 days after castration, dramatic changes in C-CAM expression were observed in the VPs of the castrated animals (Figure 7). C-CAM staining appeared throughout the ducts and glands with the apical surfaces of the luminal cells showing the highest ¯uorescence intensity (Figure 7). Weak and heterogenous expression of CCAM in the basal cells similar to that observed in castrated DLP was also detected by double-labeling immuno¯uorescence with C-CAM antibodies and Mab K903 (Figure 7). Staining with isoform-speci®c antibodies revealed increased expression of both Lform and S-form C-CAM proteins in VPs from the castrated animals. Consistent with the results of the RNase protection assay, the S-form C-CAM was expressed at a higher level than the L-form, and the two forms were co-localized on both the apical and lateral surfaces of the degenerating luminal cells. Figure 7 illustrates the staining pattern of S-form CCAM in VP from a castrated animal. A similar localization was found with the L-form speci®c antibody but the intensity of labeling was much weaker (data not shown). These results suggest that androgen suppresses the expression of C-CAM isoforms in rat VP but not DLP and that the expression of both L- and S-form C-CAM in rat VP is regulated by androgen in a coordinated fashion. The changes in C-CAM expression levels were also examined with VP and DLP from animals castrated for 7 days, when castration-induced changes in prostate reached steady-state (Lee, 1981). The patterns of CCAM expression in these tissues were similar to those observed in the prostate tissues taken 3 days after castration (data not shown), suggesting that castration aects C-CAM expression in a consistent manner throughout the entire castration period. Discussion

Figure 4 Eect of androgen on the steady-state transcript levels of L-form and S-form C-CAM in DLP. RNA was obtained from the DLP of sham-operated animals (control), castrated animals (castrated), castrated animals treated with testosterone propionate (+TP), and castrated animals treated with testosterone propionate plus 4-hydroxy¯utamide (+TP/F). The RNase protection assay was described in Materials and methods

Several conclusions can be drawn from this study. First, marked lobe-speci®c variations in the expression of C-CAM isoforms are present in the rat prostate. In DLP, both isoforms were expressed with a S- to Lform ratio of about 5 : 1 but neither isoform was detected in the VP by RNase protection or immuno¯uorescence analysis. Second, C-CAM exhibits celltype speci®c expression in the adult DLP. In normal DLP, the two isoforms co-localized to the apical surfaces of luminal epithelial cells but were not expressed at detectable levels in basal cells. Third, lobe-speci®c variations exist in the regulation of CCAM expression by androgen. In castrated animals, the ratio of the S- to L-isoform and their levels of expression in the DLP were not signi®cantly altered by androgen deprivation. However, dramatic changes in the spatial distribution did occur as indicated by the appearance of C-CAM isoforms on lateral membranes of luminal cells and on basal cells. In contrast, in VP,


castration induced expression of the messages and proteins for both isoforms in a coordinated fashion and produced high S- to L-ratio similar to DLP, suggesting that the lack of detectable C-CAM messages in normal VP resulted from androgen suppression. Because S-form C-CAM lacks most of the tumorsuppressor domain and has no tumor-suppressive activity, its role in prostatic tumorigenesis remains unclear. It is thus intriguing that S- and L-form CCAM were co-expressed in adult DLP with a stable ratio of 5 : 1 (Figures 1, 2 and 3). Localization of Lform and S-form C-CAM in the same subpopulations of prostate cells suggests that they probably do not

function independently. Instead, this observation implies that a balance in L-form and S-form C-CAM levels may be important for maintaining the normal growth and dierentiation of prostate epithelia. In this regard, it is interesting that changes in the levels and ratios of L-form to S-form C-CAM occur in hepatocellular carcinoma which have lost growth control but not in regenerating liver, a state of regulated growth (Thompson et al., 1994). In addition, there is an optimal ratio of L-form to S-form CCAM required for suppressing the growth of a colon carcinoma cell line in vivo (Turbide et al., 1997). These observations suggest that even though S-form C-CAM

Figure 5 Altered distribution of C-CAM proteins in DP after androgen deprivation. Frozen sections of DPc (DP castrated) were double immunostained with Ab669, anti-CS or anti-CL (a, d and g) and Mab K903 (b, e and h). Panels c, f and i were photographed using a dual ®lter cube that shows both FITC and Texas Red ¯uorescence. (6006)

3257


3258

may not have tumor-suppressive activity by itself, its presence at the proper level may play an important role in regulating the tumor-suppressive activity of L-form C-CAM. The fact that every tissue examined thus far have expressed both isoforms argues in favor of this conclusion. Although the DLP in rats is anatomically and biochemically distinct from the VP, the physiological dierences of these lobes have not been clearly de®ned. In humans, prostate diseases are often manifested in a lobe-speci®c manner: benign prostate hyperplasia is often associated with hyperplastic growth in the transition zone, whereas prostate cancer and prostate intraepithelial neoplasia often occur in the peripheral zone (Bostwick et al., 1993; McNeal et al., 1988). The rat DLP is believed to closely resemble the outer region of the human prostate, i.e., the peripheral zone, where prostatic carcinoma frequently occurs (Price, 1963). Because C-CAM can suppress prostate tumorigenesis, it is tempting to speculate that the high level of CCAM expression in DLP is important for maintaining balanced growth and dierentiation. This hypothesis

should be able to be tested by modulating the expression of C-CAM in rat DLP. In fact, lobespeci®c expression of genes in mouse prostate using prostate speci®c promoters (e.g., the probasin promoter (Gingrich et al., 1996; Greenberg et al., 1995; Kasper et al., 1998; Yan et al., 1997) or the prostate-speci®c antigen promoter (Cleutjens et al., 1997; Wei et al., 1997)) was recently shown to be feasible. As C-CAM has a functional role in prostate growth control (Hsieh et al., 1995) and C-CAM was mainly expressed in DLP, where prostate cancer frequently occurs, further investigating the role of C-CAM in an in vivo transgenic mouse model will be important. In the rat, C-CAM expression is regulated by androgen only in the VP but not in the DLP. Since protein expression in both the VP (e.g., prostatic binding proteins (Parker and Scrace, 1979; Parker et al., 1980)) and the DLP (e.g., probasin (Matuo et al., 1985)) is controlled by androgen, it seems unlikely that dierential expression of androgen receptors is responsible. It seems more probable that these lobespeci®c dierences in regulation result from variations in the levels of transcription factors or transcription adapters (co-activators) that are expressed in both VP and DLP but have dierent responses to androgen regulation. C-CAM will be a useful model for studying dierential regulation by androgen in dierent prostate lobes. Because induction of prostate involution by androgen depletion is a common treatment of advanced prostate cancer, understanding the mechanisms underlying lobe-speci®c gene expression in prostate may allow us to target prostate cancers more eectively. Materials and methods Animal experiments Normal Sprague-Dawley rats were castrated for the experimental group. Sham-operated animals were used as controls. One day after the operation, the animals were divided into three groups and injected intraperitoneally daily with saline, 500 mg of testosterone propionate, or 500 mg of testosterone propionate plus 5 mg of 4-hydroxy¯utamide for 3 days. The animals were sacri®ced by carbon dioxide inhalation 24 h after the last injection. Ventral prostates (VPs), dorsal prostates (DPs), and lateral prostates (LPs) were removed for further analysis. For the time course of castration study, animals were castrated and the prostates were removed at 3 or 7 days post-castration. RNase protection analysis

Figure 6 Eect of androgen on the steady-state transcript levels of L-form and S-form C-CAM in VP. An RNase protection assay was performed on RNA samples prepared from VPs as described in Figure 4. Control, sham-operated animals; castrated, castrated animals; +TP, castrated animals treated with testosterone propionate; +TP/F, castrated animals treated with testosterone propionate plus 4-hydroxy¯utamide

Total RNA was extracted from tissue samples by the singlestep acid guanidinium-phenol-chloroform extraction method (Chomczynski and Sacchi, 1987). Considering the small size of the tissues, dorsal and lateral prostates were combined for RNA isolation. The integrity of the RNA was con®rmed by gel electrophoresis. Generation of a 219-bp antisense probe corresponding to the rat L-form C-CAM cytoplasmic domain and the RNase protection assay was performed as previously described (Cheung et al., 1993b). Antibodies Antibodies that have been previously shown to recognize epitopes on either the L-form C-CAM, S-form C-CAM, or


3259

Figure 7 Detection and localization of increased levels of C-CAM proteins in VP after androgen deprivation. Frozen sections of VPs from control (a ± c) and castrated (VPc) (d ± i) rats were double immunostained with Ab669 and Mab K903 (a ± c and d ± f) and with anti-CS and Mab K903 (g ± i) (6006). Panels c, f, and i are merged image of the same tissues stained with the antibodies. Arrowheads indicate C-CAM positive endothelial cells. (6006)

both were used for the immunohistochemical analyses of rat prostate tissues. These included: Ab669, produced by immunizing rabbits with puri®ed C-CAM protein (Lin et al., 1991), that recognizes both the L- and S- C-CAM isoforms; Mab 5.4, a mouse monoclonal antibody of the IgG1 subtype, that also recognizes both isoforms (Hixson and McEntire, 1989); anti-CL, which is a polyclonal rabbit anti-peptide antibody speci®c for the cytoplasmic domain of L-form C-CAM and was generated by immunizing rabbits with keyhold limpet hemacyanin-conjugated peptides (Lin et al., 1991). We also used a rabbit polyclonal antibody anti-CS, produced against the hexapeptide GGSGSF, which is found only in the cytoplasmic domain of S-form C-CAM

(Thompson et al., 1994), and monoclonal antibody K903 or 34bE12 (Keratin-903, isotype IgG, Enzo Diagnostics, Farmingdale, NY, USA), which is speci®c for high-molecularweight cytokeratins 1, 5, 10 and 14, and was used to identify the basal cells of rat prostate. Immunohistochemical analyses Immunohistochemical analyses were performed on frozen sections of rat VP or DLP. To increase the sensitivity of the immunohistochemical analysis, an indirect immunofluorescent (IF) technique was used. Four- to ®ve-micrometer whole-mount serial sections of DLP and VP were prepared


3260

from tissues frozen in OCT (Miles, Elkhart, IN, USA) in a dry ice-hexane bath. The sections were placed on Vectabond (Vector Laboratories, Inc., Burlingame, CA, USA)-coated glass slides, air dried for 5 min, and ®xed in 100% ice-cold acetone for 10 min. To simultaneously visualize epithelial structures of the DPs and LPs, transverse sections of the whole DLP were prepared. Structures of the VP were studied in a longitudinal plane. Antibodies diluted in phosphate-buered saline (PBS) at optimal concentrations, as determined from the rat liver section assays (Lin et al., 1991), were incubated with rat prostate tissue sections at room temperature. IF labeling using mouse monoclonal antibodies was performed by using the ABC kit (Vector Laboratories) as follows: The sections were sequentially blocked in 10% normal goat serum, avidin D, and biotin for 15 min each to decrease nonspeci®c staining. These sections were then incubated with mouse monoclonal antibodies at appropriate concentrations for 1 h. Unbound antibody was washed o with PBS for 45 ± 60 min, and the rat prostate sections were incubated for 45 min with a biotinylated goat anti-mouse immunoglobulin G secondary antibody (Vector) diluted 1 : 100. After three washes in PBS for 30 min each, these sections were incubated for 30 min with streptavidinTexas Red conjugate (Pierce Chemical Co., Rockford, IL, USA) diluted 1 : 750 and washed overnight in PBS at 48C. When rabbit polyclonal anti-C-CAM antibodies were used for IF labeling, the sections were blocked in 10% normal goat serum, incubated with the polyclonal antibodies at appropriate dilution, washed three times in PBS for 45 min each time, and incubated with ¯uorescein isothiocyanatelabeled goat anti-rabbit Ig antibody (Cappel) diluted 1 : 100. The staining was followed by an overnight wash in PBS at

48C. The sections were then covered with cover slips by using carbonate-buered glycerol containing propyl gallate as a mounting medium and photographed with a Nikon E800 microscope. Comparison of patterns of in situ reactivity of Mab 5.4 and Mab K903 and rabbit polyclonal antibodies Ab669, anti-CL, and anti-CS was accomplished by double-labeling immuno¯uorescence. The sections were ®rst labeled with the monoclonal antibody and then with speci®c rabbit antisera. To avoid misinterpretation of these double-labeling IF results, the sections were also labeled in the reverse order. In addition, antibody reactivities were compared on adjacent serial sections by single labeling. Negative controls consisted of the omission of the primary antibody replaced by PBS and of incubation with isotypic immunoglobulins.

Abbreviations C-CAM , cell-cell adhesion molecule; VP, ventral prostate; VPc, ventral prostate of castrated rat; DP, dorsal prostate; DPc, dorsal prostate of castrated rat; LP, lateral prostate; DLP, dorso-lateral prostate; IF, immuno¯uorescence; PBS, phosphate-buered saline; FITC, ¯uorescein-isothiocyanate. Acknowledgements This work was supported by Grant CA64856 (to S-HL) and CA42714 (to DCH) from National Institute of Health. Animal facility was supported by a core grant CA16672 from National Institute of Health to The University of Texas, MD Anderson Cancer Center.

References Bostwick DG, Amin MB, Dundore P, Marsh W and Schultz DS. (1993). Hum. Pathol., 24, 298 ± 310. Bussemakers MJ, van Moorselaar RJ, Giroldi LA, Ichikawa T, Issacs JT, Debruyne FM and Schalken JA. (1992). Cancer Res., 52, 2916 ± 2922. Cheung PH, Culic O, Qiu Y, Earley K, Thompson N, Hixson DC and Lin SH. (1993a). Biochem. J., 295, 427 ± 435. Cheung PH, Thompson NL, Earley K, Culic O, Hixson D and Lin SH. (1993b). J. Biol. Chem., 268, 6139 ± 6146. Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, 156 ± 159. Cleutjens KB, van der Korput HA, Ehren-van Eekelen CC, Sikes RA, Fasciana C, Chung LW and Trapman J. (1997). Mol. Endocrinol., 11, 1256 ± 1265. Cress AE, Rabinovitz I, Zhu W and Nagle RB. (1995). Cancer Metastasis Rev., 14, 219 ± 228. Gingrich JR, Barrios RJ, Morton RA, Boyce BF, DeMayo FJ, Finegold MJ, Angelopoulou R, Rosen JM and Greenberg NM. (1996). Cancer Res., 56, 4096 ± 4102. Greenberg NM, DeMayo F, Finegold MJ, Medina D, Tilley WD, Aspinall JO, Cunha GR, Donjacour AA, Matusik RJ and Rosen JM. (1995). Proc. Natl. Acad. Sci. USA, 92, 3439 ± 3443. Hixson DC and McEntire KD. (1989). Cancer Res., 49, 6788 ± 6794. Hixson DC, McEntire KD and Obrink B. (1985). Cancer Res., 45, 3742 ± 3749. Hsieh JT, Luo W, Song W, Wang Y, Kleinerman DI, Van NT and Lin SH. (1995). Cancer Res., 55, 190 ± 197. Kasper S, Sheppard PC, Yan Y, Pettigrew N, Borowsky AD, Prins GS, Dodd JG, Duckworth ML and Matusik RJ. (1998). Lab. Invest., 78, 319 ± 333. Kleinerman DI, Troncoso P, Lin SH, Pisters LL, Sherwood ER, Brooks T, von Eschenbach AC and Hsieh JT. (1995). Cancer Res., 55, 1215 ± 1220.

Kunath T, Ordonez-Garcia C, Turbide C and Beauchemin N. (1995). Oncogene, 11, 2375 ± 2382. Lee C. (1981). In: The Prostatic Cell: Structure and Function. Alan R. Liss, Inc: New York, pp. 145 ± 159. Lin SH, Culic O, Flanagan D and Hixson DC. (1991). Biochem. J., 278, 155 ± 161. Luo W, Wood C, Earley K, Hung M-C and Lin S-H. (1997). Oncogene, 14, 1697 ± 1704. Matuo Y, Nishi N, Muguruma Y, Yoshitake Y, Kurata N and Wada F. (1985). Biochem. Biophys. Res. Commun., 130, 293 ± 300. McNeal JE, Redwine EA, Freiha FS and Stamey TA. (1988). Am. J. Surg. Pathol., 12, 897 ± 906. Najjar SM, Accili D, Philippe N, Jernberg J, Margolis R and Taylor SI. (1993). J. Biol. Chem., 268, 1201 ± 1206. Neumaier M, Paululat S, Chan A, Matthaes P and Wagener C. (1993). Proc. Natl. Acad. Sci. USA, 90, 10744 ± 10748. Parker MG and Scrace GT. (1979). Proc. Natl. Acad. Sci. USA, 76, 1580 ± 1584. Parker MG, White R and Williams JG. (1980). J. Biol. Chem., 255, 6996 ± 7001. Price D. (1963). Natl. Cancer Inst. Monogr., 12, 351 ± 369. Thompson NL, Lin SH, Panzica MA and Hixson DC. (1994). Pathobiology, 62, 209 ± 220. Turbide C, Kunath T, Daniels E and Beauchemin N. (1997). Cancer Res., 57, 2781 ± 2788. Umbas R, Schalken JA, Adlders TW, Carter BS, Karthaus HF, Schaafsma HE, Debruyne FM and Isaacs WB. (1992). Cancer Res., 52, 5104 ± 5109. Wei C, Willis RA, Tilton BR, Looney RJ, Lord EM, Barth RK and Frelinger JG. (1997). Proc. Natl. Acad. Sci. USA, 94, 6369 ± 6374. Yan Y, Sheppard PC, Kasper S, Lin L, Hoare S, Kapoor A, Dodd JG, Duckworth ML and Matusik RJ. (1997). Prostate, 32, 129 ± 139.