Testin Secreted by Sertoli Cells Is Associated with the Cell Surface, and Its Expression Correlates with the Disruption of Sertoli-Germ Cell Junctions but Not the Inter-Sertoli Tight Junction*

Testin is a testosterone-responsive Sertoli cell secretory product. In the present study, we demonstrated that the amount of testin secreted by Sertoli cells in vitro was comparable with several other Sertoli cell secretory products. However, virtually no testin was found in the luminal fluid and cytosols of the testis and epididymis when the intercellular junctions were not previously disrupted, suggesting that secreted testin may be reabsorbed by testicular cells in vivo. Studies using Sertoli cells with and without a cell surface cross-linker and radioiodination in conjunction with immunoprecipitation illustrated the presence of two polypeptides of 28 and 45 kDa, which constitute a binding protein complex that anchors testin onto the cell surface. The 28- and 45-kDa peptide appear to be residing on and inside the cell surface, respectively. Immunogold EM studies illustrated testin was abundantly localized on the Sertoli cell side of the ectoplasmic specialization (a modified adherens junction) surrounding developing spermatids. In contrast, very few testin gold particles were found at the site of inter-Sertoli tight junctions. When the inter-Sertoli tight junctions were formed or disrupted, no significant change in testin expression was noted. This is in sharp contrast to the disruption of Sertoli-germ cell junctions, which is accompanied by a surge in testin expression. These results demonstrate the usefulness of testin in examining Sertoli-germ cell interactions.

Testin is a testosterone-responsive Sertoli cell secretory product. In the present study, we demonstrated that the amount of testin secreted by Sertoli cells in vitro was comparable with several other Sertoli cell secretory products. However, virtually no testin was found in the luminal fluid and cytosols of the testis and epididymis when the intercellular junctions were not previously disrupted, suggesting that secreted testin may be reabsorbed by testicular cells in vivo. Studies using Sertoli cells with and without a cell surface crosslinker and radioiodination in conjunction with immunoprecipitation illustrated the presence of two polypeptides of 28 and 45 kDa, which constitute a binding protein complex that anchors testin onto the cell surface. The 28-and 45-kDa peptide appear to be residing on and inside the cell surface, respectively. Immunogold EM studies illustrated testin was abundantly localized on the Sertoli cell side of the ectoplasmic specialization (a modified adherens junction) surrounding developing spermatids. In contrast, very few testin gold particles were found at the site of inter-Sertoli tight junctions. When the inter-Sertoli tight junctions were formed or disrupted, no significant change in testin expression was noted. This is in sharp contrast to the disruption of Sertoli-germ cell junctions, which is accompanied by a surge in testin expression. These results demonstrate the usefulness of testin in examining Sertoli-germ cell interactions.
Testin, a testosterone-responsive glycoprotein secreted by rat Sertoli cells in vitro, consists of two highly homologous variants with an apparent M r of 35,000 and 37,000 (1)(2)(3)(4). Immunofluorescent microscopy and immunohistochemistry reveal that testin resides near the basal lamina of the seminiferous epithelium in most stages of the cycle, consistent with its localization at the Sertoli-germ cell junction (5)(6). However, a transient but drastic increase in testin accumulation was noted between Sertoli cells and the head of elongated spermatids at early stage VIII preceding spermiation (7), consistent with its localization at the ectoplasmic specialization, which is a modified adherens junction (see Table I). These observations sug-gest that testin may be a sensitive marker in examining the cellular events of Sertoli-germ cell interactions.
Once the mRNA sequence of testin was known, Northern blots and reverse transcription-polymerase chain reaction were used to survey the testin mRNA distribution in multiple organs from both adult and immature male and female rats. It was found that testin is predominantly expressed in the gonad (6). We postulated that this expression correlated with the rapid turnover of intercellular junctions in the testis and ovary during germinal cell development. This hypothesis was supported by the observations that testin mRNA can also be detected in non-gonadal tissues such as pre-and neo-natal rat kidney at the time of extensive tissue restructuring due to organ growth (8). Moreover, the steady-state testin mRNA level in the ovary is high at proestrus, estrus, and metestrus, correlating with the maturation of intrafollicular ova and the eventual rupturing of the follicle at ovulation that coincides with the rapid turnover of inter-granulosa cell junctions (8,9). In addition, testin expression was drastically reduced to an almost undetectable level at diestrus, during which functional regression of the corpora lutea occurs (8,9).
Other recent in vivo and in vitro studies reveal that testin is a sensitive marker to monitor the disruption of intercellular junctions in the testis, since the expression of testin is positively correlated to this event (8). For instance, a surge in testin expression and an intense accumulation of its protein in the cytosol of the testis are found when germ cells, mainly round and elongated spermatids, were depleted from the seminiferous epithelium by either lonidamine (8), busulfan (2,3), or X-irradiation (10) at the time when Sertoli-germ cell junctions were disrupted. A brief hypotonic treatment lysing germ cells in Sertoli-germ cell cocultures, thereby disrupting the inter-Sertoli-germ cell junctions, also induced a drastic increase in testin expression by Sertoli cells (8). These studies, however, cannot distinguish whether the observed drastic increase in testin expression correlates with the disruption of inter-Sertoli tight junctions or adherens and gap junctions, which are found between Sertoli cells as well as between Sertoli and germ cells (Table I; for reviews, see Refs. [11][12][13]. As such, we have used an established culture model that selectively disrupts tight junctions in vitro (14) by [Ca 2ϩ ] depletion using primary Sertoli cells cultured in vitro to assess whether such a disruption is associated with any changes in testin expression. In addition, we seek to (i) identify the distribution of testin in the epithelium by immunogold EM to define its subcellular localization, (ii) examine the effect of anti-testin IgG on the re-establishment of inter-Sertoli tight junctions after their disruption as assessed by the transepithelial electrical resistance (TER) 1

Preparation of Testicular Cell Cultures
Sertoli Cells Cultured at Low Density-Primary Sertoli cells were prepared from 20-day-old Sprague-Dawley rats as described previously (6,8). All the experiments using Sertoli cell-enriched cultures reported in this paper were derived from 20-day-old rats unless otherwise specified. Cells were plated at a density of 4.5 ϫ 10 6 cells/9 ml/100-mm dish (about 5 ϫ 10 4 cells/cm 2 ) in serum-free Ham's F12 nutrient mixture/ Dulbecco's modified Eagle's medium (F12/DMEM, 1:1, v/v) supplemented with insulin (10 g/ml), human transferrin (5 g/ml), bacitracin (5 g/ml), and epidermal growth factor (5 ng/ml) and incubated at 35°C in a humidified atmosphere of 95% air, 5% CO 2 (v/v). About 48 h after plating, cultures were hypotonically treated with 20 mM Tris-HCl, pH 7.4, for 2.5 min to lyse residual germ cells (17) to obtain Sertoli cell cultures with greater than 95% purity. Cells were washed twice with F12/DMEM, and the cells were allowed to recover for an additional 24 h before their use. Under these conditions, specialized tight junctions were not formed when assessed by various criteria as described previously (4).
Sertoli Cells Cultured at High Density-To assess the effect of antitestin IgG or [Ca 2ϩ ] depletion on the inter-Sertoli tight junction, primary Sertoli cells isolated as described above from 20-day-old rats were cultured at high cell density to allow the establishment of specialized junctions. Briefly, about 2 ϫ 10 6 cells/cm 2 were plated on Matrigel TM (1:8)-coated HA filters in the apical chamber of a bicameral unit (Millipore, Bedford, MA) (4). To assess the formation of tight junctions, TER across the Sertoli cell columnar monolayer was quantified using a Millicell electrical resistance system. Briefly, current was passed through the epithelial monolayer between two silver-silver chloride electrodes. Resistance was calculated from the change in voltage across the monolayer induced by a short (ϳ2 s) 20-A pulse of current. The resistance was multiplied by the surface area of the filter to yield the areal resistance in ohms/cm 2 . The net value of electrical resistance was then computed by subtracting the background, which was measured on Matrigel-coated cell-free chambers, from values of Sertoli cell-plated chambers. Disruption of the tight junctions were achieved by incubating the Sertoli cell monolayer in [Ca 2ϩ ]-free F12/DMEM for 15 min as described (14), which is manifested by a drastic decline in TER. Thereafter, cells were returned to [Ca 2ϩ ]-containing F12/DMEM with or without anti-testin IgG or normal rabbit serum IgG (200 g/ml) for 20 min at room temperature with gentle rocking and then returned to 35°C for the re-establishment of the tight junction as described previously (14). Disruption and reformation of the tight junction was assessed by TER measurement. IgG was purified from decomplemented sera (56°C for 30 min) by sequential ammonium sulfate precipitation and DEAE chromatography as described previously (18). Each time point contained triplicate cultures, and each experiment was repeated at least three times using different batches of cells.
Primary Sertoli Cell Cultures from 35-and 90-day-old Rats-Primary cultures of Sertoli cells from 35-and 90-day-old Sprague-Dawley rats were prepared essentially as described previously (19). The cells were suspended in F12/DMEM supplemented with various factors as described for immature Sertoli cells (see above) and plated at approxi-mately 4.5 ϫ 10 6 cells/9 ml/100-mm dish and cultured for 2 days at 35°C with 95% air, 5% CO 2 . Thereafter, cells were hypotonically treated to remove the residual germ cells (17). The resulting cell purity was about 85% when judged microscopically, and these cells were used for RNA extraction after 4 days in culture to compare the basal steadystate testin mRNA level between Sertoli cells isolated from rats of different ages.
Germ Cells-Total germ cells were isolated from 90-day-old Sprague-Dawley rat (about 300 gm b.w.) testes by a mechanical procedure without any enzymatic treatment as detailed elsewhere (20). These cell preparations consisted largely of spermatogonia, spermatocytes, and round spermatids with a relative percentage of 16:19:65 when verified by microscopic examination and DNA flow cytometry as described (20,21). Virtually all elongated spermatids were removed in the glass wool filtration step (20). Germ cells were cultured at a density of 22.5 ϫ 10 6 cells/9 ml of F12/DMEM supplemented with sodium DL-lactate (6 mM) and sodium pyruvate (2 mM) in 100-mm dishes for 20 h at 35°C to obtain germ cell-conditioned medium (GCCM) as described (20,21) or used immediately after their isolation for binding and coculture experiments. These cells were largely free of somatic cell contamination when assessed by various criteria as detailed elsewhere (20).
Sertoli-Germ Cell Cocultures-To assess the effects of germ cells on Sertoli cell testin expression, primary Sertoli cells isolated from 20-dayold rats were plated at 5 ϫ 10 4 cells/cm 2 and cultured for 2 days; thereafter, cells were hypotonically treated to remove contaminating germ cells (17) (day 0). Cells were then cultured for an additional 24 h (day 1), washed once, and cultured for 3 additional days (day 4) to ensure that the testin steady-state mRNA level had returned to base line since the culture procedure per se can induce damage on the intercellular junctions, which is a potent stimulator of testin expression (8). Thereafter, different amounts of germ cells using a Sertoli:germ cell ratio of 1:2.5, 1:5, and 1:10 were added to Sertoli cells and cultured for an additional 20 h (day 5). It must be noted that under these conditions, specialized junctions between Sertoli and germ cells had not yet formed, since their formation would require at least 24 -48 h in culture (22,23). Before their termination, cultures were subjected to a hypotonic treatment to lyse germ cells 20 min before the addition of RNA STAT-60 TM for RNA extraction to eliminate RNA contributed by germ cells in the sample to be analyzed. As such, the RNA that were examined in this experiment were largely derived from Sertoli cells. To assess the effect of the disruption of Sertoli-germ cell junctions, some cocultures that had incubated for 30 h (day 5) were subjected to a hypotonic treatment on day 5 that was 24 h before their termination on day 6. Control dishes were Sertoli-germ cell cocultures incubated until day 6 without hypotonic treatment on day 5 but subjected to a hypotonic treatment 20 min before termination to eliminate RNA contributed by germ cells in the sample to be analyzed.

Pulse-Chase Analysis of the Synthesis and Secretion of Testin by Sertoli Cells in Sertoli-Germ Cell Cocultures
To assess the effect of germ cells or GCCM on the synthesis and secretion of testin by immature Sertoli cells in vitro, Sertoli cells were prepared at a density of 4.5 ϫ 10 6 cells/9 ml/100-mm dish as described above. Twenty four h after the hypotonic treatment, germ cells (10 ϫ 10 6 cells/dish) or GCCM (10 g of protein/dish) were added and cocultured with Sertoli cells for 18 h in F12/DMEM containing 1/100th of the normal methionine concentration. The cells were then pulse-labeled for 15 min with 100 Ci of [ 35 S]methionine per dish and subsequently washed three times to remove any remaining [ 35 S]methionine. At specified time points, media and cells were harvested and stored at Ϫ20°C until used. Immunoprecipitation was performed as described (4). In one set of control experiments (Con 1), Sertoli cells were cultured alone without germ cells or GCCM and immunoprecipitated with the testin antibody at specified time points. In other controls (Con 2), Sertoli cells were cultured alone but immunoprecipitated with preimmune serum to assess the specificity of the testin antibody. Samples were resolved onto 10% T polyacrylamide gels and visualized by autoradiography.

RNA Extraction and Northern Blot Analysis
Total RNA was extracted from cell cultures or tissues using RNA STAT-60 TM as described previously (6,8,24). Northern blot analysis was performed as described previously using a ␣-32 P-labeled testin cDNA probe by nick translation (6) for hybridization. To ensure that equal amounts of RNA were loaded into each lane, some blots were rehybridized with a ␣-32 P-labeled ␤-actin cDNA probe (6), and data were normalized after densitometric scanning analysis.

Glycerol-induced Intercellular Junction Damage in the Testis
Recent studies have shown that glycerol can induce permanent damage of the blood-testis barrier and inhibit spermatogenesis by disrupting the inter-Sertoli cell tight junction near the basal lamina within 2 weeks after glycerol administration (25). As such, we seek to examine if there is any change in testin expression when the tight junctions between Sertoli cells are disrupted 2 weeks after glycerol treatment and to examine the changes of testin expression when germ cells are later depleted. A 20% glycerol solution was prepared in PBS (10 mM sodium phosphate, 0.15 M NaCl, pH 7.4) and used to induce intercellular junction damage in the testis after intratesticular injection as described (25). Adult Sprague-Dawley rats approximately 90 days of age (about 250 -300 gm b.w.) were used. Animals were anesthetized with Metofane R before treatment. Control and test animals received 200 l of PBS or 20% glycerol solution by intratesticular injection through the polar axis of each testis, respectively. Four to six rats were used per treatment group and rats were sacrificed by CO 2 asphyxiation at specified time points. Northern blot analysis and immunohistochemistry was performed as described previously (6,8,24). Tissue homogenates of the testis were also prepared as described previously (2).

Binding and Uptake of 125 I-Testin by Sertoli Cells
Purified testin isolated from SCCM as described previously (2) was radiolabeled with 125 I-Bolton-Hunter reagent (specific activity, 3056 -3324 Ci/mmol, NEN Life Science Products) to a specific activity of approximately 50 Ci/g of protein.
For binding experiments, Sertoli cells were plated in 12-well dishes at a concentration of 5 ϫ 10 5 cells/ well/ml F12/DMEM. These cells were used on day 3. Cells were washed three times with F12/DMEM containing 1% BSA. Approximately 1 ϫ 10 6 cpm of 125 I-testin were added per well in duplicate with or without a 100-fold excess of unlabeled testin and incubated at 35 or 4°C for several different time points. Cells were then washed three times in F12/DMEM containing 1% BSA (w/v) to remove unbound 125 I-testin. Cells were solubilized in 1 M NaOH for radioactivity determination. Specific binding was determined from the difference between total binding and nonspecific binding (in the presence of 100-fold excess unlabeled testin). Radioactivity in solubilized cell extracts was determined in a Packard Cobra II ␥-counter.

Membrane Preparation and Solubilization of 125 I-Testin
Primary monolayer cultures of Sertoli cells were prepared from 20day-old rats as described above and plated at a density of 4.5 ϫ 10 6 cells/9 ml/100-mm dish. Four days after hypotonic treatment, Sertoli cells were washed twice with media and incubated with 5 ϫ 10 6 cpm 125 I-testin/5 ml of F12/DMEM for 4 h at 35°C. Cells were washed three times with fresh media and scraped off in PBS containing 2 mM of phenylmethylsulfonyl fluoride and 2 mM N-ethylmaleimide to inhibit protease activity. Cell membranes were disrupted by repeated freezethawing. Membranes were pelleted for 15 min at 12,000 ϫ g and washed three times to remove residual cytosolic proteins. The membranes were then solubilized with 100 l of corresponding buffer and heated at 100°C for 10 min, and testin was visualized by autoradiography after SDS-PAGE. The membrane solubilization buffers consisted of 20 mM Tris containing 1.6% 2-mercaptoethanol with or without 1% SDS, 1% Triton X-100, or 1% Nonidet P-40. All buffers were adjusted to pH 6.8 at 22°C. Extracellular matrix (ECM) was the remaining amorphous substances in the dishes used for membrane preparation as described previously (52). It was extracted with the same buffers as described above by heating at 100°C for 10 min.

Sertoli Cell Membrane and Cell Surface Labeling
Sertoli cells were cultured at high cell density (0.75 ϫ 10 6 cells/cm 2 ) on Matrigel (1:5 diluted with F12/DMEM)-coated dishes as described (4) to allow the formation of specialized junctions. Cell membranes were prepared as described above for iodination. For surface labeling, live cells were cross-linked with 3,3Ј-dithiobis(sulfosuccinimidyl) propionate (DTSSP, a membrane-impermeable and thiol cleavage cross-linker) as described by the manufacturer (Pierce) before Na[ 125 I] labeling using Iodogen (26), and therefore only proteins on the cell surface were labeled. About 500 g of protein of Sertoli cell membrane or DTSSPcross-linked cells (12 ϫ 10 6 ) were labeled with 2.5 mCi of Na[ 125 I] in a Pierce Reacti-vial coated with 100 g of Iodogen (26). Free Na[ 125 I] was removed by dialysis against 6 liters of 20 mM Tris, pH 7.4 with 2-3 changes over a period of 2 days. 125 I-Membrane ghosts were solubilized in 20 mM Tris, pH 7.4, containing 0.1% Brij 97, 0.1% SDS, 0.1% Triton X-100, 2 mM phenylmethylsulfonyl fluoride, and 2 mM EDTA for 3 h at 37°C. Samples were dialyzed overnight in PBS to remove detergents.
Immunoprecipitation was then performed using either testin antibody or preimmune serum (1:50) overnight at 4°C with agitation on a rotator. Immunocomplexes were purified on a protein A-Sepharose column (0.75 ϫ 10 cm, inner diameter) previously equilibrated with 10 column volumes of PBS. Nonbound proteins were removed by a 15-column volume of PBS. Specifically bound proteins were eluted from the column using 0.1 M glycine, pH 3.5, and collected in 0.5-ml fractions. Aliquots (5 l) were counted in a Packard Cobra II ␥-counter. Selected fractions were resolved by SDS-PAGE onto a 10% T SDS-polyacrylamide gel and visualized by autoradiography. The monospecificity of the testin antibody used in this study has been characterized and described (1,2,4).

Ultrastructural Localization of Testin in the Seminiferous
Epithelium by Immunogold Electron Microscopy.
For immunogold EM study, adult male Sprague-Dawley rats (250 gm b.w.) were used, and the animals were sacrificed by CO 2 asphyxiation. Testes were perfused via the testicular artery with ice-cold physiological saline to remove contaminating blood for 10 s, followed by 0.1 M PBS, pH 7.2, containing 0.2% glutaraldehyde and 4% paraformaldehyde for 30 min. The testes were quickly cut into small pieces and immersed in the same fixatives containing 4% sucrose at 4°C for 2 h. After thoroughly washing with the same buffer, samples were dehydrated in ascending concentrations of ethanol at low temperature: 30% ethanol, 30 min at 4°C; 50% ethanol, 30 min at Ϫ20°C; 70% ethanol, 30 min at Ϫ35°C; 95% ethanol, 30 min at Ϫ35°C; 100% ethanol, twice for 30 min at Ϫ35°C. Lowicryl K4M infiltration and embedding solutions were prepared according to the manufacturer's specifications (Polysciences, Inc. Warrington, PA). Dehydrated specimens were infiltrated as follows: Lowicryl/ethanol 1:1, 60 min at -35°C; Lowicryl/ethanol 2:1, 60 min at Ϫ35°C; and Lowicryl overnight at Ϫ35°C. The specimens were then transferred to Lowicryl embedding solution in precooled beam capsules. Polymerization was induced by UV irradiation at Ϫ35°C overnight followed by 2 days of UV irradiation at room temperature. Semi-thin sections stained with toluidine blue were used to select relevant tissue and cells. Thin sections cut by an ultramicrotome were mounted on formaval-coated nickel grids for subsequent immunogold localization. The sections were exposed to undiluted goat serum for 1 h at room temperature, then incubated overnight at 4°C on drops of affinity-purified anti-testin IgG. Affinity chromatography using protein A-Sepharose 4B (Amersham Pharmacia Biotech) was performed according to the manufacturer protocol. After washing in PBS (4ϫ, 15 min), the grids were floated for 30 min at room temperature on drops of colloidal gold-labeled goat anti-rabbit IgG second antibody (particle size 10 nm, AuroProbe EM GAR G10, Amersham) diluted 1:40 in PBS containing 0.1% BSA. The grids were thoroughly washed with PBS and then fixed with 2.5% glutaraldehyde and post-fixed in 1% OsO 4 . After drying in the desiccator, the grids were stained by lead citrate and uranyl acetate. Normal rabbit serum and omission of testin antiserum were used as controls. Sections were viewed with an Hitachi 800 electron microscope.

General Methods
Protein estimation was performed as described previously (30) using BSA as a standard. Statistical analysis was performed by Student's t test using the GB Statistical Analysis Software package (Version 3.0) (Dynamics Microsystems, Inc., Bethesda, MD).

Distribution of Testin in Biological Fluids and Reproductive Tissue Extracts and Its Comparison to Three Other Sertoli Cell
Secretory Products-To better understand the biological features of testin, we compared its distribution in various biological fluids and compartments to three other Sertoli cell secretory products: namely androgen binding protein, clusterin, and ␣ 2 -macroglobulin in the rat by RIA. When Sertoli cells were cultured in the monolayer at 5 ϫ 10 4 cells/cm 2 , when some of the specialized junctions did not form (such as occluding tight junctions, see Table I), the amount of testin secreted by Sertoli cells in vitro was comparable with that of androgen binding protein, clusterin, and ␣ 2 -macroglobulin (Fig. 1A). However, the level of testin in rete testis fluid from intact adult testes was several orders of magnitude lower than these three other secretory proteins, which were highly concentrated in the rete testis fluid (Fig. 1B). By any measure, it is logical to expect that any secretory proteins of Sertoli cell origin be concentrated in the luminal fluid since the blood-testis barrier formed by adjacent Sertoli cells limits the transport of proteins from the systemic circulation to the interstitium unless it is being actively transported there. On the other hand, testin may be rapidly reabsorbed by either Sertoli or germ cells after its secretion by Sertoli cells. If testin is indeed secreted basally, it is expected to be detected in the serum at high concentration; however, earlier RIA data revealed a very low level of testin in the serum, and its concentration did not alter after orchiectomy (2). As such, results shown in Fig. 1B illustrate the unique feature of testin, indicating this protein is not accumulated in the luminal fluid and is likely to be reabsorbed onto the testicular cell. When the levels of testin in the cytosols of the intact testis ( Fig. 1C) and epididymis (Fig. 1D) were quantified and compared with these other Sertoli cell proteins, it was found that testin was almost negligible in both organs when compared with these other Sertoli cell proteins, further suggesting that the secreted testin might become reabsorbed by testicular cells.
Binding of 125 I-Testin by Sertoli Cells, Germ Cells, and ECM-To investigate whether testin is bound to the Sertoli cell membrane, the following experiments were performed. Highly purified Sertoli cells prepared as described under "Materials and Methods" were plated in 12-well dishes at 5 ϫ 10 5 cells/ well/ml F12/DMEM. Briefly, 125 I-testin (1 ϫ 10 6 cpm, about 10 fmol) was incubated with cells for specific time points at 35 or 4°C in F12/DMEM ( Fig. 2A). Nonspecific binding was estimated by using 100-fold excess unlabeled purified testin. Cells were then washed three times in F12/DMEM to remove unbound 125 I-testin. Cells were subsequently solubilized in 1 M NaOH for radioactivity determination. Fig. 2A shows that Sertoli cells bound 125 I-testin specifically in a temperature-dependent manner of which the binding was saturated by 2 h. We next investigated whether testin can also bind to germ cells or ECM. Sertoli cells were cultured at 4.5 ϫ 10 6 cells/9 ml/100-mm dish for 2 days, contaminating germ cells were lysed by a hypotonic treatment (17), and the cultures were incubated at 35°C for 4 days. Highly purified Sertoli cells or germ cells were then incubated with 125 I-testin (about 5 ϫ 10 6 cpm) for 4 h at 35°C. The dishes were washed, and cell membranes were isolated as described under "Materials and Methods." Cell membranes and ECM were then solubilized in the corresponding buffer, heated at 100°C for 10 min, and resolved by SDS-PAGE, and the 125 I-testin was visualized by autoradiography. In Fig. 2B, lanes 1 and 10 are the 125 I-testin used for incubation, showing the two molecular variants of testin I and testin II. Lanes 2-5 are the Sertoli cell membranes solubilized with 20 mM Tris containing 1.6% 2-mercaptoethanol; 20 mM Tris containing 1% SDS; 20 mM Tris containing 1% SDS and 1.6% 2-mercaptoethanol; and 20 mM Tris containing 1% SDS, 1% Triton X-100, 1% Nonidet P-40, and 1.6% 2-mercaptoethanol, respectively. All buffers used were adjusted to pH 6.8 at 22°C. Lanes 6 -9 are the ECM on the Petri dish extracted with the same buffers as shown in lanes 2-5. These results indicate not only that 125 I-testin binds onto the Sertoli cell membrane and ECM, but its solubilization requires the use of detergents, suggesting that its association with the cell membrane is not a nonspecific attachment. In the absence of detergents (lanes 2 and 6 versus lanes 3-5 and 7-9, Identification of a Protein Complex on the Sertoli Cell Membrane That Binds Testin-We next sought to investigate whether the binding of 125 I-testin onto the cell membrane as shown in Fig. 2 is mediated by a binding protein. Sertoli cells were cultured at high cell density on Matrigel-coated dishes for 4 days to allow the formation of specialized junctions. Plasma membranes were isolated as described under "Materials and Methods," and the whole membrane fraction was labeled with Na[ 125 I] by Iodogen (26). Immunoprecipitation was then performed on the labeled membrane proteins using either testin antibody or preimmune serum (1:50). The monospecificity of this antiserum has been characterized and established (2,4,6). Immunocomplexes were purified by protein A-Sepharose column and resolved on a SDS-polyacrylamide gel and shown in Fig. 3A. Lane 1 (S) is the 14 C-methylated protein standard. Lane 3 is 125 I-testin, where the two molecular weight variants of testin were clearly visible. Lanes 2 and 4 are the Sertoli cell membrane (SM) proteins after cell labeling and immunoprecipitation run under reducing (R) and nonreducing (NR) conditions, respectively. Radiolabeled testin I and II extracted from the membrane are clearly visible in the SM samples (Fig.  3A, lanes 2 and 4) when compared with 125 I-testin alone (lane 3). It was noted that two other cell membrane proteins that were labeled and immunoprecipitated in conjunction with testin, designated as a (45 kDa) and b (28 kDa) are also visible in lanes 2 (reducing) and 4 (nonreducing), indicating that they are single polypeptide chains. Lanes 5 and 6 are the corresponding controls of lanes 2 and 4, where the testin antibody incubation was substituted with preimmune serum, illustrating the bands a and b shown in Fig. 3A are components of the binding protein complex. These results, however, cannot distinguish whether these two peptides reside on the cell surface or are found inside the membrane. In contrast to the whole membrane labeling, Fig. 3B shows the result obtained from the surface labeling of viable Sertoli cells previously treated with a membrane-impermeable and thiol-cleavable cross-linker, DTSSP. Lane 1 (S) is the 14 C-methylated protein standard. Lane 2 is 125 I-testin, where the two molecular variants of testin are clearly visible. Lanes 3 and 4 are the Sertoli cell surface-labeled proteins after extraction and immunoprecipitation run under reducing (R) and nonreducing (NR) conditions, respectively. Radiolabeled testin I and II in conjunction with the 28-kDa protein, desig- nated b, are clearly visible in the DTSSP-treated surface-labeled sample (Fig. 3B, Lane 3) under reducing conditions. Under nonreducing conditions, the band b together with testin were not visible and possibly retarded in the gel because the complex had not been cleaved with a reducing agent, since DTSSP is a thiol-cleavable cross-linker (Fig. 3B, lane 4 versus lane 3). These results indicate that the 28-kDa protein (band b) and testin I and II reside on the cell surface, and the 45-kDa protein a is localized inside the cell membrane. In both sets of experiments, the gels were also Coomassie Blue-stained to visualize any proteins that might not have been labeled. However, no other proteins were identified. Localization of Testin in the Rat Testis by Immunogold EM-To examine the subcellular localization of testin, immunogold EM was carried out, and about 300 cross-sections were examined. Fig. 4 summarizes the result of these analyses. Immunoreactive testin as marked by gold (black) particles was abundantly localized near the surface of a Sertoli cell adjacent to the ectoplasmic specialization (a modified adherens junction) around a developing spermatid (Fig. 4A). The Sertoli cell was typified by the presence of microtubule bundles (Fig. 4A, arrowheads). Very few gold particles representing immunoreactive testin were found along the tight junction between two adjacent Sertoli cells (Fig. 4B). These data are in agreement with previously published immunofluorescent microscopy and immunohistochemistry revealing the abundant presence of testin between Sertoli and germ cells (5-7). Fig. 4C is a control section stained with IgG isolated from preimmune serum showing a cross-section between two Sertoli cells as typified by the presence of actin filament bundles (shown by the arrowheads) where no gold particles were seen. The plasma membranes may not be obvious in these sections, probably due to the omission of OsO 4 in the initial tissue fixation, since the use of this fixative caused a loss of antigenicity of the testin molecule, making localization impossible.

Effect of Anti-testin IgG on the Formation of Tight Junctions and the Expression of Testin after Their Disruption by Calcium
Depletion-Since earlier studies reveal a tight relationship between the expression of testin and the integrity of intercellular cell junctions in the testis (8), we sought to examine whether a disruption of the inter-Sertoli tight junction can induce a surge in testin expression similar to what was shown when the Sertoli-germ cell junctions were disrupted (8). It is known that epithelial tight junctions, such as those found in the Madin-Darby canine kidney cells in vitro, can be disrupted by the removal of calcium ions from the medium and can be quickly reassembled after its replacement (14,33,34). The integrity of the inter-Sertoli tight junctions was assessed by TER measurement (31,32). The inter-Sertoli tight junction was disrupted by [Ca 2ϩ ] depletion in primary Sertoli cell cultures. Sertoli cell cultures were prepared as described under "Materials and Methods," and TER measurements were taken thereafter. It was noted that tight junctions were established in these Sertoli cell cultures by day 3-4 ( Fig. 5A), when the TER reached its plateau with a measurement of about 50 -60 ohms/cm 2 , which is similar to a previous report (35). On day 5, extracellular [Ca 2ϩ ] was removed from the bicameral unit by rinsing the cell monolayer gently with Ca 2ϩ -free F12/DMEM, and the units were incubated in Ca 2ϩ -free F12/DMEM for 15 min. At the end of this 15-min period, a significant decline in TER was noted (Fig. 5A), illustrating the tight junction had become leaky. This result is consistent with other tight junction forming epithelia such as Madin-Darby canine kidney cells (14). In some experiments, cells were incubated with or without either anti-testin IgG (200 g/ml) or normal rabbit IgG (200 g/ml) together with fresh F12/DMEM containing [Ca 2ϩ ] to assess the effect of testin antibody on the reformation of tight junctions between Sertoli cells. It was noted that the tight junctions were reestablished within 90 min (Fig. 5A) as shown by an increase in TER. However, the presence of anti-testin IgG did not interfere with the reformation of tight junctions in these cultures (Fig.  5A). More important, when the testin steady-state mRNA level was assessed in the samples when the tight junctions were being disrupted and re-established, no detectable change in testin expression was noted (Fig. 5B). When the same blot shown in Fig. 5B was rehybridized with a ␤-actin probe (Fig.  5C) and the data were densitometrically scanned and normalized against ␤-actin (Fig. 5D), testin expression was found not to correlate with the disruption or formation of tight junction, which is entirely different from the testin mRNA expression

FIG. 2. Binding of 125 I-testin to primary Sertoli cell cultures (A), and its solubilization from Sertoli cells (SC), germ cells (GC), and ECM (B).
A, binding assay was performed as described under "Materials and Methods." 125 I-Testin (1 ϫ 10 6 cpm) was incubated at 35 or 4°C for specified time points. Nonspecific binding was estimated using a 100-fold excess of unlabeled testin. B, autoradiograph showing the solubilization of membrane-bound 125 I-testin from SC, GC, and ECM. 125 I-Testin (5 ϫ 10 6 cpm/5 ml/100-mm dish) was incubated with primary Sertoli cells for 4 h at 35°C as described under "Materials and Methods." Membrane proteins were solubilized in 100 l of corresponding buffer, heated at 100°C for 10 min, and resolved on a 10% T SDS-polyacrylamide gel, and testin was visualized by autoradiography. Lanes 1 and 10, 125 I-testin tracer. Lanes 2-5, SC membranes solubilized with either 20 mM Tris containing 1.6% 2-mercaptoethanol, 20 mM Tris containing 1% SDS and 1.6% 2-mercaptoethanol, 20 mM Tris containing 1% SDS, or 20 mM Tris containing 1% SDS, 1% Triton X-100, 1% Nonidet P-40, and 1.6% 2-mercaptoethanol, respectively. All buffers were adjusted to pH 6.8 at 22°C. Lanes 6 -9 is the remaining ECM on the SC dishes extracted with the same buffers as shown in lanes 2-5. Primary adult germ cells were also incubated for 4 h at 35°C with 125 I-testin (lane 11) or with 125 I-testin plus a 100-fold excess of unlabeled testin to assess nonspecific binding (lane 12). The germ cell membranes were solubilized in SDS sample buffer and heated at 100°C for 10 min.
during the disruption of Sertoli-germ cell junctions (8). In other experiments, the testin expression was also monitored up to 5 days after tight junctions were disrupted in vitro by Ca 2ϩ depletion, and no changes in testin expression were noted (data not shown). Also, Sertoli cells were cultured with anti-testin IgG or normal rabbit IgG in F12/DMEM between days 1 and 4 when tight junctions were being formed; no differences in TER were noted between these cultures and control cultures without any IgG (data not shown). These analyses clearly illustrate the lack of correlation between testin expression and the formation or disruption of inter-Sertoli tight junctions.

Changes in Testin Expression in the Testis by Glycerol-induced Disruption of the Blood-Testis
Barrier-To confirm the lack of correlation between testin expression and the disruption of inter-Sertoli tight junctions, we sought to use an in vivo animal model in which the tight junctions were disrupted followed by a progressive disruption of Sertoli-germ cell junctions due to germ cell depletion. Administration of glycerol intratesticularly is known to cause long term cessation of spermatogenesis in rats without any apparent changes on Leydig cell function, serum gonadotropin and testosterone levels, and secondary sexual characteristics (36,37). Recent studies have shown that glycerol can induce permanent damage of the bloodtestis barrier by disrupting the inter-Sertoli tight junction near the basal lamina within 2 weeks after glycerol administration (25). A disruption of the blood-testis barrier was manifested by an influx of [ 3 H]inulin and [ 125 I]albumin to the rete testis fluid, seminiferous tubule fluid, and the testicular tissue within 2 weeks post-glycerol treatment after in vivo infusion of radiolabeled substances (25). We have examined the effect of glycerolinduced disruption of inter-Sertoli tight junctions and the subsequent damage on Sertoli-germ cell junctions as a result of germ cell depletion on the steady-state testicular testin mRNA level (Fig. 6). The morphological changes in the testis after glycerol treatment were also examined immunohistochemically where testin appears as a reddish-brown precipitate (Fig. 7). Adult rats were anesthetized with Metofane R and received control (PBS) or treatment solution (200 l of PBS containing 20% glycerol) injected via a 26-gauge needle through the polar axis of each testis beginning at one pole and terminating at the other (25). 2, 4, and 8 weeks later, testes were removed for RNA extraction and Northern blot analysis. Fig. 6A is a Northern blot using about 20 g of total RNA per lane. No changes in the testicular testin steady-state mRNA level were apparent by 2 weeks at the time when tight junctions were disrupted (Fig. 6,  A and B). It was found that virtually all germ cells were still present in the epithelium by 2 weeks after glycerol administration ( Fig. 7C versus Fig. 7, A and B). However, a significant increase in the testin steady-state mRNA level (Fig. 6, A and  B), which was also accompanied by an accumulation of testin (Fig. 6, C and D) in the testis was clearly visible by 4 and 8 weeks after glycerol treatment when germ cells, in particular round and elongated spermatids, were depleted from the seminiferous epithelium at 4 -8 weeks (Fig. 7D and E versus Fig. 7,  A, B, and C). When the concentrations of testin in the testis of these rats were quantified by RIA and compared with control rats, a significant increase in testin concentration was noted (Fig. 6C). Since there was a significant decline in testicular weight by 8 weeks after glycerol treatment, the changes in testin level were taken into account with the reduction in testicular weight. Once the data were expressed as testin per pair testes (Fig. 6D), a 7-fold increase was detected in glyceroltreated rats by 8 weeks. These results are also consistent with the immunohistochemistry data, since there is drastic increase in immunoreactive testin accumulated in the tubular lumen at 4 -8 weeks after glycerol treatment (Fig. 7, D and E) compared with control rats (Fig. 7B) and rats treated with glycerol for 2 weeks (Fig. 7C). Glycerol, however, had no effect on testin expression in the epididymis up to 8 weeks post-treatment (Fig.  6, A and B), illustrating the specificity of this chemical treatment in the testis. However, immunohistochemistry analysis revealed that testin was accumulated in the epididymal lumen 4 -8 weeks post-glycerol treatment (data not shown).
Effect of GCCM or Germ Cells on Sertoli Cell Testin Steadystate mRNA Level-It is known that germ cells neither express testin mRNA (7) nor do they secrete any testin in vitro (20). But it is not known whether GCCM or germ cells can regulate testin expression. To examine such a possibility, Sertoli cells cultured at 5 ϫ 10 4 cells/cm 2 when specialized junctions did not form were incubated with increasing concentrations of GCCM proteins for a 20-h period (Fig. 8A). Fig. 8A shows that GCCM had no affect on the Sertoli cell testin steady-state mRNA level. Fig. 8B is the same blot such as the one shown in Fig. 8A but hybridized with a ␤-actin cDNA probe. Fig. 8C is the densitometrically scanned data of three separate Northern blots normalized against ␤-actin, indicating that germ cell-released proteins had no apparent effect on Sertoli cell testin expression.

. Immunogold EM localization of testin in the rat testis.
A and B are sections of Lowicryl-embedded rat testes stained with affinity-purified anti-testin IgG followed by a second antibody conjugated to gold particles. Gold particles corresponding to immunoreactive testin appeared as black dots in these figures. A, this section shows a transverse section of a SC and a developing spermatid (germ cell (GC)), as manifested by the presence of the acrosome cap (Ac) and nucleus (Nu). Gold particles accumulated at the sites of ectoplasmic specializations on the Sertoli cell side near the cell surface. Arrowheads indicate the microtubule bundles, which are one of the junctional elements of ectoplasmic specializations in Sertoli cells. Bar, 0.3 m. B, this shows the cross-section to actin filament bundles (arrowheads) between two transversely sectioned adjacent Sertoli cells exhibiting the fine structure of the tight junction. Immunoreactive testin was found to associate with actin bundles in the tight junction. Bar, 0.1 m. C, this is a control micrograph in which the anti-testin IgG was substituted with rabbit IgG. This is a parallel section of two SC, where the actin filaments (arrowheads) can be seen. No gold particles were found. Bar, 0.1 m. the corresponding spent media, suggesting GCCM did not affect Sertoli cell testin secretion. However, a transient but consistent increase in the concentration of testin in the spent media was noted using 5 g of GCCM protein/dish (Fig. 8D).
We next examined whether germ cells can affect Sertoli cell testin expression in vitro (Fig. 9). Since the isolation of Sertoli cells from the seminiferous tubule involves the disruption of Sertoli-germ cell junctions, which would have induced changes in testin expression, Sertoli cells used for this experiment were cultured for 4 days before the addition of germ cells to allow its testin steady-state mRNA level to return to the basal level. Lane 1 is total RNA derived from Sertoli cells cultured for 48 h in vitro; cells were terminated 20 min before the hypotonic treatment (Day 0). Lane 2 is RNA from Sertoli cell cultures 24 h after hypotonic treatment (Day 1). Lane 3 is the Sertoli cell RNA isolated from cultures on day 5 that served as a control for lanes 4 -6. As expected, a steady decline in testin expression during cultures is noted (Fig. 9A, lanes 1-3), suggesting that the isolation of Sertoli cells from the seminiferous tubule can enhance testin expression because these steps disrupted the Sertoli-germ cell junctions. The hypotonic treatment step on day 0 (i.e. 48 h after Sertoli cells were isolated from the tubules) could not induce a surge in testin expression in these cultures (Fig. 9A, lane 2 versus lane 1) possibly because these primary cultures were relatively free of germ cells; as such, not many Sertoli-germ cell junctions could be disrupted. Alternatively, the expression of testin at this time was already maximized, and the removal of residual germ cells could no longer elicit an additional increase in testin expression. Different germ cell numbers using a Sertoli:germ cell ratio of 1:2.5, 1:5, and 1:10 were then plated onto these Sertoli cell cultures on day 4 and incubated for an additional 20 h (day 5) to examine the effects of germ cells. Immediately before RNA extraction, each culture dish was hypotonically treated to lyse germ cells to eliminate RNA contributed by germ cells in the sample to be analyzed. As such, total RNA extracted from these dishes were largely of Sertoli cell origin. It was found that germ cells did not generate a dose-dependent and significant effect on the Sertoli cell testin expression as shown in Fig. 9, A-C (lanes 4 -6 versus lane 3), even though it is apparent that the presence of germ cells reduced Sertoli cell testin expression slightly. It must be noted that under these conditions, specialized Sertoli-germ cell junctions did not form since their formation would require an incubation period of at least 24 -48 h in vitro (22,23). In some experiments, the Sertoli-germ cell cocultures were allowed to incubate for 30 h in vitro and were subjected to a hypotonic treatment on day 5 to disrupt the Sertoli-germ cell junctions, and the cells were harvested on day 6 for analysis. A significant increase in testin expression was found as a result of the disruption of the junctions (Fig. 9A, lane 7 versus lane 8).
Effect of Germ Cells or GCCM on Testin Synthesis and Secretion-To study the effect of germ cells or GCCM on Sertoli cell testin secretion, Sertoli cells (4.5 ϫ 10 6 cells/9 ml/100-mm dish) were cocultured with germ cells (1 ϫ 10 7 ) or GCCM (10 g of protein) for 18 h in vitro. Thereafter, cells were pulse-labeled with [ 35 S]methionine for 15 min and chased with cold methionine at specified time points. Fig. 10, A and B show the relative amounts of 35 S-labeled testin in the cytosol and media, respectively, at specified time points. It was noted that testin appeared in the cytosol within 15 min with a testin I:testin II ratio of 3:1 when the x-ray film was densitometrically scanned at 600 nm (Fig. 10A). By 24 h, all the newly synthesized 35 S-labeled testin was no longer detectable in the cytosol but secreted into the medium (Fig. 10A), indicating that germ cells did not inhibit testin secretion into the media. Testin was detected in the medium within 1 h after pulse-labeling and peaked at 24 h, concomitant with the disappearance of testin in the cytosol, but the ratio of testin I:testin II had shifted, becoming 1:1.5 (Fig. 10, B versus A). The reason for such a shift in ratio is not immediately known; it is possibly the result of post-translational processing such as glycosylation. Con 1 is the corresponding control using Sertoli cells without either germ cells or GCCM, whereas Con 2 is the corresponding control using preimmune serum instead of the anti-testin antibody for immunoprecipitation. The level of [ 35 S]-labeled testin shown in Con 1 using Sertoli cells cultured alone without either germ cells or GCCM is not significantly different from the corresponding cultures with either germ cells or GCCM (Fig.  10, A and B).
Age-dependent Expression of Testin-Earlier studies have shown that the expression of testin in adult rats is confined to the gonad (6). When the steady-state mRNA level of testin from the seminiferous tubule isolated from 10, 20, and 60 days of age was examined by Northern blot, it was found that tubules from immature rats had almost 10-fold more testin mRNA than mature rats (8). However, these earlier studies did not take into consideration the increase in RNA contributed by germ cells in the samples being analyzed, since there is a drastic increase in germ cell:Sertoli cell ratio during maturation. When the changes in testin steady-state mRNA level in the testis were quantified by Northern blot, it was found that there is a steady increase in testin expression per pair of testes during maturation (Fig. 11), illustrating that testin expression correlates with the onset of spermatogenesis. We next examined the testin steady-state mRNA level in Sertoli cells isolated from testes at different ages. Unexpectedly, an age-dependent reduction of testin expression (Fig. 12A) was noted when about 10 g of total RNA was used for analysis (Fig. 12B). These blots were then rehybridized with a ␤-actin probe and densitometrically scanned at 600 nm, and the data were normalized against ␤-actin. An 8-fold reduction in testin expression was observed during maturation, when the steady-state mRNA level was compared between Sertoli cells isolated from 20-and 90-dayold rats (Fig. 12C). These results illustrate that an age-dependent reduction in testin expression likely correlates with the differentiation status of the Sertoli cell. Table I summarizes the three types of specialized junctions that are present in other epithelia that are also found in the mammalian testis (for reviews, see Refs. 11-13). All three types of junctions, namely occluding, anchoring, and communicating junctions, can be found between Sertoli cells. However, only anchoring and communicating junctions are present between Sertoli and germ cells. Moreover, some of the anchoring junctions between Sertoli and germ cells such as tubulobulbar complexes are unique to the testis (Table I). Surprisingly, the component proteins that constitute these junctions between Sertoli and germ cells, let alone the molecules that regulate the disassembly and reassembly of these junctions throughout different stages of spermatogenesis as a result of migration of developing germ cells from the basal lamina to the adluminal compartment of the seminiferous epithelium (for reviews, see Refs. 11,12), are largely unknown. Although the mechanisms and cellular events that regulate germ cell translocation in the epithelium are poorly understood, studies in organogenesis, embryogenesis, tumor growth, and metastasis have yielded some crucial information with regard to germ cell movement, since these other cellular processes also involve extensive turnover of cell-cell and cell-matrix interactions as well as cell migration (for reviews, see Refs. 38 and 39). It is known that FIG. 6. A study on the relationship between glycerol-induced disruption on inter-testicular cell junctions and testin expression as well as its protein accumulation in the rat testis. A, a Northern blot using about 20 g of total RNA/lane from testis and epididymis of rats at 2, 4, and 8 weeks (W) after an intratesticular injection of glycerol including control rats. kb, kilobases. B, same blot as shown in A but rehybridized with a ␤-actin cDNA probe. C, testin concentration in the testicular cytosol in glycerol-treated rats 8 weeks post-treatment versus control (n ϭ 4) when quantified by RIA. D, changes in testin content per pair of testes 8 weeks after glycerol treatment when the reduction in organ weight was taken into account (n ϭ 4). In both C and D, results are mean Ϯ S.D. of four rats. *, significantly different from control; p Ͻ 0.001. both embryogenesis and tumor growth involve the participation of proteases, protease inhibitors, signaling molecules, growth factors, and junctional complex components such as cell adhesion molecules (for reviews, see Refs. 40 -45). For the past decade, studies from different laboratories have identified several of these component molecules in the testis (for reviews, see Refs. 11-13 and 46). Recent studies from this laboratory have also illustrated the involvement of proteases, protease inhibitors, and cell adhesion molecules in the early stage of Sertoligerm cell interactions preceding the establishment of specialized cell junctions in vitro (47,48), suggesting that the establishment of specialized junctions between testicular cells is not simply a series of morphological events. Instead, multiple factors are involved, illustrating the complexity of junction formation in the seminiferous epithelium. The present report demonstrates testin is likely to be a sensitive marker to probe the events of cell junction disruption in the testis.

DISCUSSION
In this study, testin was shown to correlate with the disruption of Sertoli-germ cell junctions but not the inter-Sertoli tight junction. Several lines of evidence have excluded the involvement of testin in the formation and/or disruption of tight junction. First, immunogold EM revealed very few testin are found in the tight junction between Sertoli cells. Second, anti-testin IgG did not interfere or facilitate the formation of tight junctions between Sertoli cells in vitro, nor did it affect the recovery of tight junctions after [Ca 2ϩ ] depletion induced disruption of the inter-Sertoli tight junction. Third, the expression of testin is not affected by the disruption of tight junctions in vitro, as demonstrated in the [Ca 2ϩ ] depletion experiment. This observation is in sharp contrast to the disruption of Sertoli-germ cell junctions, which is accompanied by a surge in testin expression (8). Fourth, changes in testin expression after glycerol treat-ment did not coincide with the damage of the blood-testis barrier by 2 weeks (25) but rather, with the depletion of germ cells, which disrupted the Sertoli-germ cell junctions. We thus conclude that the testin that was found in both the adluminal and basal compartments of the seminiferous epithelium as visualized by immunofluorescent microscopy and immunohistochemistry (5-7) is the protein localized between Sertoli and germ cells, most likely at the adherens junction such as the desmosome-like junction, ectoplasmic specialization, and tubulobulbar complex. It is unlikely that testin is involved in the gap junction, since the component molecules of the gap junction have been very well characterized, and the primary sequence of testin does not bear any homology to any of the existing connexin family members (for review, see Ref. 13).
It has been shown that the testin steady-state mRNA level in the adult rat testis is significantly enhanced as a result of either a chemical treatment such as glycerol, lonidamine (8), and busulfan (3) or a physical treatment such as hypotonic treatment (8) and X-irradiation (10). The depletion of germ cells after these treatments would undoubtedly disrupt the Sertoli-germ cell junctions. It is our belief that several factors may be operating independently or synergistically that regulate testin expression as shown in these in vitro and in vivo experiments. First, germ cells may regulate testin expression via cell-cell contacts or through a factor(s) released from germ cells. As such, a depletion of germ cells leads to a change in testin expression. Second, testin may be a structural component of the Sertoli-germ cell junction. Thus, when the intercellular junction is damaged, another yet-to-be identified factor is released to stimulate the production of testin to replace the lost structural component or to trigger another cascade of events. Third, testin may be a stress-induced protein in response to the FIG. 7. Morphological changes and the associated pattern of testin immunohistochemical localization in adult rat testes after intratesticular glycerol treatment. Immunoreactive testin appears as a reddish-brown precipitate as denoted by arrowhead. A and B are photomicrographs of the cryostat sections of testes from a control rat. C, D, and E are photomicrographs of rats treated with glycerol after 2, 4, and 8 weeks, respectively. Four animals in each treatment group including control were processed for microscopic examination, and at least 50 -100 sections were examined. A single set of representative data are shown here. A, control rat showing the cross-section of a normal seminiferous tubule at stage VII of the spermatogenic cycle stained with preimmune serum indicating the specificity of the staining. B is the same as in A, except that it was stained with testin antibody. Immunoreactive testin was found between spermatocytes/spermatogonia and Sertoli cells at the basal compartment. Testin was also found between Sertoli cells and the heads of the elongated spermatids (es) at the adluminal compartment. C, two weeks after glycerol treatment, it is noted that all elongated spermatids were depleted, but the number of round spermatids (rs) and spermatocytes (p) remained relatively unchanged. Testin is still localized in the basal compartment between Sertoli and germ cells. Four (D) and eight (E) weeks after glycerol treatment, a massive reduction of germ cells from the epithelium and an accumulation of immunoreactive testin was found in the lumen of the tubule. stress and/or cellular death caused by these experimental manipulations.
The present study using germ cells or their conditioned medium cocultured with Sertoli cells failed to demonstrate a significant effect on Sertoli cell testin steady-state mRNA level. This result seemingly suggests that germ cells do not play a major role in regulating testin expression. However, one must note that these cocultures were terminated at 20 h, at the time when specialized Sertoli-germ cell junctions had not yet formed, since morphological analysis has shown that the establishment of specialized junctions between Sertoli and germ cells in vitro, such as desmosome-like adherens junctions, requires a culture period of 24 -48 h (22, 23). As such, it remains to be determined whether testin is involved in the formation of Sertoli-germ cell junctions. The fact that there is a surge in testin expression when the Sertoli-germ cell junctions are disrupted may suggest testin somehow participates in the destruc- tion of cell junctions by acting as a protease or that it protects the testis against tissue damage by acting as a protease inhibitor. Alternatively, the increase in testin expression is the result of cell junction disruption that is associated with other cellular events. However, testin was found to be neither a protease nor a protease inhibitor (6), making the former possibility unlikely. Although the glycerol or lonidamine (8)-induced Sertoli-germ cell junction damage can elicit an increase in testin accumulation in the testicular cytosol by up to 20 -30fold, most of the testin was accumulated in the lumen and was likely flushed out. Therefore, it is very unlikely that testin is being used for the reassembly of the damaged junction. The present study, however, illustrates that the testin steady-state mRNA level per pair of testes increases steadily during testicular maturation when the increase in organ weight and the increase in germ cell to Sertoli cell ratio are taken into consideration. These results suggest that the rapid assembly and disassembly of intercellular junctions at the onset of spermatogenesis is likely to be one of the most critical factors in regulating testin expression.
Studies on the binding of testin onto the Sertoli and germ cell membrane reveal that the dissociation of testin from the cell surface after its binding requires the use of a detergent illus-trating that other biochemical changes may take place when testin couples onto the binding protein complex. This result also supports the notion that testin may be a structural component between Sertoli and germ cells. The amount of testin binding to Sertoli cells is limited to only 0.1-0.5% that of total 125 I-testin in the incubation mixture, regardless of an increase in Sertoli cell number used for the in vitro assay. This result together with the detergent solubilization experiment strongly suggests that the interaction of testin with its binding protein complex is not a classical ligand-receptor interaction. The fact that a secretory protein can become tightly associated with the cell surface via a receptor-binding protein in a fashion dissimilar to a classical ligand-receptor interaction is not without precedence. Wnt, a growing class of multi-functional signaling molecules (glycoproteins with apparent M r between 34,000 and 42,000) involved in both tumorigenesis and patterning events during development and tissue differentiation by coordinating the organization of groups of cells in the developing vertebrate (49,50), are secretory proteins that can become tightly associated with the cell surface or ECM and originally thought to be via a nonreceptor-mediated mechanism (51)(52)(53). A recent study, however, has demonstrated that Dfz2, a 694-amino acid polypeptide, is the receptor protein of the signaling molecules of the Wnt gene family (54). The binding of the Wnt proteins onto the receptor, similar to testin, cannot be assessed by conventional receptor-ligand assays such as a Scatchard plot, probably due to their low abundancy. A comparison between testin and nine members of the Wnt family, including Wnt-1 and Wnt-2 (55), using DNASIS, PROSIS, and BESTFIT programs at the levels of amino acid and nucleotide sequences revealed that they share 10 -30% identity (data not shown). As such, testin is likely a distant member of the Wnt gene family, and it does share some of the unusual feature of the Wnt.
In the present and other earlier studies (3,8,10), it was shown that disruption of testicular cell junctions by glycerol, lonidamine, busulfan, or hypotonic treatment can lead to a surge in testin expression. Such a change in expression may be a response to cellular stress, when junctions are being disrupted. Studies in muscle, in particular, cardiac and skeletal myocytes, reveal that 5-30% of the cell population undergo plasma membrane disruption under physiological conditions, due to the contractile nature of these cells (56 -58). Likewise, about 3% and 6% of the epidermal and endothelial cells in the skin (56) and aorta (59), respectively, are being disrupted. Even though the level of cell wounding for the Sertoli and germ cell in the testis is not known, extensive plasma membrane disruption in testicular cells is expected to occur due to the rapid morphological changes during spermatogenesis. Until recently, membrane resealing was thought to be a passive event. The "wound hormone" hypothesis (60) suggests that chemical mediators of tissue restructuring such as growth factors, stored in cytosol, are released during membrane disruptions (for review, see Ref. 61). For instance, basic fibroblast growth factor (bFGF), which lacks a signal peptide sequence, is a potent growth-promoting factor when it is released extracellularly due to plasma membrane damage (62,63). Studies by immunoprecipitation has demonstrated the release of bFGF by germ cells (64), suggesting the bFGF found in GCCM is likely the result of germ cell wounding, since bFGF under the normal physiological condition is not a secretory protein because of the lack of a signal peptide. Other studies have demonstrated that the use of trypsin to isolate germ cells from the tubules (64) can alter many of the cell surface properties (20,22,65). As such, the secretion of bFGF by germ cells as demonstrated in this earlier study (64) may be the result of trypsin-induced plasma membrane damage that causes of release of bFGF. It is possible that bFGF is an important growth-promoting and signaling molecule that participates in repairing the disrupted testicular cell junctions during spermatogenesis and whose release from germ cells is the result of degeneration (66 -68) and apoptosis (69). Testin, on the other hand, is a secretory protein with a definite signal sequence (6). However, its protein level in the rete testis fluid and cytosols of the testis and epididymis as well as its expression are virtually undetectable except in the gonad. When Sertoli cells are being cultured in vitro, the amount of testin secreted into the spent medium is comparable with other Sertoli cell secretory products. It is our belief that upon dissociation of the testicular cells from the tubule during the preparation of Sertoli cells, testin expression is induced and remains elevated, as demonstrated in the present study. This observation thus supports the postulate that the expression of testin may be induced by cell wounding.
FIG. 12. The testin steady-state mRNA level in Sertoli cells isolated from rats at 20, 35, and 90 days of age, which illustrates an age-dependent reduction in expression. A, about 10 g of total RNA was used per lane, and the Northern blot was hybridized with a 289-base pair testin cDNA probe as described under "Materials and Methods." B, the same blot such as the one shown in A but stained with ethidium bromide, showing the integrity of the RNA; the same amount of RNA was used per each lane. C, densitometric scanning data of two blots such as the one shown in A at 600 nm but normalized against ␤-actin. D, day.