The Prostate 38:268–277 (1999) Early Castration-Induced Upregulation of Transforming Growth Factor ␤1 and Its Receptors Is Associated With Tumor Cell Apoptosis and a Major Decline in Serum Prostate-Specific Antigen in Prostate Cancer Patients Pernilla Wikström,1 Patrick Westin,2 Pär Stattin,1 Jan-Erik Damber,1* and Anders Bergh2 1 2 Department of Urology and Andrology, Umeå, Sweden Department of Pathology, Umeå University, Umeå, Sweden BACKGROUND. The mechanism behind castration-induced apoptosis in prostate cells is unknown, but data from other species suggest that transforming growth factor ␤1 (TGF-␤1) may be involved. METHODS. By using quantitative RT-PCR and immunohistochemistry, expression of TGF␤1 and its receptors type I and II (RI and RII) was studied in normal and tumor areas of core biopsies taken before and 2–11 days after castration therapy. The TGF-␤ responses were related to changes in apoptotic index and to changes in serum prostate-specific antigen (PSA). RESULTS. In normal prostate tissue, apoptosis was generally increased by castration, and apoptosis was accompanied by an increase in TGF-␤1 and RII mRNA levels (P < 0.05). In tumors, apoptosis was seen only in 44% of the cases and in these, but not in the others, TGF-␤1, RI, and RII mRNA levels were increased (P < 0.05). In the patients showing a prognostically favorable PSA response (nadir PSA <5 ng/ml), but not in the others, RI and RII mRNA levels were significantly upregulated (P < 0.05). CONCLUSIONS. Short-term upregulation of TGF-␤1 and its receptors is associated with apoptosis in human prostate and prostate cancer, and possibly with a favorable clinical outcome after castration therapy. Prostate 38:268–277, 1999. © 1999 Wiley-Liss, Inc. KEY WORDS: TGF-␤1; apoptosis; prostate cancer; castration therapy INTRODUCTION Castration treatment has remained the first-line therapy for metastatic prostate cancer over the past 50 years , since it initially relieves clinical symptoms for most patients. However, almost all tumors relapse to an androgen-independent stage within a few months or up to several years later. This wide timespan in progression-free survival indicates tumor differences at the cellular level. At present, the best way to predict clinical outcome after castration is to measure the serum level of prostate-specific antigen (PSA) 3–6 months after therapy initiation [2,3]. Patients with © 1999 Wiley-Liss, Inc. normalized PSA values at that time (PSA-responders) have longer predicted progression-free survival times than non-PSA-responders. However, a recent study showed that it may be possible to predict clinical reGrant sponsor: Swedish Cancer Society; Grant number: Project 1760; Grant sponsor: University Hospital of Northern Sweden; Grant sponsor: Maud and Birger Gustavsson Foundation; Grant sponsor: Lions Cancer Research Foundation; Grant sponsor: Umeå University; Grant sponsor: Swedish Society for Medical Research. *Correspondence to: J.-E. Damber, Department of Urology and Andrology, Umeå University, 901 85 Umeå, Sweden. Received 27 March 1998; Accepted 30 July 1998 TGF-␤1 and Apoptosis in Prostate Cancer sponse at an earlier time point by examining shortterm cellular effects induced by castration therapy . In the rat ventral prostate (VP), androgen deprivation causes massive cell death within days due to induction of apoptosis [5,6]. Castration-induced apoptosis in the prostate is believed to be mediated by different intermediate factors. It has been suggested that transforming growth factor ␤1 (TGF-␤1) is a crucial factor for the apoptotic process in the VP . TGF-␤1 is a multifunctional growth factor with a number of effects in the development, differentiation, and growth of epithelial tissues . The expression of TGF␤1 and its receptors, TGF-␤ receptor type I and type II (RI and RII), is negatively regulated by androgens and increases during castration-induced apoptosis in the rat VP [9–11]. Moreover, TGF-␤1 has been shown to induce apoptosis in normal prostatic epithelial cells both in vivo and in vitro [12,13]. In the rat prostatic Dunning R3327 PAP adenocarcinoma, on the other hand, neither apoptosis nor TGF-␤1 or RII expression increase shortly after castration treatment [11,14,15]. Lack of TGF-␤ induction after castration has also been demonstrated in other Dunning tumor sublines . Taken together, these results suggest that castrationinduced apoptosis in androgen-dependent prostatic tissues involves a TGF-␤1 response that might be lost in androgen-independent prostatic tumors. In earlier studies, we found that human prostate tumors respond to castration in a highly heterogeneous way. Only approximately one third of tumors, mainly belonging to PSA-responding patients, showed increased apoptotic index 1 week after treatment [4,17]. This could be compared to a sevenfold increase in apoptotic cells in adjacent normal prostatic glands . However, it is not known if castrationinduced apoptosis in human prostate and prostate tumors involves a TGF-␤ response. Neither is it known if the lack of short-term apoptosis induction in some prostate tumors after androgen withdrawal could be due to some failure in mediating such a TGF-␤ response. Human prostate tumors often express high levels of TGF-␤1 [18,19], and overproduction of TGF␤1 has been shown to be associated with angiogenesis, metastasis, and short survival in prostate cancer . This may be due to possible tumor-stimulating properties of TGF-␤1 such as inhibition of immune responses  and stimulation of angiogenesis and cell motility [22,23]. Furthermore, low expression of RI and RII has been associated with short prostate cancer-specific survival [20,24], which may indicate some failure of prostate tumor epithelial cells in responding to TGF-␤1 growth-inhibiting and/or apoptotic signals. The aim of this study was to examine if short-term 269 effects on TGF-␤1, RI, and RII expression are associated with apoptosis and clinical response after castration therapy in patients with advanced prostate cancer. MATERIALS AND METHODS Tissue Collection and Processing Several ultrasound-guided core biopsies were taken shortly before and 2–11 days after castration therapy in a series of patients with advanced prostate cancer. Local tumor stage was evaluated by digital rectal examination, according to the 1992 UICC classification , and radionuclide bone scan was performed for metastasis staging at the time of diagnosis (Table I). Biopsies were immediately frozen in liquid nitrogen or fixed in phosphate-buffered formalin, before being embedded in TissueTek (Miles, Inc., Elkhart, IN) and paraffin, respectively. The biopsies were cut into 4-m-thick sections and stained with hematoxylin and eosin (HE). The tumors were classified into high (G1), moderate (G2), and low (G3) differentiation by evaluating the HE-stained slides from the fixed biopsies, according to the World Health Organization (WHO) classification system (Table I) . The HE-stained cryosections were used for localizations of normal areas (biopsy areas where no malignant cells were detected) and tumor areas (Fig. 1). Biopsy parts corresponding to these areas were microdissected from the frozen biopsies, by using a sterile scalpel blade, and further processed for total RNA extraction. By using this procedure, total RNA from normal and tumor prostate tissue were isolated from 9 and 18 pairs of frozen pre- and posttherapy biopsies, respectively. Clinical Classification Response to castration therapy was defined, as previously described , as a serum level of PSA of 艋5 ng/ml 3–6 months posttreatment (PSA-responders), and nonresponse as a nadir PSA 艌10 ng/ml (nonPSA-responders, Table I). Determination of Apoptotic Indexes Apoptotic indexes (number of apoptotic cells per 1,000 cells) in the HE-stained sections from the frozen tumors were determined by counting approximately 250–1,000 and 300–2,500 normal and tumor cells, respectively, at 400× magnification. Apoptotic cells were defined as single rounded cells or fragments with 270 Wikström et al. TABLE 1. Clinical Characteristics of Patients With Advanced Prostate Cancer Treated With Castration Therapy and Included in the mRNA Experiments Tumor stagec T1–T2 T3–T4 Tumor graded G1 G2 G3 Metastasis (bone scan) PSA before therapye PSA nadir AI before therapy AI after therapy Time for biopsyf Ap-responders (n = 8)a Non-Ap-responders (n = 10) PSA-responders (n = 10)b Non-PSA-responders (n = 7) 1 7 2 8 2 8 1 6 1 4 3 7 1,400 (1,000) 26 (21) 8.8 (1.5) 19 (3.6) 6.4 (1.1) 5 5 8 550 (210) 25 (11) 17 (2.0) 13 (1.8) 6.3 (0.47) 1 5 4 7 1,100 (850) 2.3 (0.44) 12 (1.5) 14 (1.9) 5.6 (0.92) 3 4 7 720 (230) 61 (23) 14 (3.4) 16 (4.4) 7.3 (0.29) a Apoptotic (Ap) response defined as increased apoptotic index (AI) in post- compared to pretherapy biopsy. Indexes calculated as number of apoptotic cells per 1,000 cells in HE-stained sections. b Response defined as serum prostate-specific antigen (PSA) 艋5 ng/ml, and nonresponse as PSA 艌10 ng/ml 3–6 months after therapy. One patient had a nadir PSA value of 8 ng/ml. c According to UICC . d According to WHO . e Values expressed as means and SEM (in parentheses). f Days between therapy and posttherapy biopsy. Total RNA Preparation and Competitive Reverse Transcription-Polymerase Chain Reaction (RT-PCR) Fig. 1. Identification of normal (N) and tumor (T) areas in HEstained cryosection from a pretherapy prostate cancer core biopsy (×200). densely aggregated chromatin and condensed cytoplasm, often lying in ‘‘halos’’ of extra cellular space (Fig. 2) . If more than one apoptotic body was seen per ‘‘halo,’’ these were considered to originate from the same cell and were counted as one. Apoptotic response to castration therapy was defined as increased apoptotic index in the posttherapy biopsy compared to the pretherapy biopsy (Apresponders), and apoptotic nonresponse as unchanged or decreased apoptotic index after treatment (non-Apresponders, Table I). Total RNA was isolated from the frozen prostatic tissues by the TRIzol extraction method (Life Technologies AB, Täby, Sweden). The mRNA levels for TGF-␤1, RI, RII, and the housekeeping gene cyclophilin  were quantified, by using competitive RT-PCR and PCR primers as previously described [11,15]. Briefly, 50 ng of total RNA were reverse transcribed together with appropriate amounts of internal RNA standards (IS) [11,15] for TGF-␤1, RI, RII, and cyclophilin. Each RNA sample was titrated with three amounts (double samples) of IS, in ranges of 8.5–170, 0.52–10, 2.3–46, and 34–680 amol for TGF-␤1, RI, RII, and cyclophilin, respectively. After RT completion, samples were divided into four PCR tubes, in order to amplify cDNA for TGF-␤1 (248 bp), RI (177 bp), RII (215 bp), and cyclophilin (362 bp) separately. During 32 cycles of PCR (95°C, 30 sec; 63°C, 30 sec; and 72°C, 45 sec), the TGF-␤ templates were competitively amplified with cDNA for their corresponding IS (266, 163, 199, and 315 bp, respectively). Resulting PCR products were analyzed in an automatic laser fluorescence system (ABI PRISM™ 377 DNA sequencer, Perkin Elmer, Askim, Sweden). The data were processed by the ABI PRISM™ GeneScan software (Perkin Elmer), and RNA levels were calculated from tem- TGF-␤1 and Apoptosis in Prostate Cancer 271 TABLE II. Clinical Characteristics of Patients With Advanced Prostate Cancer Treated With Castration Therapy and Included in the Immunohistochemistry Experiments Tumor stageb T1–T2 T3–T4 Tumor gradec G1 G2 G3 Metastasis (bone scan) PSA before therapyd PSA nadir Time for biopsye PSAresponders (n = 12)a Non-PSAresponders (n = 9) 4 8 1 8 3 7 2 6 770 (720) 2.0 (0.38) 7.0 (0.64) 2 4 3 8 1,600 (460) 84 (22) 6.8 (0.40) a Response defined as serum prostate-specific antigen (PSA) 艋5 ng/ml, and nonresponse as PSA 艌10 ng/ml 3–6 months after therapy. One patient had a nadir PSA value of 8 ng/ml. b According to UICC . c According to WHO . d Values expressed as means and SEM (in parentheses). e Days between therapy and posttherapy biopsy. Fig. 2. Cryosections showing castration-induced apoptosis (arrows) in tumor epithelial cells of a prostate cancer core biopsy (×400). A: Tumor cells before therapy. B: Tumor cells after therapy. plate- to IS-cDNA ratios, as previously described . The TGF-␤1, RI, and RII levels were corrected for the corresponding cyclophilin levels in each RNA sample and expressed as relative levels (amol/amol cyclophilin mRNA) in the resulting figures. Immunohistochemistry In addition to quantify the mRNA levels for TGF-␤1 and its receptors before and after castration, we wanted to use immunohistochemistry (IHC) to examine TGF-␤1, RI, and RII protein expression in response to castration therapy. Unfortunately, the biopsy material used in the mRNA experiment was not sufficient for these studies, and we had to use core biopsies from another series of patients. The two series of patients were similar according to tumor and metastasis classification and PSA levels before and after therapy, and the biopsies were collected and processed in the same way (Tables I and II). Four-micron sections from formalin-fixed tumor biopsies were deparaffinated and rehydrated according to standard procedures, washed with phosphatebuffered saline (PBS), and heated in a microwave oven at 600 W for 2 × 7.5 min and 1 × 5 min in 0.01 M citrate buffer, pH 6.0, as earlier described . To quench endogenous peroxidase activity, slides were immersed in 3% H2O2 in methanol for 20 min. Slides were incubated overnight, at 4°C, with the following antibodies: anti-TGF-␤1 (10 g/ml, anti-TGF-␤1 neutralizing antibody, R&D Systems, Oxon, UK), anti-RI (0.25 g/ml, V-22, Santa Cruz Biotechnology, Santa Cruz, CA), and anti-RII (1 g/ml, anti-human TGF-␤ RII neutralizing antibody, R&D Systems). For detection, the ABC technique was used, with aminoethylcarbazole as chromogen, according to the manufacturer’s instructions (Vector Laboratories, Burlingame, CA). Sections were counterstained with Mayer’s hematoxylin solution. Specificity of TGF-␤ immunoreactions was examined by preincubation of the primary antibodies with a 25-fold (w/w) excess of the corresponding control peptides (recombinant human TGF␤1 and TGF-␤1 soluble receptor II, R&D Systems, and V-22P, Santa Cruz Biotechnology). The immunoreactivity in tumor and normal areas 272 Wikström et al. TABLE III. Short-Term Effects Induced by Castration Therapy on Apoptosis and TGF-␤1, RI, and RII mRNA Levels in Patients With Advanced Prostate Cancer Normal tissue (n = 9) AIa TGF-␤1b RI RII Time for biopsyc Tumors (n = 18) Untreated mean (SEM) Castrated mean (SEM) Untreated mean (SEM) Castrated mean (SEM) 5.0 (0.64) 0.12 (0.017) 0.022 (0.0034) 0.23 (0.085) 14.0 (2.2)* 0.37 (0.15)* 0.060 (0.032) 0.71 (0.36) 7.0 (0.65) 13.0 (1.6) 0.11 (0.012) 0.014 (0.0023) 0.094 (0.014) 16.0 (2.0) 0.21 (0.045)* 0.020 (0.0025) 0.17 (0.032)** 6.3 (0.55) a Apoptotic indexes (AI) calculated as number of apoptotic cells per 1,000 cells in HEstained sections. b Relative mRNA levels expressed as amol/amol cyclophilin mRNA (see Materials and Methods). c Days between therapy and post-therapy biopsy. *P < 0.05, significant increase after castration therapy. **P < 0.01, significant increase after castration therapy. was evaluated without knowledge of any patient data. Immunoreactivities for TGF-␤1, RI, and RII were classified as negative (−), moderate (+), or intense (++), and assessed as either increased or unchanged by castration therapy. Statistics Groups were compared by the Mann-Whitney Utest and paired observations by the Wilcoxon paired test. To test correlations, the Spearman rank sum test was applied. P 艋 0.05 was considered statistically significant. RESULTS Effects of Castration Therapy on Apoptosis, and on TGF-␤1, RI, and RII Expression Apoptotic indexes were determined and TGF-␤1, RI, and RII mRNA levels were quantified in normal and tumor parts of prostate cancer biopsies taken before and 2–11 days after castration therapy (Table III). Furthermore, TGF-␤1 and TGF-␤ receptor protein expression was examined in pre- and posttherapy biopsies by using IHC (Table IV). In normal prostatic tissue, apoptotic indexes were increased after castration in 8 of 9 cases (89%), and TGF-␤1, RI, and RII mRNA levels were increased in 7 (78%), 5 (56%), and 7 (78%) cases, respectively. In the one normal case where no induction of apoptosis was seen after treatment, there was also no increase in TGF-␤1 or TGF-␤ receptor mRNA levels. The average TGF-␤1 mRNA levels and apoptotic indexes were significantly higher in the post- than in the pretherapy biopsies (3.1- and 2.8-fold; P = 0.028 and 0.011, respectively), while the increase in RI and RII mRNA levels were nonsignificant (P = 0.21 and 0.14, Table III). In the tumor tissue, apoptotic response to castration was more heterogeneous than in normal tissue, with only 8 of 18 cases (44%) showing increased apoptotic indexes after therapy. For the whole group, castration did not induce a significant increase in apoptosis. The TGF-␤1 and RI mRNA levels were increased by castration in 12 (67%, P = 0.031 and 0.102, respectively), and the RII mRNA levels in 13 (72%, P = 0.006), of the tumors (Table III). Immunoreactivity for TGF-␤1, RI, and RII was found mainly in normal and tumor epithelial cells (Fig. 3), which is in line with recent results demonstrating protein expression of TGF-␤1 and its receptors, and mRNA expression of TGF-␤1, preferentially in prostate epithelial cells . The epithelial immunostaining in the pretherapy biopsies was classified as negative (−), moderate (+), or intense (++), and was found to be unchanged or intensified (Fig. 3) in the posttherapy biopsies (Table III). In accordance with the mRNA results (Table III), protein levels of TGF-␤1, RI, and RII seemed to be heterogeneously upregulated by androgen ablation in the prostate tissues (Table IV). TGF-␤1 and TGF-␤ receptor protein induction were more frequently seen in normal than in tumor tissue (Table IV). The specificity of the immunoreactions has previously been determined , and control slides incubated with preblocked antibodies showed no staining (results not shown). TGF-␤1 and Apoptosis in Prostate Cancer 273 TABLE IV. Short-Term Effects Induced by Castration Therapy on TGF-␤1, RI, and RII Immunoreactivity in Patients With Advanced Prostate Cancer Increased immunoreactivity after therapy TGF-␤ RI RII No. of cases (normal/tumor) Time for biopsy (normal/tumor)a Normal tissue no. (%) Tumor no. (%) 13/21 13/21 16/22 6.5 (0.27)/6.8 (0.44) 7.3 (0.36)/6.7 (0.42) 7.3 (0.55)/7.0 (0.33) 11 (85) 9 (69) 10 (63) 9 (43) 11 (52) 12 (55) a Days between therapy and posttherapy biopsy. Values expressed as means and SEM (in parentheses). TGF-␤1, RI, and RII mRNA Expression in Relation to Castration-Induced Apoptosis Normal tissue with increased apoptotic index after castration showed three- and fivefold inductions of TGF-␤1 and RII mRNA, respectively (P = 0.025, Fig. 4A). The RI mRNA level was increased twofold, but this was not statistically significant (P = 0.123, Fig. 4A). In the tumors, the relative changes in TGF-␤1, RI, and RII mRNA levels observed after castration therapy were correlated to the corresponding changes in apoptosis indexes (rs = 0.60, 0.63, and 0.61; P = 0.016, 0.005, and 0.008, respectively). TGF-␤1, RI, and RII mRNA levels were increased in the Ap-responding tumors after castration (2.5-, 1.8-, and 2.3-fold; P = 0.036, 0.036, and 0.017, respectively, Fig. 4B), while neither TGF-␤1 nor the TGF-␤ receptor mRNA levels were significantly increased in the non-Ap-responding tumors (Fig. 4C). TGF-␤1, RI, and RII mRNA Expression in Relation to PSA Response After Castration Therapy There was no significant correlation found between the relative changes in tumor TGF-␤1, RI, or RII mRNA levels and the corresponding changes in patient PSA levels after castration therapy. In accordance with the Ap-responding patients, however, the PSAresponding patients showed increased RI and RII mRNA levels after treatment (1.7- and 1.5-fold, P = 0.047 and 0.022, respectively, Fig. 5A). There was also a tendency for TGF-␤1 induction in this group, although the increase was nonsignificant (P = 0.114, Fig. 5A). Neither TGF-␤1, nor the TGF-␤ receptor mRNA levels, were significantly increased in the non-PSAresponding tumors after therapy (Fig. 5B). DISCUSSION Androgen ablative therapy today, due to its ability to initially relieve clinical symptoms for most patients, is the most frequently used method in the treatment of advanced prostate cancer. The beneficial therapy responses are believed to be achieved by castrationinduced apoptosis and decreased cell proliferation among androgen-dependent epithelial cells [5,6,30]. Only a minority of human prostate tumors, however, respond with increase in apoptosis. Recent studies have shown increased, unchanged, or even decreased apoptosis in human prostate cancer 1 week after castration , and this heterogeneity in short-term apoptotic response was demonstrated to be differential for subsequent clinical outcome . The molecular mechanism for castration-induced apoptosis in prostate epithelial cells is not clear, but it has been thought to involve a TGF-␤1-dependent pathway (see Introduction). Several studies have demonstrated a relationship between TGF-␤1 and castration-induced apoptosis in animal models [9,10,30,31]. However, the present study is the first to report short-term effects on TGF␤1, RI, and RII expression in the human prostate and advanced prostate cancer after castration. Induction of TGF-␤1, RI, and RII expression was found to be related to the apoptotic response in tumor cells after castration therapy. The TGF-␤1, RI, and RII mRNA levels were significantly increased in the Apresponding, but not in the non-Ap-responding tumors after treatment. This is in line with previous results showing increased expression of TGF-␤1 and its receptors in apoptosis-responding model systems, such as the rat VP and the human PC-82 tumor [9,10,30,31], but not in apoptosis-nonresponding Dunning tumors after castration [11,15,16]. Taken together, these results suggest that TGF-␤1 is involved in mediating castration-induced apoptosis in the prostate, and furthermore, that androgen-independent tumors may evade castration-induced apoptosis due to defects in their TGF-␤1 response to androgen ablation. Defects in TGF-␤ receptor expression have been associated with TGF-␤1 insensitivity in many systems, 274 Wikström et al. Fig. 3. Sections from Ap-responding tumors before (A, C, E) and after (B, D, F) castration therapy, showing increased immunoreactivity for TGF-␤1 (A, B), RI (C, D), and RII (E, F) after castration (×400). including the prostate [32–36], and loss of epithelial expression of RI and RII has been shown with human prostate cancer progression [37,38]. Lack of RI or RII expression in the epithelial cells would be one obvious explanation for the absence of castration-induced TGF-␤1 response in advanced prostate cancer. In this study, however, TGF-␤ receptor expression in the intact tumors was not predictable for the apoptotic re- sponse to castration. All untreated tumors expressed similar amounts of RI and RII mRNA, and the Apresponding group contained intact tumors with negative as well as intense receptor immunoreactivity. Our results suggest that the ability of androgen-independent tumors to avoid castration-induced apoptosis is not simply due to loss of RI or RII expression in the tumor cells, but may be influenced by the inability of TGF-␤1 and Apoptosis in Prostate Cancer 275 Fig. 5. Competitive RT-PCR results, showing relative TGF-␤1, RI, and RII mRNA levels before (open columns) and after (shaded columns) castration therapy in PSA-responding (A) and non-PSAresponding (B) prostate tumors. RI and RII mRNA levels were increased by castration in the PSA-responding tumors (*P < 0.05). No significant induction of TGF-␤1, RI, or RII was seen in the non-PSA-responding tumors after castration. Fig. 4. Competitive RT-PCR results, showing relative TGF-␤1, RI, and RII mRNA levels before (open columns) and after (shaded columns) castration therapy in Ap-responding normal prostate areas (A), and in Ap-responding (B) and non-Ap-responding (C) prostate tumors. TGF-␤1 and RII mRNA levels were increased by castration in the Ap-responding normal areas, and the TGF-␤1, RI, and RII mRNA levels were increased in the Ap-responding tumors (*P < 0.05). No significant induction of TGF-␤1, RI, or RII was seen in the non-Ap-responding tumors after castration. some tumors to upregulate the expression of these receptors in response to castration. The TGF-␤1 signaling pathway in prostate cancer cells therefore needs to be thoroughly investigated. In addition to being associated with apoptosis, the tumor TGF-␤ response to castration therapy seemed to be predictable for the subsequent PSA-response. The RI and RII mRNA levels were significantly increased by castration in the group of PSA-responding tumors, but not in the group of non-PSA-responding tumors. These results are consistent with results by Stattin et al. , indicating the possibility of predicting clinical outcome at an early time-point by studying short-term effects of castration therapy. There was, however, no significant induction of TGF-␤1 in the PSAresponding tumors after castration, although an overall induction of TGF-␤1 was seen in 67% of the tumor cases. Speculatively, induction of TGF-␤1 in tumor cells that do not go into apoptosis after castration may be nonbeneficial to the patient, due to tumorpromoting effects of TGF-␤1 such as inhibition of immune responses  and stimulation of angiogenesis [20,22], cell motility , and metastasis . Apoptosis was more frequently seen in normal prostate tissue than in tumors after castration. Only one normal case showed lack of castration-induced apoptosis, and in this tissue there were also no increases in TGF-␤1 or TGF-␤ receptor expression. The average increase of TGF-␤1, RI, and RII mRNA in the normal Ap-responding tissue was more pronounced 276 Wikström et al. than the increase in the Ap-responding tumors, although the increase in RI mRNA was not statistically significant. Lack of statistical confirmation of RI induction in this group, as well as the big standard deviations seen for TGF-␤1, RI, and RII in normal tissue, suggest that normal, nonmalignant prostate tissue could be rather heterogeneous. Possible reasons for this are premalignant cellular defects in normallooking epithelial cells, but also regional differences in normal epithelial cell function and/or functional changes related to the proximity of cancer tissue. 11. 12. 13. 14. CONCLUSIONS The present study suggests that increased expression of TGF-␤1, RI, and RII is associated with castration-induced apoptosis in the human prostate and in advanced prostate cancer. Moreover, short-term effects on RI and RII seem to be predictable for PSA response and thus probably also for the long-term clinical outcome after castration therapy. ACKNOWLEDGMENTS Mrs. Birgitta Ekblom, Mrs. Elisabeth Dahlberg, and Ms. Pernilla Andersson have skillfully contributed to this paper by their technical assistance. 15. 16. 17. 18. 19. REFERENCES 20. 1. Huggins C, Stevens REJ, Hodges CV. Studies on prostate cancer. II. The effects of castration on advanced carcinoma of the prostate. Arch Surg 1941;43:209–223. 2. Fowler JJE, Pandey P, Seaver LE, Feliz TP, Braswell NT. Prostate specific antigen regression and progression after androgen deprivation for localized and metastatic prostate cancer. J Urol 1995;153:1860–1865. 3. Petros JA, Andriole GL. Serum PSA after antiandrogen therapy. Urol Clin North Am 1993;20:749–756. 4. Stattin P, Westin P, Damber J-E, Bergh A. Short-term cellular effects induced by castration therapy in relation to clinical outcome in prostate cancer. Br J Cancer 1998;77:670–675. 5. Kerr JFR, Searle JW. Deletion of cells by apoptosis during castration-induced involution of the rat prostate. Virchows Arch [B] 1973;13:87–102. 6. Kyprianou N, Isaacs JT. Activation of programmed cell death in the rat ventral prostate after castration. Endocrinology 1988;122: 552–562. 7. Isaacs JT, Lundmo PI, Berges R, Martikainen P, Kyprianou N, English HF. Androgen regulation of programmed death of normal and malignant prostatic cells. J Androl 1992;13:457–464. 8. Massagué J. The transforming growth factor-␤ family. Annu Rev Cell Biol 1990;6:597–641. 9. Kyprianou N, Isaacs JT. Expression of transforming growth factor-␤ in the rat ventral prostate during castration induced programmed cell death. Mol Endocrinol 1989;3:1515–1522. 10. Kim IY, Ahn H-J, Zelner DJ, Park L, Sensibar JA, Lee C. Expression and localization of transforming growth factor-␤ receptors 21. 22. 23. 24. 25. 26. 27. 28. type I and type II in the rat ventral prostate during regression. Mol Endocrinol 1996;10:107–115. Wikström P, Bergh A, Damber J-E. Expression of transforming growth factor-␤ receptor type I and type II in rat ventral prostate and Dunning R3327 PAP adenocarcinoma in response to castration and oestrogen treatment. Urol Res 1997;25:103–111. Martikainen P, Kyprianou N, Isaacs JT. Effect of transforming growth factor-␤1 on proliferation and death of rat prostatic cells. Endocrinology 1990;127:2963–2968. Sutkowski DM, Fong C-J, Sensibar JA, Rademaker ED, Sherwood ER, Kozlowski JE, Lee C. Interaction of epidermal growth factor and transforming growth factor beta in human prostatic epithelial cells in culture. Prostate 1992;21:133–143. Brändström A, Westin P, Bergh A, Cajander S, Damber J-E. Castration induces apoptosis in the ventral prostate but not in an androgen-sensitive prostatic adenocarcinoma in the rat. Cancer Res 1994;54:3594–3600. Lindström P, Bergh A, Holm I, Damber J-E. Expression of TGF␤1 in rat ventral prostate and Dunning R3327 PAP prostate tumor after castration and estrogen treatment. Prostate 1996;29: 209–218. Steiner MS, Zhou Z-Z, Tonb DC, Barrack ER. Expression of transforming growth factor-␤1 in prostate cancer. Endocrinology 1994;13:2240–2247. Westin P, Stattin P, Damber J-E, Bergh A. Castration therapy rapidly induces apoptosis in a minority and decreases cell proliferation in a majority of human prostatic tumors. Am J Pathol 1995;146:1368–1375. Thompson TC, Truong LD, Timme TL, Kadmon D, McCune BK, Flanders KC, Scardino PT, Park SH. Transforming growth factor ␤1 as a biomarker for prostate cancer. J Cell Biochem 1992;16: 54–61. Truong LD, Kadmon D, McCune BK, Flanders KC, Scardino PT, Thompson TC. Association of transforming growth factor-␤1 with prostate cancer: an immunohistochemical study. Hum Pathol 1993;24:4–9. Wikström P, Stattin P, Franck-Lissbrant I, Damber J-E, Bergh A. Transforming growth factor ␤1 is associated with angiogenesis, metastasis, and poor clinical outcome in prostate cancer. Prostate 1998;37:19–29. Torre-Amione G, Beauchamp RD, Koeppen H, Park BH, Schreiber H, Moses HL, Rowley DA. A highly immunogenic tumor transfected with a murine transforming growth factor type ␤1 cDNA escapes immune surveillance. Proc Natl Acad Sci USA 1990;87:1486–1490. Roberts AB, Sporn MB, Assoian RK, Smith JM, Roche NS, Wakenfield LM, Heine UI, Liotta LA, Falanga V, Kehrl JH, Fauci AS. Transforming growth factor type ␤: rapid induction of fibrosis and angiogenesis in vivo and stimulation of collagen formation in vitro. Proc Natl Acad Sci USA 1986;83:4167–4171. Yang EY, Moses HL. Transforming growth factor ␤1-induced changes in cell migration, proliferation, and angiogenesis in the chicken chorioallantoic membrane. J Cell Biol 1990;111:731–741. Kim IY, Lang S, Oefelein MG, Oyasu R, Kozlowski JM, Lee C. Loss of expression of transforming growth factor-␤ receptors is a poor prognostic indicator in prostate cancer patients. J Urol [Suppl] 1997;157:445. UICC. TNM classification of malignant tumours, 4th ed. Berlin: Springer; 1992. Mostofi FK, Sesterhenn IA, Sobin LH. International histological classification of prostate tumours. Geneva: WHO; 1980. Kerr JFR, Wyllie AH, Currie AR. Apoptosis: a basic biological phenomenon with wide ranging implications in tissue kinetics. Br J Cancer 1972;26:239–257. Danielson PE, Forss-Petter S, Brow MA, Calavetta L, Douglass J, TGF-␤1 and Apoptosis in Prostate Cancer 29. 30. 31. 32. 33. 34. Milner RJ, Sutcliffe JG. p1B15: a cDNA clone of the rat mRNA encoding cyclophilin. DNA 1988;7:261–267. Taylor CR, Shi S-R, Chaiwun B, Young L, Imam SR, Cote RJ. Strategies for improving the immunohistochemical staining of various intranuclear prognostic markers in formalin-paraffin sections. Hum Pathol 1994;25:263–270. Kyprianou N, English HF, Isaacs JT. Programmed cell death during regression of PC-82 human prostate cancer following androgen ablation. Cancer Res 1990;50:3748–3753. Kyprianou N, Isaacs JT. Identification of a cellular receptor for transforming growth factor-␤ in rat ventral prostate and its negative regulation by androgens. Endocrinology 1988;123: 2124–2131. Kalkhoven E, Roelen BAJ, de Winter JP, Mummery CL, van den Eijden-van Raaij AJM, van der Saag PT, van der Burg B. Resistance to transforming growth factor ␤ and activin due to reduced receptor expression in human breast tumor cell lines. Cell Growth Differ 1995;6:1151–1161. Kim IY, Ahn H-J, Zelner DJ, Shaw JW, Sensibar JA, Kim J-H, Kato M, Lee C. Genetic change in transforming growth factor ␤ (TGF-␤) receptor type I gene correlates with insensitivity to TGF-␤1 in human prostate cancer cells. Cancer Res 1996;56:44– 48. Park K, Kim S-J, Bang Y-J, Park J-G, Kim NK, Roberts A, Sporn 35. 36. 37. 38. 277 MB. Genetic changes in the transforming growth factor ␤ (TGF␤) type II receptor gene in human gastric cancer cells: correlation with sensitivity to growth inhibition by TGF-␤. Proc Natl Acad Sci USA 1994;91:8772–8776. Sun L, Wu G, Willson JKV, Zborowska E, Yang J, Rajkarunanayake I, Wang J, Gentry LE, Wang X-F, Brattain MG. Expression of transforming growth factor ␤ type II receptor leads to reduced malignancy in human breast cancer MCF-7 cells. J Biol Chem 1994;269:26449–26455. Wang J, Sun LZ, Myeroff L, Wang X, Gentry LE, Yang J, Liang J, Zborowska E, Markowitz S, Willson JKW, Brattain MG. Demonstration that mutation of the type II transforming growth factor ␤ receptor inactivates its tumor suppressor activity in replication error-positive colon carcinoma cells. J Biol Chem 1995; 270:22044–22049. Guo Y, Jacobs SC, Kyprianou N. Down-regulation of protein and mRNA expression for transforming growth factor-␤ (TGF␤1) type I and type II receptors in human prostate cancer. Int J Cancer 1997;71:573–579. Kim IY, Ahn H-J, Zelner DJ, Shaw JW, Lang S, Kato M, Oefelein MG, Miyazono K, Nemeth JA, Kozlowski JM, Lee C. Loss of expression of transforming growth factor-␤ type I and type II receptors correlates with tumor grade in human prostate cancer tissues. Clin Cancer Res 1996;2:1255–1261.