Effects of estrogen and androgen on the ultrastructure of secretory granules and intercellular junctions in regressed canine prostate.код для вставкиСкачать
THE ANATOMICAL RECORD 197~111-132 (1980) Effects of Estrogen and Androgen on the Ultrastructure of Secretory Granules and Intercellular Junctions In Regressed Canine Prostate FREDERICK B. MERK, IRWIN LEAV, PAUL W.L. KWAN, AND PETER OFNER Departments of Pathology, Anatomy, and Urology.Tufts University School of Medicine, Boston, Massachusetts, 0211I ; Laboratory of Pharmacology, Harvard School of Dental Medicine, Boston, Massachusetts, 021 15; and Steroid Biochemistry Laboratory, Lemuel Shattuck Hospital, Boston, Massachusetts, 02130 ABSTRACT Epithelial cells in the prostate of the castrated or hypophysectomized dog were studied by thin-section and freeze-fracture electron microscopy to determine in vivo responses to estradiol-17P 17-cyclopentylpropionate (ECP) and testosterone cyclopentylpropionate (TCP). Particular attention was given to changes in specific organelles and intercellular junctions that might reflect hormone action. The few secretory granules that remain in the regressed epithelium (vestigial granules) serve as markers of prior androgen responsiveness. Pharmacologic doses of ECP caused regressed glandular cells to acquire a novel phenotype. Characteristic features of these estrogen-modified glandular (EMG)cells are newly formed secretory granules and tonofilament bundles that coexist with vestigial granules, thus demonstrating bipotentiality of response. Glandular cell-tight junctions appear unaltered by the endocrine manipulations. Although EMG cells have squamous cell features, tight junctions remain intact. Desmosomes in the canine prostate are dimorphic and are classified 70F and l O O F according to the width of the filaments that converge on the dense plaques. In intact dogs, l O O F desmosomes are associated with basal-reserve cells, whereas only the 70F variety is found between glandular cells. TCP treatment does not alter this distribution. Following ECP administration, both 70F and l O O F desmosomes are present between EMG cells. The coexistence of newly formed secretory granules and tonofilaments of lOOF desmosomes in the same EMG cell represents estrogen-induced bidirectional differentiation. Our findings indicate that androgens and estrogens are individually capable of controlling the expression of secretory granules and desmosomes. In intact animals, male and female sex hormones may act in concert to direct epithelial cell differentiation of the prostate. Differentiation and function of prostatic cells have traditionally been considered to be exclusively androgen-directed processes. It is now well documented that in a number of species the gland responds to pharmacologic doses of estrogens by undergoing a proliferative change termed squamous metaplasia (Triche and Harkin, 1971;Kroes and Teppema, 1972;Ofner et al., 1974;Kjaerheim et al., 1974;Helpap and Steins, 1975; Hohbach, 1977). Although unphysiologic blood levels of female hormone are required to induce the squamous metaplasia, the occurrence of this abnormal differentiation of the prostatic epithelium suggests that estro- 0003-276X/80/1972-0111$03.800 1980 ALAN R. LISS, INC. gen action may also be involved in androgendependent normal epithelial growth and in promoting growth in benign prostatic hyperplasia and prostatic carcinoma. Estrogen receptors have been identified in the cytosols of hyperplastic (Hawkins e t al., 1976) and carcinomatous (Bashirelahi and Young, 1976) human prostate. With these considerations in mind, we focused our attention on the effects of estrogen on prostatic epithelial cell differentiation and the Received February 14, 1979; accepted January 4, 1980 111 112 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER 17p-hydroxysteroid pathway of testosterone transformation, characteristic of androgen metabolfsm by normal prostatic epithelium. The dog was chosen as our experimental model because it shares with man the propensity to develop age-related benign prostatic hyperplasia (Berg, 1958) and prostatic carcinoma (Leav and Ling, 1968).To avoid the interactive effect of endogenous androgens and pituitary hormones with those of female sex steroids, we have recently utilized castrated or hypophysectomized dogs to characterize direct estrogen actions on the prostate (Leav et al., 1978). Our results show that the pharmacologic doses of estradiol-17P 17-cyclopentylpropionate(ECP) employed, not only cause basal-reserve cell proliferation and the attendant squamous metaplasia but also stimulate regressed glandular cells, with accompanying major alterations in prostatic androgen and estrogen metabolism. The regressed glandular cells, which had been responding as tall columnar epithelium to endogenous androgen prior to castration or hypophysectomy, undergo redirected differentiation in response to estrogen treatment. We have designated this cell type “estrogenmodified glandular” (EMG) cells (Leav et al., 1978). Published reports concerning prostatic ultrastructure have dealt primarily with hormone-induced intracellular events. In the present study emphasis is placed on sex hormone-mediated alterations in cytoplasmic and cell-surface components associated with intercellular junctions. These structures, which represent a significant cohesive element in the organization and development of normal tissue (Staehelin, 1974;McNutt, 19771, are frequently modified during pathologic processes including neoplastic transformation (McNutt and Weinstein, 1973; Weinstein et al., 1976). Our objective has been to determine the effects of hormonal manipulations on the ultrastructure of epithelial cell junctions in the canine prostate. Following the reorganization of prostatic acini that accompanies estrogeninduced metaplasia, we find that some desmosomes between EMG cells differ from those between glandular cells, but that tight junctions appear unchanged. MATERIALS AND METHODS Thirteen German Shepherd crossbreeds ranging from 3 to 5 years of age were utilized in this study. The experimental dogs were treated according to a protocol described in detail elsewhere (Leav et al., 1978). In brief, all experi- mental dogs were conditioned for 2&25 days after purchase from a licensed dealer. Each of the animals was given a complete physical examination including rectal palpation of the prostate. In no case was the prostate abnormally enlarged. Androgen depletion was accomplished either by castration (Lacroix, 1959) or by hypophysectomy (Markowitz et al., 1964).Levels of circulating 17p-hydroxy-C19-steroids (testosterone plus 5a-dihydrotestosterone, not separated) and circulating estrogens (estradiol-l7p and estrone, separately) were monitored before and at varying intervals after surgery. The plasma was stored a t -20°C until radioimmunoassay (RIA) could be performed (Collins et al., 1972). Estradiol-17p 17-cyclopentylpropionate (ECP) was administered to the dogs by intramuscular injection and testosterone cyclopentylpropionate (TCP) was given in the same manner. Both compoundswere purchased from the Upjohn Co., Kalamazoo, Michigan. We examined prostate glands taken from seven groups of dogs: 1)two intact untreated animals (dogs # 1and 2); 2) two animals killed four weeks after castration (dogs #3 and 4); 3) two animals, one killed two weeks, the other four weeks after hypophysectomy (dogs #5 and 6, respectively); 4) two castrated animals that each received a first injection of ECP (8 mg) four weeks after orchiectomy, the second 8 mg dose one week later and were killed one week after the second injection (dogs #7 and 8); 5) two castrated animals that each received one 8 mg injection of ECP two weeks after orchiectomy and were killed one week later (dogs #9 and 10); 6) two hypophysectomized animals that each received a first injection of ECP (8 mg) two weeks after the operation, a second 8 mg dose one week later and were killed one week after the second injection (dogs #11 and 12);7) one hypophysectomized animal that received a 600 mg dose of TCP two weeks after the operation and was killed 56 hours later (dog #13). Our results with the 13 animals are outlined in Table 1. The dogs were weighed and then killed by a n overdose of sodium pentobarbital. The prostate glands were quickly excised and weighed and the tissues were divided for various experimental procedures including thin-section and freeze-fracture electron microscopy. Specimens of sellae turcicae were taken from hypophysectomized dogs and decalcified. Sections were prepared for histologic examination to verify completeness of the hypophysectomies. Tissues for thin-section electron microscopy 113 EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS TABLE 1. Concentration ofplasma 17phydroxy-Cl9-steroid and weights ofprostates from intact and treated dogs Dog # Protocol’ 1 2 3 4 5 6 7 8 9 10 11 12 13 Intact Intact Castrate Castrate Hypox Hypox Castrate-ECP Castrate-ECP Castrate-ECP Castrate-ECP Hypox-ECP Hypox-ECP Hypox-TCP Pre-ablation Post-ablation 17p-hydroxy-C,,-steroidt(ng%*) - 110 200 70 150 135 75 70 400 130 165 135 - 20 < 10 < 10 12 12 12 < 10 < 10 < 10 15 14’ Prostate (g) 25.0 38.0 10.5 7.0 4.0 6.0 18.0 8.0 11.4 11.0 11.0 15.0 16.5 ‘Refer to Materials and Methods. 17P-hydroxy-C,,-steroidincludes testosterone and 5rr-dihydrotestosterone. The latter steroid cross-reads 56% with testosterone at 50% displacement. Cross-reactivity of other Cl.-steroidsis 3% or less. *ng/100 ml plasma. ‘RIA value three days before TCP treatment. t were fixed by immersion in 2.5% glutaraldehyde-Wo paraformaldehyde solution in 0.1 M phosphate buffer (pH 7.3) for three hours a t room temperature. They were then rinsed in the same buffer overnight, postfixed in 1% osmic acid in 0.1 M phosphate buffer for one hour at 4”C,dehydrated with a n ethanol series, and embedded in Epon 812. Silver sections were cut using a diamond knife and stained with uranyl acetate followed by lead citrate. In order to avoid biased sampling, the sections from each prostate were obtained from not fewer than four Epon blocks. The evaluation of desmosomes was carried out a t a final magnification of 200,000 x . Desmosomes were not included in the data unless both dense plaques appeared distinct. The widths of desmosomal filaments and plasma membranes were measured on the original plates and on prints enlarged exactly 2.5 X, using a Bausch & Lomb magnifying reticle. Illustrations of thinsectioned material were printed on highcontrast paper to enhance delineation of the filaments. Specimens for freeze-fracture electron microscopy were prepared according t o the method of Moor and Miihlethaler (1963).Tissue blocks 1 mm3 that had been fixed in 0.1M cacodylate-buffered Wo glutaraldehyde and stored in the same buffer were soaked overnight in 20%0 glycerol-0.1 M cacodylate. They were quenched in Freon 22 (cooled with liquid nitrogen) and fractured in a Balzers model BAF’301 freeze-etch device (Balzers High Vacuum Corp., Hudson, New Hampshire). Plati- num-carbon replicas were cast a t -115°C or -100°C and were subsequently cleaned in a sodium hypochloriteKOH mixture. All specimens were examined in a Philips 300 electron microscope which was calibrated with a carbon grating having a spacing of 0.463 pm per line. The terminology of Branton et al. (1 975) is used in descriptions of the freeze-fracture replicas. RESULTS Support o f experimental data No pituitary tissue was observed in sections of sellae turcicae from the hypophysectomized dogs. As a consequence of castration or hypophysectomy, levels of circulating 17phydroxy-C19-steroids declined sharply and in many instances approached the sensitivity limits (10 ng%) of our RIA (Table 1).No significant differences in the reduced levels of male sex steroids were apparent when the two ablative procedures were compared. The RIA values were obtained immediately prior to ablation, just before hormone treatment and a t sacrifice. Levels of plasma estradiol-17p ?nd estrone were also monitored. Following either ablative procedure, titers of these hormones decreased about 3&50%0.After ECP administration, they were increased 200-500% above preablation levels assuring us that our data represent a bona fide estrogen effect. Intercellular junctions in canznp prostate A terminal bar or “junctional complex” (Farquhar and Palade, 1963) is located between prostatic glandular cells a t the lirrninal poles 114 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER (Figs. 1-8,14,15, and 18).The elements of each complex generally appear in the following order: a distally located tight junction (zonula occludens), an intermediate junction (zonula adherens), and a proximally oriented desmosome (macula adherens). Two types of desmosomesjoin epithelial cells in the canine prostate. Both types are present either as a component of thejunctional complex or as independent structures along the plasma membrane. We classify them “100F and “ 7 0 F according to the criteria outlined by McNutt and Weinstein (1973).These classifications depend on the width of the filaments (100F or 70F) associated with the dense plaques (McNutt and Weinstein, 1973). We find that lOOF desmosomes, which frequently have a long profile (Fig. 131, are associated with distinct 9-10 nm wide tonofilaments (Fig. 16).The tonofilaments often extend deep into the cytoplasm, giving the desmosome a “shaggy” appearance (Fig. 16). However, 70F desmosomes have short, fine filaments that are 5-7 nm in width (Fig. 8) or are indistinct (Fig. 14). The 70F desmosomes are generally smaller than the lOOF variety and they usually have a “square” appearance because the total length of both sets of filaments, extending in opposite directions into adjacent cells, is about equal to the width of the junctional profile (Figs. 8 and 14). Although the type of desmosome cannot be positively identified at low magnification, the characteristic square or shaggy appearance is often apparent (Figs. 1-3). (RER) and lack secretory granules (Fig. 1).The other population consists of well-differentiated columnar (glandular) cells. They contain abundant RER and Golgi membranes, but the most prominent feature in the cytoplasm is many membrane-bound secretory granules (Fig. 1).The average diameter of the largest profiles (presumably granules that have been sectioned through the equator) is 1.05 i 0.17 Pm. Following either castration or hypophysectomy there is atrophy of glandular epithelium in the prostate (Fig. 2). Because morphological responses to both forms of ablation are essentially the same, they are described together. After ablation, no change is detected in the architecture of basal-reserve cells. However, glandular cells have increased nuclear to cytoplasmic ratios because of loss of cytoplasmic volume and the nuclei appear pleomorphic. There is a n increase in the number of autophagic vacuoles and lytic-dense bodies. Marked reductions in the amounts of RER, Golgi membranes, and secretory granules are observed. The few secretory granules that remain (vestigial granules) retain their characteristic appearance but some have undergone extraction artifact. The extracted material is presumably lipid, which is a major constituent of canine prostatic secretory granules (Gouvelis et al., 1971). Vestigial granules are found either sequestered in autophagic vacuoles or free in the cytoplasm (Fig. 2). Ultrastructural comparison of normal and Intercellular junctions in acini of intact regressed glandular cells and ablated dogs Two populations of epithelial cells are found Typical junctional complexes (Farquhar and in acini of intact-untreated dogs (Fig. 1).One of Palade, 1963) are found between glandular them is the undifferentiated basal-reserve cell cells of the intact dog prostate (Figs. 1and 14). (Timms et al., 1976) that is usually oriented Freeze-fracture replicas of normal tissue reveal with it long axis parallel to the basement that the tight junctions are complete fibrillar lamina. Basal-reserve cells have scant networks, consisting of 5-8 intramembranous amounts of rough endoplasmic reticulum strands. Type A-1 gap junctions (Staehelin, Fig. 1. Portion of a prostatic acinus from a normal intact dog (#l). Basal-reserve cells (B), which are located at the periphery of the acinus, appear inactive. Secretory (glandular)cells extend from the basal lamina (BL) to the lumen (L). They contain moderate amounts of rough endoplasmic reticulum (RER) and numerous secretory granules. A junctional camplex (JC) is present at the apical-lateral surfaces of two cells and its proximal component is a small “square”desmosome (arrow). x 13,700. Fig. 2. Acinus from a 4-week castrate(control) dog (#3). Involution ofthe glandular cellsis accompanied by an appearance of secondary lysosomes and by reduction in the number of cytoplasmic organelles including secretory granules. A small numberofvestigial granules remainand some of them are incorporatedinto developing-autophagicvacuoles (AV).Although a few vestigial granules have undergone extractionartifacts (longarrows), they retain the appearance of the secretory granules found in intact animals. Considerable cytoplasmic debris is present in the lumen. Junctional complexes, which include small “square” desmosomes, are similar to those of intact untreated dogs (short arrows). x 7,000. EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS 115 116 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS 1974) appear on the P-fracture face as relatively small aggregates of subunits (30-50 particles). Similar observations have been made in prostates of other species (Sinha and Bentley, 1975; Sinha et al., 1977). Both 70F and lOOF desmosomes are present in thin-sectioned prostatic epithelium obtained from intact dogs. From six blocks we examined thin sections that included profiles of more than 25 acini, and found that basal-reserve cells are joined to glandular cells almost exclusively by l O O F desmosomes (Fig. 13). In contrast, an examination of 105 desmosomes between glandular cells provided no clear indications that any are the 100F variety. The fine filaments, associated with desmosomes of the junctional complexes,usually appear indistinct and form a filamentous mat that appears similar to the mats associated with intermediate junctions (Fig. 14). Serial sections of the 70F desmosomes fail to reveal any contact with 10-nm filaments. Compact fascicles of tonofilaments, which are very electron-dense, occasionally appear in the cytoplasm of glandular cells but they are not found in contact with 70F desmosomes. We have found that the only desmosomes with distinct tonofilaments in glandular cell cytoplasm are those also joined to basal-reserve cells. Based on thin-section and freeze-fracture data, the ultrastructure of tight and gap junctions within regressed acini appears essentially the same as in intact untreated animals. Tight junctions with the full complement of junctional strands and small gap junctions persist between the regressed glandular cells. Similar observations have been made in other species (Tice et al., 1975; Sinha and Bently, 1975). Following ablation-induced regression, 100F desmosomes join basal-reserve cells and glandular cells. Although the 70F desmosome continues to be the predominating form between glandular cells (Fig. 151, a few lOOF desmosomes now appear (18out of 112 counted, or 16%). Response of regressed prostatic acini to estrogen stimulation No cytological differences have been observed between the ECP-stimulated epithelia of castrated or hypophysectomized dogs. Following estrogen administration, squamous metaplasia is evident in most of the acini. In addition, the regressed glandular cells are also stimulated. However, the extent of response in both types of cells is not uniform and varies from acinus to acinus. 117 Three types of epithelial cells are present in the acini (Leav et al., 1978).The basal-reserve cells are somewhat enlarged, but their cytoplasmic organelles do not appear modified by the female hormone. Squamous cells, which appear elongated, are fairly uniform in appearance (Fig. 3). They are generally located in the periphery of the acini and, as squamous metaplasia becomes advanced, these cells become stratified. Squamous cells are characterized by large perinuclear bundles of tonofilaments but secretory-, keratohyalin-, and membranecoating granules are absent. The basal-reserve cells and their squamous-cell progeny are joined to one another almost exclusively by l O O F desmosomes. The third type is the pleomorphic estrogen-modified glandular (EMG) cells (Fig. 31, which form a single layer. They combine features of secretory and squamous cells. The EMG-cell contains numerous tonofilaments that are scattered or arranged in bundles. Coexisting with the tonofilaments are abundant RER, Golgi membranes, and large numbers of secretory granules (Fig. 4). Two distinct populations of membrane-bound granules are frequently found in the same EMG cell (Figs. 5 and 6). One type is easily recognized as the vestigial granules. They are relatively sparse and are encountered in various stages of lysosome-induced degradation (Fig. 5). These granules, which have the same texture and dimensions as those found in glandular cells prior to ablation, can be distinguished from lipid droplets by virtue of their limiting membrane (Figs. 5 and 6). They are consequently considered to be structures that were largely formed under the influence of androgen. The other type is very numerous. They are usually about % the size, less “electrondense,” and more peripherally located than the vestigial granules (Figs. 5 and 6). Since the small granules were never found in regressed acini, their development must be considered to be an estrogen-mediated phenomenon. Intercellular junctions between the modified glandular cells of estrogen-treated dogs The EMG cells, laden with secretory granules, a r e easily recognized in freezefracture replicas (Fig. 91, and the tight junctions associated with these cells (Fig. 10) closely resemble those in intact and ablated animals. We never observe focal attenuations or discontinuities in the network of intramembranous strands. Gap junctions are not readily apparent in EMG cell membranes. However, unusual junctional combinations (Orwin et al. 118 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS 1973; Alroy and Weinstein 1976; Elias and Friend 1976) consisting of granulofibrillar areas encircled by tight-junctional strands are occasionally seen (Fig. 11). Following estrogen treatment, l O O F desmosomes are the only type observed between squamous and EMG cells. They are also found between EMG cells (54 out of 127 counted, or 43%).The 100F desmosome is present on EMG cell surfaces at many locations, including the proximal position of junctional complexes (Figs. 4 and 7). The tonofilaments of these junctions tend to be long and straight (Figs. 7 and 16) and they often intermingle with cytoplasmic bundles of tonofilaments (Fig. 16). However, 70F desmosomes are also found between EMG cells. Many of them appear unaffected by the hormone treatment (compare Figs. 7 and 8). They are associated with short filaments that are thready (Fig. 8) or indistinct (Fig. 5).Serial sections indicate that the dense plaques are usually not in contact with 10-nm filaments (Figs. 5a and b). However, some small desmosomes are associated with a few tonofilaments (lower part of Fig. 7). These junctions probably represent a spectrum that spans the two prototypes described above. Response of regressed acini to androgen stimulation In this preliminary experiment we found that, during the brief period (56 hours) following androgen administration, full restoration of the secretory epithelium has occurred. The basal-reserve cells appear unchanged, but the glandular cell cytoplasm is packed with secretory granules (Fig. 18) that have the same “electron density” as their counterparts in intact dogs. Freeze-fractured tight junctions (Fig. 12) are indistinguishable from those in intact or ablated dogs, and large gap junctions (Fig. 17) are present. Desmosomes of the junctional complexes and elsewhere between glandular cells are predominantly (89 of 100 counted) the 70F variety (Fig. 18). Compact fascicles of tonofilaments, similar to those described in intact 119 dogs, are sometimes present in the cytoplasm but they are usually not in contact with the desmosomes. DISCUSSION When the prostate of the androgen-depleted dog is exposed to pharmacological levels of circulating estrogen, two populations of epithelial cells are affected simultaneously. One population is the basal-reserve cells, which proliferate (Helpap and Steins, 1975;Hohbach, 1977;Leav et al., 1978), and whose progeny differentiate into squamous cells, giving rise to squamous metaplasia. The other population is the regressed glandular cells. When stimulated with estrogen, they undergo cytoplasmic transformation from one adult cell type into another and therefore are also participants in the metaplastic process. However, unlike the basal-reserve cells, they probably do not divide. Autoradiographic analysis of 3H-thymidine incorporation in prostatic explants, exposed to estrogen in vivo, indicates a high labeling index over the basal(squamous cell) regions of the acini and a low index over the superficial (EMG cell) regions (Helpap and Steins, 1975; Leav et al., 1978). The regressed prostatic glandular cell in our canine model is a unique tool for monitoring ultrastructural responses to individual sex steroid hormones in vivo. Following estrogen stimulation, these cells undergo pronounced changes in appearance. However, their original identity is not lost because the vestigial granules, a distinctive cytoplasmicmarker, are retained. Their presence indicates that these cells were responding to androgen prior to ablation. The small secretory granules that appear followingECP treatment were never present in the regressed epithelium. Consequently we consider them to be an indicator of estrogen action. Coexistence of the small estrogeninduced secretory granules andlor tonofilament bundles with vestigial granules is a distinguishing feature of EMG cells and is indicative of bipotential hormone responsiveness Fig. 4. Detailed view of EMG cells from a castrated ECP-treated dog (#7). Small membrane-boundsecretory granules are situated near the plasma membrane and they coexist in the cytoplasm with wide bundles of tonofilaments (TF),Golgi apparatuses(G),and rough endoplasmic reticulum (ER),and lytic-densebodies (DB)are common.The EMGcells arejoinedby tightjunctions (TJ) and lOOF desmosomes (D). A junctional complex from the same specimen is illustrated in the inset. The desmosomal(1OOF)component is tangentially sectioned and distinct tonofilaments are seen converging on the dense plaques. x 41,000; inset x 45,000. 120 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS 121 Fig. 5. Estrogen-modifiedglandular cells from a hypophysectomized, ECP-treated dog (#12). Two populations of secretory granules coexist in the same cell. Small granules, which appear in large numbers, are arranged subjacent to the luminal membrane. A second population of granules, which is larger and more electron-dense than the first, is illustrated in various stages of lysosome-induced degeneration (LYS). These are vestigial granules. At the far left is a junctional complex that includes a “square” 70F desmosome (arrow).A high magnificationview (inset a) and a non-consecutive serial section (inset b) show that this tangentially sectioned desmosome is associated with fine as well as indistinct filaments and that it has no apparent contact with tonofilaments. Inset c is a high magnification view of the area within the rectangle (from a serial section). The outer membrane is that of the lysosome, whereas the inner membrane is part of the vestigial granule. Arrows indicate regions where the “unit”character of the membranes is evident. x 14,500;insets a and b x 63,000;inset c x 95,000. Fig. 6. Estrogen-modifiedglandular cells from the same specimen as the preceding figure (dog # 12).Ajunctional complex that joins two EMG cells includes a lOOF desmosome (D). Tonofilaments of the lOOF desmosome (see Fig. 7) and small luminally oriented secretory granules coexist in the same cell with a vestigial granule (V).Vestigial granules, including the one in the inset (alsofrom dog #12), have amorphous centers, whereas lytic-densebodies (DB)have heterogeneous interiors. A rim of unextracted material, which includes the limiting membrane of the vestigial granule, is seen in the inset (arrow).The outer membrane is probably that of a lysosome whose body is entirely out of the plane of section (compare with Fig. 5). x 14,500 inset x 44,000. 122 F.B. MERK, I. LEAV, P.W.L. KWAN,AND P. OFNER EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS Fig. 7. High magnification of the EMG-EMG cell desmosomes illustrated in Figure 6. The coarse filaments (long arrows) associated with the lOOF desmosome near the lumen are long and they have the rigid appearance characteristic of tonofilaments. The dense plaques of the other desmosome are associated with a fine network of cytoplasmic filaments (short arrows)that stain less intensely than tonofilaments. However, a few coarse (%10nm) filaments (long arrows) also converge onto the dense plaque. This junction may represent a transition stage between 70F and lOOF desmosomes. x 65,000. Fig. 8. Detailed view of a n EMG-cell junctional complex that includes a “square” 70F desmosome (dog #12). Structural differences between the filaments associated with 70F and IOOF desmosomes will be appreciated by comparing Figures 7 and 8, both printed a t the same magnification. The short rippled microfilaments that converge on the dense plaque of this 70F desmosome are 5-7 nm wide and do not exceed the width of transversely sectioned plasma membrane (at asterisk). x 65,000. 123 124 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER Fig. 9. Freeze-fracture replica from the prostate of a hypophysectomized ECP-treated dog (#ll).The EMG cells are characterized in replicas by the presence of secretory granules near the luminal membranes (compare with Figs. 4-61, Except for a small region of PF face luminal membrane (P) and nuclear envelope tN), the cells have been cross-fractured, clearly revealing the proximity of the secretory granules to the lumen (L).The granulemembrane relationship facilitates identification of EMG cell membranes for junctional analysis. X 43,000. EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS 125 Fig. 10. Freeze-fractured tight junction (zonula occludens) from a hypophysectomized ECP-treated dog (#ll).This EMGcell junction consists of an uninterrupted fibrillar network that is 5-8 strands in depth. The luminal membrane with cross-fractured microvilli (MV) is oriented above the junction and the lateral membrane is below. x 46,500. Fig. 11. Unusual combinationof a tight junction and desmosomeswithin an EMG-cell plasma membrane (dog # 11). Some PF-face strands of the tight junctions encircle granulofibrillar areas of the membrane that are freeze-fractured desmosomes (D). It is not possible to determine whether these desmosomes are 70F or 100F. x 54,000. Fig. 12. Freeze-fracture replica of a tightjunction from a hypophysectomized TCP-treated dog (#13). The tight junctional network, ranging from 6 to 9 strands in depth, forms a continuous belt within the membrane of an androgen-stimulated glandular cell. Thisjunction closely resemblesthe zonulae occludentes found in prostatesof intact-untreated dogs. x 40,000. 126 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS 127 Fig. 13. lOOF desmosome from a prostatic acinus of an intact normal dog (#l).This junction, which has a long profile, is located between a basal-reserve cell and a glandular cell. x 82,000. Fig. 14. Junctional complex between prostatic glandular cells of a n intact normal dog (#U. Two epithelial cells arejoinednear the luminal surface (L)by ajunctional complex which includes a tightjunction (TJ),intermediatejunction (IJ), and “square” 70F desmosome (DES). Although unit membranes and dense plaques of the desmosome are distinct, the associated filaments are not individually resolved, They form a filamentous mat that resembles the one adjacent to the intermediate junction. X 82,000. Fig. 15. Junctional complex between regressed epithelial cells of acastrated dog (#3). A large 70F desmosome is the proximal component ofthisjunctional complex. A smaller 70F desmosome is also present. Adjacent to the dense plaques, fine filaments are seen in longitudinal section (long arrows) and in cross section (short arrows). The width of the 5-6 nm filaments is less than transversely sectioned plasmalemma (asterisk) and unit membrane within the desmosome itself. X 114,000. Fig. 16. Two “shaggy” lOOF desmosomes in prostatic epitheliumofa castrated ECP-treateddog (#7). The desmosomesjoin a n EMG cell (below)with a squamous cell (above).Tonofilaments associated with the desmosomesextend deeply into the cytoplasm and mingle with bundles of tonofilaments. The tonofilaments (between arrows) have a width of 9-10 nm. Most of the plasma membranes between the cells appear thickened because of tangential plane of section. A short portion of transversely sectioned plasmalemma (asterisk) is narrower than the tonofilaments. x 82,000. Fig. 17. Freeze-fractured gap junction from the glandular epithelium of a hypophysectomized TCP-treated dog (#13). A large aggregation of 7-9 nm particles is present on the P fracture face. x 90,000. Fig. 18. Detailed view of a junctional complex in the glandular epithelium of a hypophysectomized TCP-treated dog (#13). Fifty-six hours after androgen administration abundant secretory granules are present in the cytoplasm. Large and small “square” 70F desmosomes are similar to those that join glandular cells in intact-untreated animals. x 65,000. 128 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER (Fig. 19). In addition, the EMG cells also exhibit bidirectional (divergent) differentiation. This phenomenon is expressed in the cytoplasm as development of secretory (small granules) and squamous (tonofilaments) components within the same cell (Fig. 19). Thus estrogen appears to be capable of imposing both secretory and squamous cell features onto a framework that previously exhibited only secretory cell characteristics. In the normal prostatic acinus, a continuous layer of glandular epithelium forms a lining between the lumen and the basement lamina. Tight junctions (zonulae occludentes) between the glandular cells are the paracellular permeability barrier to diffusion (Claude and Goodenough, 1973; Staehelin, 1974;McNutt, 1977). When the glandular cells acquire squamous characteristics, as the result of estrogen treatment, no modifications of the tight-junctional networks are observed. This observation may be significant in view of the nature of tightjunctional development in other epithelia. squamous cells in mammalian tissues such as epidermis and vagina are joined by focal (discontinuous) tight junctions (Elias et al., 1978). However, when a squamous epithelium is exposed to retinoic acid (vitamin A) in organ culture, tonofilaments and other squamous features are progressively replaced by glycogen and mucous granules (Elias and Friend, 1976). Concurrently, complete (continuous) tight junctions are assembled at the surfaces of cells undergoing the vitamin A-induced mucous metaplasia (Elias and Friend, 1976). Our experimental protocol produces an opposite effect on epithelia. The glandular cells acquire tonofilaments and other squamous features, yet the completeness of the tight junctions is not affected. Attenuation and breakdown of tightjunctional elements can take place during rapid cell proliferation (Staehelin, 19741, amphibian neurulation (Decker and Friend, 19741, and neoplastic transformation (Merk et al., 1977; Sinha et al., 1977). Our failure t o observe junctional breakdown may be due to insufficient time for disassembly to occur. A more likely explanation is that although EMG cells have acquired some squamous cell features they continue to perform glandular cell functions, including provision of a permeability barrier for the underlying cells. Desmosomes serve as anchorage sites for intercellular adhesion. These cell-surface specializations are associated with cytoplasmic filaments that protect the plasma membrane by distributing mechanical stress over a large volume (McNutt and Weinstein, 1973; Staehelin, 1974). The present study demonstrates that two morphologically distinct forms of desmosomes are present in the canine prostate. The 70F and lOOF desmosomes have been described previously (McNutt and Weinstein, 19731,but this is the first report that the phenotypic expression of desmosomes is under hormonal control in a target tissue. Our finding is Fig. 19. Schematic representation of responses that regressed prostatic cells have to male and female sex steroids. Atrophic glandular cells (GI a t the bottom of the figure are joined to one another by 70F desrnosomes (1). Basal-reserve cells (B) form lOOF desmosornes (2) with their glandular cell neighbors. The cytoplasm of the regressed glandular cells contains numerous lysosornes including lytic-dense bodies (3).Also present are vestigial secretory granules (41,which frequently appear partially extracted. When regressed glandular cells are exposed to TCP, they rapidly acquire a n appearance similar to that of the glandular cells found in intact animals (upper left). Desrnosomes joining glandular cells, either as components of junctional complexes or individually, are of the 70F variety (1).The lOOF desmosomes are present only between glandular and basal cells (2). Under the influence of androgen, the glandular cells have the characteristics of a secretory epithelium. They are laden with abundant, densely stained secretory granules. Broad bundles of tonofilaments are lacking. When regressed glandular cells are exposed to pharmacological levels of ECP (illustrated a t center right) they acquire a phenotype that is not observed in normal canine prostate. Although estrogen-modified glandular cells (EMG) continue to exhibit 70F desmosomes (1) and vestigial granules (41,new structures have appeared in response to the female hormone. They are a population of small secretory granules, broad bundles of tonofilaments, and lOOF desmosornes (2) that join the EMG cells. Basal-reserve cells (B) also respond to estrogen stimulation. They proliferate and their squamous cell progeny (S), characterized by lOOF desmosomes and tonofilament bundles, are located between some EMG cells and the basement lamina. The coexistence of vestigial granules (a marker of previous androgen responsiveness) and the small secretory granules (a marker of estrogen responsiveness) in EMG cells indicates that a potential to respond to either sex hormone exists in the regressed glandular cells, i.e., a bipotentiality of response. A collateral phenomenon is observed in the EMG cells. Tonofilament bundles and lOOF desmosomes (characteristics of squamous-cell differentiation) are formed concurrently with secretory granules (characteristic of secretory-cell differentiation) under the influences of estrogen. Coexistence of these structures in the EMG cells is an indication of bidirectional differentiation. 129 EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS - Androgen Induced Secretory Granules INTACTor TCP-TREATED Estrogen-Induced Secretory Granules Bidirect ionaI Differentiation c3/ HYPOX.or CASTRATE 130 F.B. MERK, I. LEAV, P.W.L. KWAN. AND P. OFNER not unexpected, because previous studies have shown that another cell-surface structure, the gap junction, also responds to hormones (Merk et al., 1972; Bjersing and Cajander, 1974; Decker, 1976; Dahl and Berger, 1978). The predominance of 70F desmosomes in glandular epithelium of the canine prostate appears to be a species-specific phenomenon. We have frequently seen l O O F desmosomes between glandular cells of normal prostate in other species including man (Merk, Leav, Kwan, and Ofner, unpublished observations). The 70F desmosomes observed between glandular cells of normal canine prostate resemble the desmosomes found during embryonic development by Lentz and Trinkaus (1971). The W3 nm filaments of these junctions frequently mingle with morphologically similar filaments in the ectoplasm. Ishikawa et al. (1969) have shown that heavy meromyosin (HMM)is bound to the thin filaments of epithelial cells in the same “arrowhead’ manner that it is bound to known samples of fibrous actin. However, thicker 10 nm filaments, including those associated with squamous-cell desmosomes (i.e., lOOF), are not decorated with HMM. The data of Ishikawa et al. (1969) indicate that fine filaments in the ectoplasm of epithelial cells are actin or actin-like protein and presumably have contractile properties, in contrast to the thicker filaments that are probably non-motile and subserve a cytoskeletal function. McNutt and Weinstein (1973) have suggested that the thin filaments of intermediate junctions (zonulae adherentes) and 70F desmosomes are also F-actin because they have a close association and morphological similarity with HMMdecorated filaments. Recent evidence supports the concept that F-actin is associated with components of the junctional complex. The protein a-actinin, which is considered to be a membrane anchor for actin, has been localized by indirect immunofluorescent microscopy in the junctional complexes of intestinal epithelia (Craig and Pardo, 1979). Our inability to resolve consistently the filaments associated with 70F desmosomes and intermediate junctions in our osmium-fixed tissues also supports the F-actin concept. Purified actin filaments, prefixed with glutaraldehyde, are readily fragmented by the action of osmic acid solutions (Maupin-Szamier and Pollard, 19781, whereas the 10 nm tonofilaments appear stable. If the proposal of McNutt and Weinstein (1973) is correct, the different properties that motile actin filaments and non-motile tonofilaments confer on 70F and lOOF desmosomes, respectively, may have functional significance. During the early stages of desmosomal morphogenesis in regenerating epidermis (Krawczyk and Wilgram, 19731, tonofilaments are attached shortly after the dense plaques are formed. This finding indicates that the lOOF desmosome is formed de novo and consequently the 70F variety, under ordinary circumstances, is not an immature form of the l O O F but is in itself a fully differentiated structure. Previous reports about morphological changes in mature desmosomes of epithelial cells have been limited to multilayered tissues where the alteration has been shown to be dependent on the location of the cells as they progress from one position to another (Petry et al., 1961; Farquhar and Palade, 1965). Our observations of 70F and lOOF desmosomes between EMG cells have been made on a single-layered epithelium. The presence of transition-stage desmosomes,which combine 70F and lOOF features (lower part, Fig. 71, suggests that some 70F desmosomes may be transformed into lOOF when the phenotype of glandular cells is modified by androgen depletion and estrogen administration. 100F desmosomes are part of the squamouscell phenotype that is expressed during squamous metaplasia irrespective of the causative agent. Consequently we do not interpret their appearance as a hormone-specific response. However 100F desmosomes are not normally part of the glandular-cell phenotype of the intact canine prostate. Therefore the appearance of these junctions between EMG cells must either by directly attributable to androgen depletion and estrogen administration or be a nonspecific phenomenon resulting from a rapid cellular response t o hormone stimulation. To test the latter possibility, we examined the glandular epithelium of a hypophysectomized dog killed just 56 hours after androgen treatment, and we infrequently observed lOOF desmosomes between glandular cells. Our observation that lOOF desmosomes are more commonly found between EMG cells than between fully restored glandular cells suggests that their formation is not the result of a generalized rapid response to sex hormone stimulation but represents a specific effect due to androgen depletion and estrogen administration. It is likely that androgen also actively regulates the phenotypic expression of desmosomes in the canine prostate. We have noticed that, when the ratio of circulating androgen to estrogen is high, 70F desmosomes predominate be- EFFECTS OF SEX STEROIDS ON CELL JUNCTIONS tween glandular cells. However, following ablation, the decline in levels of circulating endogenous androgens is much greater than that of estrogen (Leav et al, 1978). This decrease in the androgedestrogen ratio is accompanied by a n increased incidence of lOOF desmosomes between glandular cells. Similarly, lOOF desmosomes also predominate in the undifferentiated prostatic epithelium of sexually immature 6-week-old dogs (Merk, Leav, and Cavazos, unpublished results). Our observations suggest that the phenotypic expression of the two types of desmosome is regulated by the androgen-estrogen ratio. Whether androgen exerts its influence by promoting formation of 70F desmosomes or by inhibiting the lOOF variety cannot be determined a t present. Prostatic glandular cells, which have developed in an environment favoring androgen, express an aberrant phenotype when exposed to pharmacological levels of estrogen. Under these conditions the dimorphic nature of EMG-cell desmosomes is evidence of bidirectional differentiation a t the cell surface. Coexistence of 70F and l O O F desmosomes on the surfaces of the metaplastic EMG cells can be compared with a similar phenomenon found on unusual cells present in canine mammary adenoacanthomas. Glandular and squamous cells in these tumors are joined by 70F and lOOF desmosomes, respectively. However, a third population, the “intermediate” cell, includes both the glandular and squamous cell types of desmosomes and therefore exhibits bidirectional differentiation of the plasma membrane (Alroy and Weinstein, 1976). Freeze-fracture replicas of epithelial cells, which combine squamous and glandular cell characteristics, sometimes display unusual combinations of tight junctions and granulofibrillar areas (desmosomes).The combinations are present on “intermediate” cell surfaces (Alroyand Weinstein, 1976)and are also found between cultured peridermal cells of the chick embryo (Elias and Friend, 1976) and between keratinizing Henle’s cells of the wool follicle (Orwin et al., 1973).These combinations, which are not ordinarily encountered between normal adult glandular cells, are found in the glandular epithelium of canine prostate following estrogen administration (Fig. 11). We conclude that glandular cells of the canine prostate possess cellular mechanisms that render them responsive t o either androgen or estrogen. During reorganization of the acini, which follows endocrine manipulations, cyto- 131 plasmic and cell-surface modifications reflect the changes occurring in the hormonal environment. When glandular cells are exposed to the abnormal stimulus of pharmacological doses of estrogen, the hormone regulates the expression of many types of structures that had previously responded to androgen, including RER, secretory granules, and desmosomes. Under physiological conditions, estrogeninduced morphological responses may be modified or masked by androgen. The differentiation of glandular cells in the canine prostate may be brought about by androgens and estrogens working in concert to establish the characteristic morphology found in intact animals. ACKNOWLEDGMENTS The authors are grateful to Bennett M. Stein, M.D., who performed the hypophysectomies, and to Fred W. Quimby, D.V.M., for assistance and advice in animal care. We also wish to thank Mr. Gary Goodrich for the steroid RIAs, Mr. Steven Halpern for photographic assistance, Mr. Dan Casper for the schematic illustration, and Ms. Lela Silverstein and Ms. Arlene Kronman for help in preparing the manuscript. This research was supported by grants CA 15776 and CA 16377 from the National Cancer Institute, U.S. Department of Health, Education, and Welfare. LITERATURE CITED Alroy, J., and R.S. Weinstein (1976) Unusual junctional complexes in canine mammary adenoacanthomas. J. Natl. Can. Inst., 561667-670. Bashirelahi, N., and J.D. Young (1976) Specific binding protein for 17P-estradiol i n prostate with adenocarcinoma. Urology, 8:55%558. Berg, O.A. (1958) Parenchymatous hypertrophy of the canine prostate gland. Acta Endocrinol., 27: 14&154. Bjersing, L., and S. Cajander (1974) Ovulation and the mechanism of follicle rupture. IV. Ultrastructure of membrana granulosa of rabbit graafian follicles prior to induced ovulation. Cell Tissue Res., 153:l-14. Branton, D., S. Bullivant, N.B. Gilula, M.J. Karnovsky, H. Moor, K. Muhlethaler, D.H. Northcote, L. Packer, B. Satir, P. Satir, V. Speth, L.A. Staehelin, R.I. Steere, and R.S. Weinstein (1975) Freeze-etching nomenclature. Science, 19054-56. Claude, P., and D.P. Goodenough (1973) Fracture faces of zonulae occludentes from “tight” and “leaky” epithelia. J. Cell Biol., 58:39@400. Collins, W.P., M.D. Mansfield, N.S. Alladina, and I.F. Sommerville (1972) Radioimmunoassay of plasma testosterone. J. Steroid Biochem., 3:33%348. Craig, S.W., and J.V. Pardo ( 1979) Alpha-actinin localization in the junctional complex of intestinal epithelial cells. J. Cell Biol., 80:203-210. Dahl, G., and W. Berger (1978)Nexus formation i n the myometrium during parturition and induced by estrogen. Cell Biol. Intl. Rep., 2t381-387. 132 F.B. MERK, I. LEAV, P.W.L. KWAN, AND P. OFNER Decker, R.S. (1976) Hormonal regulation of gap junction differentiation. J. Cell Biol., 69:66%685. Decker, R.S., and D.S. Friend (1974) Assembly of gap junctions during amphibian neurulation. J . Cell Biol., 62t32-47. Elias, P.M., and D.S. Friend (1976) Vitamin-A-induced mucous metaplasia. An in vitro system for modulating tight and gap junction differentiation. J. Cell Biol., 68: 17%188. Elias, P.M., N.S. McNutt, and D.S. Friend (1978) Membrane alterations during cornification of mammalian squamous epithelia: A freeze-fracture, tracer and thin-section study. Anat. Rec., 189t577-594. Farquhar, M.G., and G.E. Palade (1963) Junctional complexes in various epithelia. J. Cell Biol., 17:375-412. Farquhar, M.G., and G.E. Palade (1965) Cell junctions in amphibian skin. J. Cell Biol., 26:26%291. Gouvelis, A,, J. Baker, and H. Rosenkrantz (1971) Histochemistry of the canine prostate. Invest. Urol., 8:42& 441. Hawkins, E.F., M. Nijs, and C. Brassinne (1976) Steroid receptors in the human prostate. 2. Some properties of the estrophilic molecule of benign prostatic hypertrophy. Biochem. Biophys. Res. Commun., 70t856861. Helpap, B., and R. Steins (1975) The cell proliferation of epithelial metaplasia in the prostate gland. An autoradiographic in vitro study. Virchows Arch. B. Cell Pathol., 19:6%76. Hohbach, C. (1977) Ul t rast ruct ural a n d enzymehistochemical studies of the prostate of the dog under the effect of estradiol. Beitr. Pathol., 160:26&273. Ishikawa, H., R. Bischoff, and H. Holtzer (1969) Formationof arrowhead complexes with heavy meromyosin in a variety of cell types. J. Cell Biol., 43:31%328. Kjaerheim, A,, E. Dahl, and K.J. Tveter (1974) The ultrastructure of the accessory sex organs of the male rat. 8. Effects of estrogen on the prostate. Lab. Invest.,31:391-397. Krawczyk, W.S., and G.F. Wilgram (1973)Hemidesmosomes and desmosome morphogenesis during epidermal wound healing. J. Ultrastr. Res., 45:9%101. Kroes, R., and J.S. Teppema (1972) Development and restitution of squamous metaplasia in the calf prostate after a single estrogen treatment. Exp. Mol. Pathol., 16:28&301. Lacroix, J.V. (1959) Orchiectomy. Canine Surgery, 4th edition, M. Mayer, J.V. LaCroix and P.H. Hoskins eds., Am. Vet. Pub. Inc., Santa Barbara, California, pp. 54&548. Leav, I., and G.V. Ling (1968) Adenocarcinoma of the canine prostate. Cancer, 22: 132%1345. Leav, I., F.B. Merk, P. Ofner, G. Goodrich, P.W.L. Kwan, B.M. Stein, M. Sar, and W.E. Stumpf (1978) Bipotentiality of response to sex hormones by the prostate of castrated and hypophysectomized dogs. Direct effects of estrogen. Am. J. Pathol., 93:6%92. Lentz, T.L., and J.P. Trinkaus (1971) Differentiation of the junctional complex of surface cells i n the developing Fundulus blastoderm. J. Cell Biol., 48:455472. Markowitz, J., J. Archibald, and H.G. Downie (1964) Hypophysectomy in dogs. In: Experimental Surgery, Williams and Wilkins Co., Baltimore, pp. 63&643. Maupin-Szamier, P., and T.D. Pollard (1978) Actin filament destruction by osmium tetroxide. J. Cell Biol., 77:837-852. McNutt, N.S. (1977) Freeze-fracture techniques and applications to the structural analysis of the mammalian plasma membrane. In: Dynamic Aspects of Cell Surface Organization. G. Poste and G.L. Nicolson eds., North-Holland Publishing Co., New York, pp. 75126. McNutt, N.S., and R.S. Weinstein (1973) Membrane ultrastructure at mammalian intercellular junctions. In: Progress in Biophysics and Molecular Biology, Vol. 26, J.A.V. Butler and D. Noble, eds., Pergamon Press, New York, pp. 45-101. Merk, F.B., C.R. Botticelli, and J.T. Albright (1972) An intercellul a r response to estrogen by granulosa cells in the rat ovary; a n electron microscope study. Endocrinology, 90:992-1107. Merk, F.B., B.U. Pauli, J.B. Jacobs, J. Alroy, G.H. Friedell, and R.S. Weinstein (1977) Malignant transformation of urinary bladder in humans and in N-[4(5-Nitro-2-furyl)2-thiazolyl] formamide-exposed Fischer rats: Ultrastructure of the major components of the permeability barrier. Cancer Res., 37:284%2853. Moor, H., and K. Miihlethaler (1963) Fine structure in frozen-etched yeast cells. J. Cell Biol., 17:60%623. Ofner P., I. Leav, and L.F. Cavazos (1974) C,,-steroid metabolism in male accessory sex glands. Correlation of changes in fine structure and radiometabolite patterns in the prostate of the androgen-deprived dog. In: Male Accessory Sex Organs, Structure and Function in Mammals. D. Brandes, ed., Academic Press, New York, pp. 267-305. Onvin, D.F.G., R.W. Thomson, and N.E. Flower (1973) Plasma membrane differentiation of keratinizing cells a t the wool follicle. 111. Tight junctions. J. Ultrastr. Res., 45:3&40. Petry, G., L. Overbeck, and W. Vogell (1961) Sind desmosomen statische oder temporare zellverbindungen? Natunvissenschaften, 48: 166- 167. Sinha, A.A., and M.D. Bentley (1975) Morphological changes in the ventral lobes of prostate in normal, castrated and testosterone-treated mice studied by the techniques of freeze fracture and electron microscopy. J. Cell Biol., 67t404A. Sinha,A.A., M.D. Bentley, and C.E. Blackard (1977) Freezefracture observations on the membranes and junctions in human prostatic carcinoma and benign prostatic hypertrophy. Cancer, 40: 118%1188. Staehelin, L.A. (1974) Structure and function ofintercellular junctions. Intl. Rev. Cytol., 39: 191-283. Tice, L.W., S.H. Wollman, and R.C. Carter (1975) Changes in tight junctions of thyroid epithelium with changes in thyroid activity. J. Cell Biol., 66t657-663. Timms, B.G., J.A. Chandler, and F. Sinowatz (1976) The ultrastructure of basal cells of rat and dog prostate. Cell Tissue Res., 2731545-554. Triche,T.J., and J.C. Harkin (1971)Anultrastructural study of hormonally induced squamous metaplasia in the coagulating gland of the mouse prostate. Lab. Invest., 25:596-606. Weinstein, R.S., F.B. Merk, and J. Ahoy (1976) The structure and function of intercellular junctions in cancer, Adv. Cancer Res., 23:23-89.