I. CHANGES IN SERTOLI-GERM CELL INTERACTIONS AND JUNCTION DYNAMICS DURING SPERMATOGENESIS: AN OVERVIEW

Top Previous Next References

In mammals, the functional unit of the testis is the seminiferous tubule. Each seminiferous tubule is about 1 m in length and 0.5 mm in diameter (for review, see Ref. 366). Figure 1 shows the cross-section of a typical seminiferous tubule from an adult rat testis. The close morphological association between Sertoli cells and germ cells at different stages of their development (such as spermatogonia, spermatocytes, round spermatids, and elongated spermatids) is clearly visible in the seminiferous epithelium (Fig. 1). As a result of such morphological intimacy between Sertoli and germ cells, it is conceivable that extensive interactions and communications take place between these cells throughout spermatogenesis both at the biochemical and molecular level. Indeed, morphometric analysis of the adult rat testis has shown that each Sertoli cell is associated with ~30-50 germ cells at each stage of the spermatogenic cycle in the epithelium (480, 497), illustrating not only that germ cell development relies heavily on the Sertoli cell but that extensive communications take place to coordinate the various events of spermatogenesis. Studies from the past two decades have indeed demonstrated that germ cells largely rely on Sertoli cells for structural and nutritional support (for reviews, see Refs. 105, 221, 379). For instance, in the rat, the entire process of germ cell development, except for the early phase of spermatogenesis from type B spermatogonia up to preleptotene and leptotene spermatocytes, is segregated from the systemic circulation because of the blood-testis barrier (BTB) created by tight junctions (TJ) between Sertoli cells near the basal lamina (128, 129, 131, 380, 400, 401). As such, germ cells and Sertoli cells develop an intimate and elaborate cellular network for cell-cell communications via paracrine factors and signaling molecules, so that Sertoli cells can provide developing germ cells with the needed nutrients and biological factors (for reviews, see Refs. 61, 62, 133, 147, 221, 416). Indeed, in vitro studies have shown that there is bidirectional trafficking between Sertoli and germ cells and that each cell type regulates the function of the other (20, 21, 221, 416).

View larger version (148K): [in this window] [in a new window]   Fig. 1. Cross-section of a seminiferous tubule from an adult Sprague-Dawley rat showing the organization of testicular cells and the intimate relationships between Sertoli and germ cells. S, Sertoli cell nucleus; SG, spermatogonium; PS, pachytene spermatocyte; RS, round spermatid; ES, elongated spermatid. [Adapted from Mruk and Cheng (308); courtesy of Dr. Li-Ji Zhu.]

Throughout spermatogenesis, different biochemical, cellular, and molecular events take place in the seminiferous epithelium leading to the formation of eight spermatids (haploid) from a single type B spermatogonium (diploid) (for reviews, see Refs. 117, 379). Furthermore, preleptotene and leptotene spermatocytes must migrate progressively from the basal to the adluminal compartment of the seminiferous epithelium traversing the BTB, while differentiating into haploid spermatids (Fig. 1). Without this timely movement of developing germ cells across the seminiferous epithelium, spermatogenesis cannot go to completion, and infertility will result. Moreover, this event of cell movement is accompanied by extensive restructuring of cell-cell actin-based adherens junctions (AJs) between Sertoli and germ cells, such as ectoplasmic specializations (ES) (for reviews, see Refs. 329, 330, 402). Although the subject of spermatogenesis, in particular its morphological changes and hormonal regulation, has been extensively studied (for reviews, see Refs. 117, 221, 378, 379, 404), the subject of cell junction restructuring pertinent to spermatogenesis from a biochemical and molecular standpoint has largely been neglected. In this review, we attempt to provide an updated review in this subject area. However, it must be noted that much of the information discussed herein is derived from investigations in other epithelia, but a significant amount of work has also been done in the testis investigating molecules pertinent to junction dynamics in the past decade. As such, every effort was made to refer to recent studies in the testis.

In mammals, spermatogenesis is composed of three distinct phases of cellular and molecular changes (for reviews, see Refs. 117, 379). 1) Mitosis is proliferation of type A spermatogonia, some of which will differentiate into type B spermatogonia (for reviews, see Refs. 123, 126). These in turn will differentiate into preleptotene and leptotene spermatocytes, which are the germ cells that will traverse the BTB entering into the adluminal compartment. This phase takes place in the basal lamina outside of the BTB (Fig. 2). 2) The meiotic phase is when primary spermatocytes divide and differentiate into secondary spermatocytes and haploid spermatids. This phase largely takes place behind the BTB in the adluminal compartment (Fig. 2). 3) Spermiogenesis is the morphogenesis of spermatids into spermatozoa, which is accompanied by extensive changes in the nucleus such as nuclear condensation (Fig. 2) (94, 105, 369). The fully developed spermatids (spermatozoa) will then leave the seminiferous epithelium via spermiation. In the mouse, the differentiation of haploid round spermatids into spermatozoa can be morphologically divided into 16 steps (379) (Fig. 2). During steps 1-7 of spermiogenesis, round spermatids develop acrosomes and flagellae (Fig. 2). At step 8, the heads of spermatids orientate toward the basal compartment of the epithelium. From steps 9 to 13, spermatids undergo a series of morphological changes, which include nucleus condensation and elongation of the flagella. At steps 14 and 15, alignment of mitochondria along the elongating flagella takes place. At step 16, spermatids are translocated to the adluminal surface of the epithelium to be released into the lumen at spermiation. Throughout these steps (Fig. 2), developing germ cells remain attached to the epithelium via a modified type of cell-cell anchoring junction with actin filament attachment sites (i.e., AJs) specific to the testis known as the ES (Fig. 2) (for reviews, see Refs. 343, 374, 380, 470).

View larger version (30K): [in this window] [in a new window]   Fig. 2. A schematic drawing that illustrates extensive changes in tight junction (TJ) and cell-cell actin-based adherens junction (AJ) dynamics during spermatogenesis and spermiogenesis in the mouse. This figure was prepared based on reviews and reports cited in sections III and V. Among the AJs in the testis, four functional complexes are known to exist to date, which include cadherin/catenin complex, nectin/afadin complex, tubulobulbar complex, and ES (see sect. V). The ES is composed of basal and apical ES constituted possibly by 64- and 61-integrins, respectively (310, 470); however, their binding partner(s), if any, is not known. It is possible that laminin 111-chains and 3-chains constitute the binding partners for the basal and apical integrins in the ES, respectively (244a) (see sect. VC1). While it is certain that 61-integrins are found between Sertoli cells and developing spermatids in the apical ES (470), it remains to be determined if 64-integrins can be found between Sertoli cells and developing spermatocytes and spermatogonia (type B) in the basal ES, or it is restricted only to the interface of Sertoli cells and the basement membrane. ES, ectoplasmic specialization, a modified testis-specific AJ.

In the seminiferous epithelium, the association between Sertoli and germ cells throughout these three phases of development as described above are arranged into defined stages. In the rat, these stages follow one another giving rise to the wave of the seminiferous epithelium along the seminiferous tubule (for review, see Ref. 339). Using periodic acid-Schiff (PAS) staining to visualize changes in the shape of the nucleus and acrosome of germ cells during their development in the seminiferous epithelium, the spermatogenic cycle is divided into 14 stages in the rat (258) (Fig. 3) and 12 stages in the mouse (379). Interestingly, different stages of the cycle in the seminiferous epithelium can be readily visualized by the transillumination pattern of freshly isolated rat seminiferous tubules by stereomicroscopy largely because of the changes in nuclear condensation of the sperm head (for review, see Ref. 339) (Fig. 3). In the rat, one spermatogenic cycle takes ~12-14 days to complete (130, 258) and ~8-9 days in the mouse (379). Studies using [³H]thymidine, however, have shown that it takes ~54 and ~35 days for a single spermatogonium to complete spermatogenesis and give rise to eight haploid spermatids in the rat (105, 117, 379) and the mouse (379), respectively. This is because developing germ cells must go through the cycle 4.5 times before they can become fully developed spermatids (spermatozoa) that are released into the tubular lumen at stage VIII of the cycle (403). Given this extensive cell movement across the epithelium, it is conceivable that there is extensive restructuring at the interface of Sertoli-germ cells at the adherens junction level. Furthermore, the biochemical, molecular, and cellular events pertinent to spermatogenesis are under endocrine control of the hypothalamic-pituitary-testicular axis (for reviews, see Refs. 146, 339, 403, 428). In this review, we limit our discussion on the biology and regulation of Sertoli-germ cell interactions to the level of cell junctions, in particular, occluding and anchoring junctions, and how current advances in these two areas have widened the possibilities of developing innovative male contraceptives.

View larger version (111K): [in this window] [in a new window]   Fig. 3. Shown are stages I-XIV of the cycle of the seminiferous epithelium in the rat as determined by transillumination pattern of the freshly isolated seminiferous tubules from adult rats (A) and cross-sections of the tubules in the testis that show the unique association between Sertoli cells and developing germ cells in each stage of the cycle (B-J). A: this is the schematic drawing of the different zones of a seminiferous tubule from adult rats under a transillumination stereomicrocope, which can be divided into different zones, namely, pale, weak spot, strong spot, and dark zone, representing different stages of the spermatogenic cycle. The different zoning pattern is the result of different numbers of condensed nuclei in developing germ cells associated with Sertoli cells at different stages. [Modified from Parvinen (339).] B-J: these are micrographs of cross-sections, ~8 µm, of frozen testes from adult rats stained with hematoxylin corresponding to seminiferous tubules at stage I (B), II-IV (C), V (D), VI-VII (E), VIII (F), IX (G), X-XI (H), XII-XIII (I), and XIV (J). They illustrate the unique pattern of association of developing germ cells, such as spermatocytes, spermatids, and elongated spermatids, with Sertoli cells in the seminiferous epithelium that gives rise to the different zones under transillumination microscopy shown in A, representing different stages of the spermatogenic cycle from I to XIV. Frozen sections were fixed in modified Bouin's fixative and stained with hematoxylin using techniques as previously described (171, 512, 513, 516, 517). Bar = 20 µm.

	II. JUNCTION DISASSEMBLY AND REASSEMBLY ARE PERTINENT TO GERM CELL MOVEMENT DURING SPERMATOGENESIS

Top Previous Next References

As described in section I, extensive interactions between Sertoli and germ cells take place in the seminiferous epithelium to coordinate the intermittent events of disassembly and reassembly of Sertoli cell AJs and TJs and Sertoli-germ cell AJs to facilitate the movement of germ cells across the epithelium (see Figs. 1-3). In this section, as well as in sections III and V, we have reviewed the current status of research in the dynamics of cell junctions in the testis. Lastly, we have introduced two new innovative approaches for male contraception that are based on current research by perturbing the functions of TJs (see sect. IV) and AJs (see sect. VI) in the testis.

A.  Cell Junctions in the Testis

1.  Introduction

Morphological studies of the testis performed in the 1970s and 1980s have identified many junction types common to other epithelia (Table 1) (61, 62, 117, 221, 342, 377, 380). The three types of junctions found in the testis are as follows: occluding, anchoring, and communicating gap junctions (GJ) (see Table 1). Other structural modifications of AJs that use actin filaments as their attachment sites (see Table 1), such as ES and tubulobulbar complexes, are unique to the testis (Table 1). In general, the relative locations of these junctions in different epithelia are the same; TJs occupy the most apical portion of cells, followed by AJs, and then by a parallel row of desmosomes. Altogether these structures form the "junctional complex." On the other hand, GJs and additional desmosomes are not integrated tightly with TJs and AJs and can be scattered around the epithelium, which in turn anchor cells onto the extracellular matrix (ECM). As such, TJs are furthest away from the ECM (for review, see Ref. 4). In the testis, however, the location of TJs relative to anchoring junctions and GJs is different from other epithelia since Sertoli cell TJs that create the BTB are found at the basal compartment of the seminiferous epithelium adjacent to the basal lamina closest to the basement membrane (see Table 1 and Figs. 1-3), which is a specialized form of ECM in the testis (for review, see Ref. 127). The classification of the three junction types in mammalian tissues and their constituent proteins are summarized in Table 1. The only known occluding and communicating junction types are the TJ and GJ, respectively (Table 1). However, there is more than one type of anchoring junction, which are classified as follows (see Table 1). In this review, we limit our discussion on TJs and anchoring junctions (in particular the cell-cell actin-based adherens junctions, AJs), since a sufficient body of work has been conducted in these areas, forming the basis for future studies to understand the biology of junction dynamics in spermatogenesis and for contraceptive development. An excellent review on GJs in the testis can be found elsewhere (see Ref. 133).

Anchoring junctions are abundant in many tissues, in particular those subjected to mechanical stress (for review, see Ref. 4). Anchoring junctions physically connect cytoskeletal elements of one cell to a neighboring cell or to the ECM. Altogether, there are four types of anchoring junctions (for review, see Ref. 4): 1) adhesion belts or adherens junctions (AJs), 2) focal contacts, 3) desmosomes, and 4) hemidesmosomes (for review, see Ref. 4). Anchoring junctions that connect two cells together are known as adhesion belts (or adherens junctions, AJs) and desmosomes; those that connect cells to the extracellular matrix are focal contacts and hemidesmosomes (see Table 1). AJs and focal contacts utilize actin filaments as attachment sites, whereas desmosomes and hemidesmosomes use intermediate filaments as attachment sites (Table 1). AJs, zonula adherens or adhesion belts, are located in the apical domains of cells below TJs and are constituted largely by Ca²⁺-dependent cell adhesion molecules (CAMs), also known as cadherins (for reviews, see Refs. 4, 177, 443). Cadherins in turn attach to intracellular attachment proteins, such as - and -catenin. This cadherin/catenin complex associates with the underlying actin filament bundles via its interactions with vinculin and -actinin (for reviews, see Refs. 177, 443). It must be noted that in the testis, AJs between Sertoli and germ cells are morphologically different from those in other epithelia, and the best-described modified testis-specific AJ structures are ES and tubulobulbar complexes (62, 378, 470, 471). Herein, we focus our review largely on the cell-cell actin-based AJs. Because several excellent reviews on desmosomes, focal contacts, and hemidesmosomes in other epithelia can be found elsewhere (for reviews, see Refs. 4, 9, 45, 52-56, 229, 246, 247, 396, 397), we will not elaborate on these junction types in this review.

Although germ cell movement across the seminiferous epithelium is one of the most important and interesting biological phenomena during spermatogenesis, very few studies have been performed to examine the participating molecules and the mechanism by which this event is regulated. Our belief is that these studies are significant because not only can they expand our knowledge of spermatogenesis pertinent to junction restructuring, but a thorough understanding of the biology of germ cell movement should lead to the development of novel and safer male contraceptives. We have hypothesized that if germ cells are induced to translocate across the epithelium rapidly, even before they complete their development, germ cells found in the seminiferous tubular lumen will be immature and lack the ability to fertilize the ovum (308). This can also be achieved by perturbing cell adhesion in the testis prompting the depletion of germ cells from the seminiferous epithelium. Alternatively, if germ cell movement is hampered and germ cells are retained in the epithelium for a prolonged period of time, they will become "aged" and be removed by Sertoli cells via phagocytosis (308). In both instances, infertility will result. A disruption of fertility by this approach is likely to induce minimal side effects since the hypothalamus-pituitary-testicular axis is not disrupted. However, a model to study the events of germ cell movement is lacking. If such a model were available, it could be used to study the cascade of events leading to germ cell movement. This, in turn, could be used to identify target genes and/or proteins that perturb cell movement and/or cell adhesion. Once the technique to obtain staged tubules for in vitro studies became available (339), studies were performed to identify target genes that associated with germ cell movement, such as spermiation (61, 62, 147). For instance, it was shown that the expression and/or accumulation of proteases, such as cathepsin L (a cysteine protease), was highest in stages V-VII of the cycle preceding spermiation (499, 508), which occurs in stage VIII, illustrating its possible involvement in disrupting elongated spermatid-Sertoli cell junctions at spermiation. Subsequent studies by in situ hybridization have also confirmed the predominant but transient expression of cathepsin L at stages V-VII (86, 345). Interestingly, the expression of cathepsins D (an aspartic protease) and S (also a cysteine protease), in contrast to cathepsin L, was shown to be predominant not only at stage VII preceding spermiation but also high at stages VIII-XII (86). These results thus challenge the notion that cathepsin L is involved in spermiation. Needless to say, there are many biochemical and molecular events occurring at stages V-VII, some of which are for purposes entirely unrelated to the events of spermiation or cell movement. As such, it is crucial that in vitro functional studies be performed first to determine which target genes and/or proteins are involved in the dynamics of TJs or AJs. This should be followed by in vivo studies to examine if these changes indeed occur in the seminiferous epithelium during junction restructuring, such as analyzing junction-associated proteins in cross-sections of staged tubules in the testis during the spermatogenic cycle. This can be followed by additional studies to investigate if a specific cellular event can be perturbed by manipulating the function of a molecule or a group of molecules to assess whether this can disrupt spermatogenesis. In this context, it is noteworthy to mention that the expression of a target gene, such as occludin, that is not stage specific in the testis (110) does not necessarily preclude its involvement in junction dynamics. Indeed, occludin is an important regulator of TJ dynamics (138, 298). Thus a molecule that is crucial to junction restructuring may not necessarily be stage specific unless it is linked to a specific event, such as spermiation.

Sertoli cells cultured in vitro in chemically defined serum-free medium have been used to study the biology and regulation of the TJ-permeability barrier for almost two decades (60, 167, 218-220, 329). This model was used in our laboratory to identify several target genes that are implicated in the regulation of Sertoli cell TJ dynamics, such as occludin, claudin-11, ZO-1, ₂-macroglobulin, and others (82-86, 169, 263, 264, 270, 271, 389, 495). Furthermore, it was shown that transforming growth factor (TGF)-3 perturbed the Sertoli cell TJ-permeability barrier in vitro via its effects on the expression of occludin, ZO-1, and claudin-11 (270), utilizing the p38-mitogen-activated protein (MAP) kinase signaling pathway (271). The novelty of this model to study Sertoli TJ dynamics is as follows. First, Sertoli cells are cultured in chemically defined serum-free medium; as such, it can be used to identify molecules that play a crucial physiological role in regulating TJ dynamics. Indeed, with the use of this model, studies from various laboratories have shown that testosterone, cAMP (82, 218, 219), and protease inhibitors (329) are important regulators of Sertoli cell TJ dynamics. Second, this model coupled with the use of other techniques, such as restricted diffusion of FITC-dextran, [³H]inulin, ¹²⁵I-BSA; polarized secretion of Sertoli cell proteins, such as ₂-macroglobulin, testin, rat androgen binding protein, and clusterin; and the measurement of transepithelial electrical resistance (TER) across the Sertoli cell epithelium, can quantitatively assess the dynamics of the TJ barrier (82, 83, 166, 167, 270, 271). Indeed, this model was used in conjunction with CdCl₂ to study the Sertoli cell TJ dynamics (82, 219). Preliminary studies have illustrated that this is a reliable in vitro model to delineate the possible cascade of events pertinent to TJ disassembly and reassembly (82).

Three theories are found in the literature explaining the events of Sertoli cell TJ disassembly/reassembly during spermatogenesis. First, the "zipper theory," which proposes that inter-Sertoli TJs or occluding zonules, consisting of fibrils that completely encircle the basal domains of Sertoli cells, break down to accommodate the passage of preleptotene or leptotene spermatocytes while new occluding zonules reform under the migrating preleptotene spermatocytes (142, 342-344). However, there are no in vivo studies showing leakage of tracers into the tubular lumen even for a short period of time. Second, the "intermediate cellular compartment theory" proposed by Russell (375a) suggests the presence of a compartment occupied by germ cells in transit from the basal to the adluminal compartment. However, subsequent morphological studies have shown that there is only one occluding zonule per Sertoli cell at any one time, making it unlikely that such a compartment exists, because if it does, one should be able to visualize a preleptotene spermatocyte trapped in between two occluding zonules (343, 344, 380). Also, both theories do not explain what triggers the dissociation and association of TJ fibrils to facilitate the movement of preleptotene and leptotene spermatocytes across the BTB. For instance, if the disrupted fibrils, which are composed of TJ-associated proteins, such as occludins and claudins, can be repaired rapidly, a sophisticated signaling/trafficking system between Sertoli and germ cells must exist. How can this be achieved and regulated? Because germ cells do not possess an extensive cytoskeleton network, their movement is likely dependent on Sertoli cells. What is the mechanism(s) that signals Sertoli cells to facilitate germ cell movement? Are both Sertoli and germ cells equipped with the necessary signaling molecules to regulate their interactions that triggers cell movement? As such, there are many open questions that remain to be addressed. Third, the "stress theory" or "repetitive removal of membrane segments theory" proposed by Pelletier and Byers (342, 343) suggests that the continuous upward migration of a large number of germ cells creates a stress against the Sertoli cell occluding zonule. This may result in junction proliferation, changes in orientation, and disintegration in the fibrils composing the occluding zonule (344, 342). This theory, however, does not explain what triggers and facilitates the upward movement of germ cells. Also, how can stress induce changes in junction orientation and breakage in the TJ fibrils?

	III. OCCLUDING JUNCTIONS: TIGHT JUNCTIONS AND THE BLOOD-TESTIS BARRIER

Top Previous Next References

The only known occluding junction type in mammalian tissues, including the testis, is the TJ (zonulae occludens) (see Figs. 2 and 4). Three functions can be ascribed to TJs. First, TJs formed between adjacent cells play an essential role in compartmentalization by creating a barrier to restrict the diffusion of solutes through the paracellular pathway (297, 432). Second, TJs also create a boundary between the apical and basolateral domains of a cell, which differ in protein and lipid composition. This in turn creates and maintains epithelial and endothelial cell polarity (367). These two roles of TJs are known as the "barrier" and "fence" functions, respectively. Third, the BTB also creates an immunological barrier that sequesters antigenic determinants residing on germ cell surfaces from the systemic circulation. This barrier also excludes the entry of circulating immunoglobulins and lymphocytes into the adluminal compartment (for reviews, see Refs. 131, 400, 401).

View larger version (29K): [in this window] [in a new window]   Fig. 4. Current model on the molecular structure of tight junctions and the constituent proteins in the testis. A schematic drawing that illustrates the molecular architecture of TJs is shown. Only three classes of TJ-integral membrane proteins, namely, occludin, claudin, and JAM, are known to date. Each TJ-integral membrane protein has distinctive transmembrane, extracellular, and intracellular domains. The COOH terminus of occludin is associated with ZO-1 intracellularly, which in turn associates with ZO-2, ZO-3, AF-6, ZAK, and others. Cytosolic proteins enclosed in dashed boxes are TJ-associated signal transducers whose functions are not yet known. This diagram was prepared based on earlier published models (for reviews, see Refs. 88, 138, 289, 298, 368, 431, 432, 458, 459, 472, 482).

In the testis, TJs are different from those found in other epithelia in several ways. First, testicular TJs are not assembled until puberty, which is ~15 days after birth in the rat (117, 131, 379, 400). Second, the Sertoli cell-TJ complex in the testis has a unique architecture and location within the seminiferous epithelium, which is formed by the close apposition of adjacent Sertoli cell membranes at the basal compartment of the epithelium and designated the BTB (141, 401). Furthermore, the BTB is closest to the basal lamina in the testis. This is in sharp contrast to TJs found in other epithelia, which are furthest away from the basement membrane. The BTB in the testis also creates a specialized microenvironment for germ cell development because developing germ cells do not have access to the systemic circulation. As such, they must rely on Sertoli cells for the provision of nutrients and biological factors for their maturation. Third, the BTB is dynamic in nature since it must open (disassemble) and close (reassemble) periodically to allow preleptotene and leptotene spermatocytes to gain entry into the adluminal compartment for further development.

Morphologically, TJs form a continuous circumferential seal near the apex of both epithelial and endothelial cells, farthest from the basal lamina (4, 139), whereas in the testis, they are located in the basal compartment of the seminiferous epithelium adjacent to the basement membrane, which is a modified form of ECM in the testis (127) (Figs. 1 and 2). Thus the relative location of TJs, AJs, and ECM in the testis is in reverse order compared with other epithelia. There are three classes of TJ-integral membrane proteins: occludins, claudins, and junctional adhesion molecules (JAM, a member of the immunoglobulin superfamily) (138, 285, 298).

A) OCCLUDIN FAMILY. Occludin is a 60- to 65-kDa single polypeptide and a Ca²⁺-independent intercellular adhesion molecule (467). It is also a putative TJ-integral membrane protein that contributes to the barrier and fence functions of TJs (for reviews, see Refs. 68, 138, 298, 458, 459). Recently, a new variant of occludin designated occludin 1B has been identified in the mouse testis, which has an additional 193-bp insertion and a unique NH₂-terminal sequence compared with occludin (313). Each occludin molecule consists of four transmembrane domains, two extracellular loops (loop 1, amino acid residues 90-138; and loop 2, residues 199-243 from the NH₂ terminus in the rat), one intracellular loop, a small NH₂-terminal cytosolic domain, and a large COOH-terminal cytosolic domain (Fig. 4) (for reviews, see Refs. 88, 138). Among these domains, the first extracellular domain is rich in Tyr and Gly (~60%), and these structural characteristics are well conserved among different mammalian species (12). Its cytoplasmic COOH-terminal domain associates with ZO-1 at a stoichiometric ratio of 1:1 (153). Overexpression of chicken occludin in Madin-Darby canine kidney (MDCK) cells increased the "tightness" of the TJ barrier as manifested by an increase in transepithelial electrical resistance across the cell epithelium, whereas truncation of occludin at the COOH terminus caused the influx of N-(4,4-difluoro-5,7-dimethyl-4-bora-3,4-diaza-S-indacene-3-pentanoyl)sphingosyl phosphocholine (BODIPY)- sphingomyelin across the TJ barrier in vitro (28). Furthermore, introduction of a peptide corresponding to the second extracellular domain of occludin can perturb the TJ-permeability barrier of Xenopus A6 epithelial cells in vitro (496) and Sertoli cells in vitro (83) and can disrupt the BTB in vivo (83). These results thus suggest the notion that the second extracellular loop of occludin is important to confer TJ functionality. However, occludin alone cannot generate a bona fide TJ barrier. For instance, occludin-deficient embryonic stem cells can still differentiate into polarized epithelial cells having TJs (384). Instead, studies have shown that occludins and claudins both contribute to maintain TJ function in epithelial cells (for review, see Ref. 459). Other studies have shown that the first extracellular loop of occludin contributes significantly to cell adhesive function (138, 298, 467). Studies by immuno-freeze fracture electron microscopy have also shown that occludin is concentrated within TJ fibrils (148) and is phosphorylated both at Ser/Thr and Tyr residues (386, 457). However, an additional minor pool of occludin is found in the basolateral membrane, which is less phosphorylated, and is not assembled into TJ fibrils (100, 386). This lateral pool of occludin is likely to serve as a reservoir of molecules for rapid expansion of TJs. For instance, earlier studies predating the discovery of occludin have shown that application of trypsin to the basolateral surface of MDCK cells could induce a rapid increase in the number of TJ fibrils and transepithelial electrical resistance (TER) (275), illustrating that mammalian cells possess a mechanism to rapidly assemble TJs in response to an external stimulus. With the use of peptide mass fingerprinting analysis coupled with electrospray ionization tandem mass spectroscopy, it has been shown that Ser-338 of occludin is the putative phosphorylation site induced by protein kinase C (PKC) in MDCK cells in vitro (13). Taking these observations collectively, it is possible that preleptotene and leptotene spermatocytes are the source of the stimulus that regulates TJ dynamics by inducing localized changes in the phosphorylation of occludin in the TJ fibrils. Before this hypothesis can be tested, one must first assess whether there are changes in occludin phosphorylation at the time of junction assembly both in vitro and in vivo. Other studies have shown that two adjacent occludin molecules are capable of lateral oligomerization, which in turn form interlocking TJ fibrils, perhaps within the membrane bilayer (72). However, claudin-1, a structural molecule that constitutes TJ fibrils by forming paired TJ strands in the apposing membrane of adjacent cells with claudin-2 or -3 (459), is largely unphosphorylated (71), seemingly suggesting that TJ dynamics are regulated by multiple signaling pathways, some of which are not regulated by protein phosphorylation. Also, different TJ-integral membrane proteins may be regulated by distinctly different signaling pathways. It has been reported that occludin and occludin-1B are found in the rat and mouse testis but not human and guinea pig Sertoli cells (110, 303), suggesting that the testis is equipped with other yet to be identified TJ-integral membrane proteins to maintain the BTB integrity and its function. With the use of immunohistochemistry, it was shown that the accumulation of occludin closely associated with the assembly of the BTB at the site of TJs in the mouse testis during postnatal development; however, its pattern of localization in the seminiferous epithelium was not stage specific (110). Between postnatal day 23 and adulthood, occludin was diffusely localized in the seminiferous epithelium (110). Furthermore, intratesticular injection of glycerol in adult rats that damaged the BTB by disrupting the organization of microfilaments and microtubules in Sertoli cells was also shown to disrupt the cellular occludin distribution (490). Taken collectively, these results illustrate that much work remains to be done to define the role of occludin in regulating TJ dynamics in the testis. A more recent study has demonstrated that expression of occludin mutants in transfected epithelial cells with a modified NH₂-terminal cytoplasmic domain (but not the COOH-terminal cytoplasmic domain since its deletion failed to perturb cell transepithelial migration) upregulated the migration of neutrophils across the TJ barrier without affecting TER and paracellular permeability (201). This report clearly illustrates that occludin regulates the transepithelial migration of cells across the TJ barrier utilizing its NH₂-terminal cytoplasmic domain. In transgenic occludin^/ mice, TJs in epithelia that were examined did not appear to be morphologically affected; for instance, the TJ barrier function of the intestinal epithelium remained intact (385). However, other phenotypic changes were detected in occludin^/ mice, such as testicular atrophy, calcification of the brain, and thinning of the bone (385). Furthermore, although the testes and seminiferous tubules of occludin^/ mice in their early postnatal stage appeared to be normal without any detectable damage to spermatogenesis, the tubules displayed atrophy and the seminiferous epithelium was devoid of germ cells in adulthood (384). These findings thus suggest that the functions of occludin and TJs may be more complicated than originally conceived. Ironically, this is not entirely unexpected since TJs are implicated as the platform for signal transduction (for review, see Ref. 510).

B) CLAUDIN FAMILY. Claudins are another family of TJ-integral membrane proteins with an apparent molecular mass of ~22 kDa that confer TJ functionality and cell adhesiveness to epithelial cells (151, 153a, 154, 458, 459). Claudins share a similar molecular topology with occludin. Each claudin molecule consists of one short NH₂-terminal cytoplasmic domain, two extracellular loops, four putative transmembrane domains, and a long COOH-terminal cytoplasmic domain (Fig. 4). Nevertheless, the primary amino acid sequences of claudins do not share any significant homologies with occludins, illustrating that these two classes of TJ proteins are distinct. At least 24 claudins have been identified in TJs in different epithelia (for reviews, see Refs. 138, 151, 154, 155, 298, 301, 302, 459). With the use of immunogold freeze-fracture labeling, claudins have been shown to localize within TJ fibrils (310, 302). Recent studies have shown that claudins are the principal TJ-component proteins primarily responsible for constructing the seal-forming elements in TJs (for reviews, see Refs. 458, 459). In contrast to occludin, claudins, such as claudin-1, found in TJ fibrils are largely unphosphorylated at the site of TJs in MDCK cells (71), illustrating that protein phosphorylation, while it plays an important role in TJ assembly, is possibly not the only determining factor that regulates TJ assembly. Seven different claudins (i.e., claudin-1, -3, -4, -5, -7, -8, and -11) are found in the testis. For instance, claudin-11 is found exclusively in the brain [also known as oligodendrocyte-specific protein (OSP) restricted to the myelin sheaths of oligodendrocytes in the central nervous system], the testis (restricted to Sertoli cells) (301, 302), the choroid plexus, and the collecting ducts in the kidney (164). The expression of claudin-11 in the testis appears to be limited to the Sertoli cell, since it was not found in germ cells (190). Moreover, the expression of claudin-11 by mouse Sertoli cells in vitro is inhibited by follicle stimulating hormone (FSH) and tumor necrosis factor (TNF)- with an ED₅₀ at 4 and 4.5 ng/ml, respectively (190). Its expression in the mouse testis in vivo was induced at postnatal days 6-16, coinciding with the assembly of the BTB that occurs at ~10-16 days postnatally (190). These results suggest that the intricate interactions between claudin-11, FSH, and cytokines play a crucial role in the assembly of the BTB and the regulation of TJ dynamics in the testis. Furthermore, gene knock-out experiments showed that null mice lacking claudin-11 are sterile and have hindlimb weakness and retardation of nerve conduction (164). TJ intramembranous strands were absent in the central nervous system myelin and between Sertoli cells in the testis in claudin-11 knock-out mice (164). These results thus illustrate the significance of claudin-11 in spermatogenesis. A recent report has also demonstrated the significance of claudin-1 in TJ function since claudin-1^/ mice died within 1 day after birth with wrinkled skin (151a). Although both occludin and claudin-4 were found in the TJs in these claudin-1^/ mice, the epidermal TJ barrier could not form (151a), seemingly suggesting that claudin-1 is primarily responsible for the integrity of the epidermal TJ barrier.

Different members of the claudin family are expressed in different tissues and/or organs. For instance, claudin-2 is predominantly expressed in the kidney and liver, whereas claudin-5 can be found in virtually all tissues examined to date (151, 154, 298). Claudins-4, -7, and -8 are present in both lung and kidney (150, 301, 302). Recently, a new member of the claudin family designated paracellin-1/claudin-16 was identified. Its mutation results in renal hypomagnesemia in humans (413). Two different claudins can also be coexpressed in a single cell, raising the possibility that heterogeneous claudins can form heteromeric TJ strands (155). Indeed, recent studies have demonstrated that claudins in one cell can associate heterotypically and homotypically with claudins in the apposing cell (155). When occludin was transfected into claudin-expressing fibroblasts, it was recruited to claudin fibrils, suggesting that claudins are the major contributor of fibril formation. Furthermore, claudins appear to confer stronger cell adhesion between cells than occludin (244), indicating they form transcellular contacts required to seal the intercellular space.

C) OCCLUDIN AND CLAUDINS: PARTNERS IN CREATING SEALED TJS. Claudins and occludin can be organized into oligomers and hetero-oligomers. Freeze-fracture analysis has shown that TJ fibrils are composed of particles of ~10 nm in diameter, which is similar in size to the connexons found in GJs (148). Interestingly, the folding topology of connexons is similar to claudins and occludin, and these molecules have similar M_r. By analogy, it is likely that the ~10 nm TJ particles are composed of more than one claudin or occludin molecule. For instance, each connexon in the GJ is composed of a ring of six identical protein subunits called the connexin at the periphery with a central pore, which permits the passage of chemical signals between two cells (for reviews, see Refs. 133, 250, 340). When claudin and occludin are coexpressed in fibroblasts, claudin recruits occludin to fibrils (154), suggesting they can form a functional protein complex. With the use of claudin^/ L cells transfected with claudins, it was found that claudin-1 and -3 and claudin-2 and -3 can form a TJ complex but claudin-1 and -2 cannot (for review, see Ref. 459). Thus, among the 24 known claudins, only some claudins can interact with selected members. Indeed, it has recently been proposed that the homotypic or heterotypic claudin-claudin complexes formed between apposing cells are composed of aqueous TJ pores to permit transport of small molecules across the TJ barrier (for review, see Ref. 459). Still, the detailed molecular architecture that forms these TJ pores in the paracellular space and their regulation is unclear. Additional mutational and structural studies are likely to resolve the issue of how these proteins interact with their partner molecules in apposing cells. It has been reported that the number of TJ strands correlate with the tightness of the TJ barrier (93, 278, 350). It is also known that the Sertoli cell TJ is one of the tightness in the mammalian body (128, 131, 159), suggesting that the number of junctional strands in Sertoli cell TJs is relatively high (343). However, the Sertoli cell TJ-permeability barrier in vitro was shown to have an electrical resistance of ~100 ·cm² (28, 82, 169, 218, 219, 270, 496), which is only about one-tenth of that observed in MDCK cells and keratinocytes in vitro (28, 179, 496). These results seemingly suggest that the tightness of the BTB is approximately one-tenth of that of TJs found in the kidney and other epithelia. Janecki et al. (219) first reported that the tightness of the Sertoli cell TJ-permeability barrier in vitro can be significantly increased in the presence of testosterone. Furthermore, a recent report (82) has shown that testosterone at 1 x 10⁷ M, which is ~100 times higher than the level of androgen found in the systemic circulation, but similar to the level detected in the testis, such as in the rete testis fluid (462), can protect the Sertoli TJ-permeability barrier in vitro from the disruptive effects of CdCl₂ (82, 219). More important, it was shown that testosterone could stimulate the expression of occludin by Sertoli cells in these same cultures (82). These results further demonstrate that the dynamics of TJs in the testis are regulated by androgens. In addition, proteases and protease inhibitors also appear to play a role in the BTB function, since chloroquine, a protease inhibitor, has been shown to facilitate the assembly and maintenance of Sertoli TJ in vitro (329), consistent with another study reporting that the assembly of junctions between testicular cells in vitro is regulated by both proteases and protease inhibitors (309).

D) JAM. JAM is the third type of TJ-integral membrane protein identified in TJs by using monoclonal antibodies against endothelial cells (285) and is localized to the intercellular boundaries of both epithelial and endothelial cells (334). To date, at least three JAM molecules are known, designated JAM-1, JAM-2, and JAM-3 (22, 23, 108, 334, 491). Although it is not known if JAM-3 is present in the testis, JAM-1 and JAM-2 have recently been identified in the mouse testis by Northern blot analysis (23). The molecular mass of JAM found in different tissues ranges from 36 to 41 kDa, possibly the result of carbohydrate heterogeneity. JAM-2 binds to JAM-3 (22). Unlike occludins and claudins, JAM exhibits a distinct topographic structure (Fig. 4). Each JAM molecule consists of a putative intracellular domain, a transmembrane domain, and an extracellular domain. The extracellular domain is composed of two V-shaped immunoglobulin-like loops with two interchained disulfide bonds (285) (Fig. 4). JAM has also been reported to facilitate homotypic cell-cell adhesion (285). When JAM was transfected into Chinese hamster ovary (CHO) cells, cell permeability to dextran (molecular mass ~38.9 kDa) was reduced by 50%, increasing the tightness of the TJ barrier (285), which is similar to results obtained when cadherin was transfected into CHO cells. However, it is not known if JAM is indeed a component of TJ fibrils and if it can form a functional TJ barrier. The use of anti-JAM monoclonal antibody, however, can prohibit the migration of monocytes through endothelial cells when chemotaxis assays were conducted both in vitro and in vivo (118). Also, administration of anti-JAM monoclonal antibody inhibited leukocyte accumulation in the cerebrospinal fluid and infiltration in the brain parenchyma. Furthermore, the tightness of the blood-brain permeability barrier was reduced in the presence of anti-JAM monoclonal antibody. On the basis of these observations, it has been concluded that JAM is a new target in limiting the inflammatory response accompanying meningitis (118). On the other hand, combined treatment of human endothelial cells with TNF- and interferon (IFN)- induced the redistribution but not the amount of JAM promoting transmigration of leukocytes across endothelial cells (331). It has been suggested that this redistribution of JAM is crucial for the transendothelial migration of leukocytes at inflammatory sites in response to proinflammatory cytokines (331). Another striking feature of JAM is that its mRNA is found in megakaryocytes, which do not possess TJs; also, the expression of JAM is either absent or at a very low level in hepatocytes, which contain well-developed TJs (285), suggesting JAM may have other yet-to-be identified physiological functions. A recent study has shown that JAM associates with ZO-1 when assessed by in vitro binding and coprecipitation experiments. More recent studies using immunofluorescent microscopy have shown that the COOH terminus of JAM binds to the PDZ3 domain of ZO-1 (214). JAM also coprecipitates with cingulin, another cytoplasmic TJ protein, through its interaction with the NH₂ terminus of cingulin (31). Although the role of JAM in tissues without TJs is still unclear, studies using immunoreplica electron microscopy have shown that JAM has an intimate spatial relationship with TJ strands in epithelial cells, such as fibroblasts (214). These results seemingly suggest that aggregates of JAM are tethered to claudin-based TJ strands through ZO-1, contributing to the overall architectural integrity of TJs in epithelia (214). Results of these studies clearly illustrate that JAM is an important TJ candidate protein. However, it remains to be determined if JAM-1 and JAM-2, which are found in the testis (23), can indeed facilitate the transmigration of preleptotene and leptotene spermatocytes across the BTB in the testis. In light of their unusual properties in regulating transendothelial migration of neutrophils and monocytes, this class of molecules should be aggressively studied in the testis to understand their role in TJ dynamics during spermatogenesis.

A) ZO-1 AND OTHERS. ZO-1 is the first TJ-associated cytoplasmic protein subjected to extensive investigation (433). ZO-1, ZO-2, and ZO-3 are members of the membrane-associated guanylate kinase (MAGUK) protein family of signal transducers. It is characterized by a PSD-95-Discs Large-zona occludens-1 (PDZ) domain, a Src homology 3 (SH3) domain, and a guanylate kinase (GUK) homology region (for reviews, see Refs. 297, 432). ZO-1, a 225-kDa polypeptide, was first identified in the membrane fraction of hepatocytes. It was also localized to Sertoli cell junctional complexes (59, 433). In the mature testis, ZO-1 is associated with Sertoli cell occluding and nonoccluding junctional membrane specializations, such as the subsurface filamentous specializations facing germ cells (59), suggesting that this protein is not restricted to TJs. In two MDCK cell strains, which exhibit high and low electrical resistance, the level of ZO-1 is similar, indicating the abundance of ZO-1 in a given tissue does not correlate with the tightness of TJs (433). ZO-1 is an asymmetric phosphoprotein that is peripherally associated with the cytoplasmic surface of the plasma membrane (10) at the sites of TJs (433). ZO-1 is also implicated in creating a mechanochemical linkage between the submembrane cytoskeleton and integral membrane components of the TJ instead of contributing to the TJ sealing properties (for review, see Ref. 432). Recent studies have shown that the PDZ domain of ZO-1 is the binding site for the COOH termini of claudins and occludins (153, 213), indicating ZO-1 is a linker protein coupling the transmembrane TJ proteins to the cytoskeleton. ZO-1 coexists with occludin in a stoichiometric ratio of 1:1 (155) (Fig. 4). The expression of ZO-1 becomes progressively restricted to the developing junction in the testis at the site of TJs between days 7 and 14 in the rat and mouse (59), correlating with the assembly of the BTB that takes places at ~13-15 days of age in rats. In addition, two ZO-1 isoforms are known: the ⁺ and forms, which are predominantly expressed at puberty and adulthood in the guinea pig testis, respectively, implicating the significance of ⁺ in the BTB assembly at puberty (344). Under conditions designed to preserve protein-protein interactions, ZO-2 (160 kDa) and ZO-3 (130 kDa) were shown to coimmunoprecipitate with ZO-1 (178, 188, 223). Both ZO-2 and ZO-3 share strong sequence homology with ZO-1 (for review, see Ref. 432). Both ZO-1 and ZO-2 are found in the testis; however, it is not known if ZO-3 is present in the testis. A recent study has identified a new MAGUK protein designated Pals1 in TJs in the kidney (368), which functions as an adapter protein linking the mammalian homologs of Crumbs and Discs Lost (the mammalian homolog of Drosophila Discs Lost is Patj). Both are proteins crucial for epithelial cell polarity in Drosophila, and this complex, the Patj-Pals1-Crumbs complex, in turn anchors to the TJ by interacting with the PDZ domain of ZO-1 (368, 405) (Fig. 4). Again, it remains to be determined if such complex-regulating cell polarity exists in the testis.

B) CINGULIN. Cingulin is the first TJ protein identified by using monoclonal antibodies (90, 91) and is found in the rat and mouse testis (see Table 1). Cingulin is a 140-kDa phosphoprotein found at the cytoplasmic plaque of TJs from epithelial and endothelial cells (90) and is further away from the site of the TJ than ZO-1 (431). Nonetheless, cingulin is essential to conferring TJ function by maintaining an impermeable barrier (88, 89). Cingulin shares structural homology with other cytoskeletal proteins, such as myosin rod and paramyosin (88, 91). Recent in vitro immunoprecipitation studies have shown that cingulin interacts with other TJ proteins, in particular ZO-1, ZO-2, ZO-3, AF-6, and myosin (100), suggesting an in vivo interaction of cingulin with ZO-1 and ZO-2 may occur. In addition, cingulin interacts with occludin in vitro (100). Studies by immunofluorescent microscopy have localized cingulin to the same site of ZO-1 at the basal compartment near the BTB in the rat and mouse testes (62). Furthermore, cingulin is also found at the spermatid-Sertoli cell contact sites consistent with its localization at the ectoplasmic specialization (62).

C) SYMPLEKIN. Symplekin is a 150-kDa protein initially identified by the use of monoclonal antibodies raised against lateral membrane junctional extract (for review, see Ref. 432). It has been localized to the site of TJs in polarized epithelia, as well as in Sertoli cells by immunohistochemistry and immunogold electron microscopy, but not at TJs in endothelial cells (236). In addition to its localization in TJs, symplekin is a ubiquitous component of the nucleoplasm in both epithelial and nonepithelial cells (236). Molecular cloning of symplekin has shown that its cDNA consists of a well-conserved Kozak sequence; however, symplekin does not share any homology with ZO-1, ZO-2, or any of the members of the MAGUK family of proteins (463).

D) OTHERS. With the use of electron microscopy techniques in conjunction with specific antibodies, protein kinases, heterotrimeric G proteins, small GTP-binding proteins (such as Rab3B and Rab13), AF6, Src substrate, and c-yes have all been localized to the cytoplasmic surface of TJs (for review, see Ref. 510). However, the functions of many of these molecules are not known (for review, see Ref. 120). Furthermore, these protein complexes eventually anchor to actin. Although actin is not exclusively found at the site of TJs, actin is known to regulate TJ permeability since agents that disrupt the actin cytoskeleton can induce influx through the paracellular space (for review, see Ref. 432). More recent studies have shown that the TJ is a platform for trafficking and signal transduction because many of the proteins necessary for membrane trafficking and signal transduction are found at TJs (see Table 2). Indeed, recent studies have identified a new family of proteins designated junctophilins (JP), which consists of JP-1, -2, and -3 (322), in the junctional complexes in excitable cells, such as heart and skeletal muscle (446). Although JP-1 and JP-2 were not found in the testis but apparently confined to heart and skeletal muscle, JP-3 is largely confined to the brain and testis, localized in the junctional complexes between the plasma membrane and endoplasmic/sarcoplasmic reticulum (446). JP proteins appear to play an important role in facilitating functional cross-talks between cells through the ionic channels on the plasma membrane and endoplasmic/sarcoplasmic reticulum in excitable cells (211, 322).

Recent studies on TJ-associated signaling molecules (see Tables 1 and 2; Fig. 4) have shown that the TJ has emerged as a platform for signal transduction to coordinate different cellular functions, in particular the dynamics of TJ functionality to permit the timely passage of cells, such as preleptotene and leptotene spermatocytes and monocytes across the BTB and epithelium, respectively. However, the precise regulation of these events is not yet known. For instance, small GTPases, Cdc42 and Rac, known to be involved in actin cytoskeleton dynamics and cell polarity, bind to a protein complex containing Par6, Par3/ASIP, and PKC that is found in TJs (for review, see Ref. 510) (Table 2, Fig. 4). When either Par6, Par3, or Cdc42 is overexpressed in epithelial cells, it causes dislocation of ZO-1 from the site of TJs, rendering the loss of cell polarity, indicating the Cdc42/Par6/Par3/PKC complex, along with ZO-1, may be critical to maintain cell-cell contact and cell polarity at the site of TJs (224, 266). Furthermore, small GTPases, such as Rho, Rac, and Cdc42, are important molecules regulating TJ dynamics via their effects on F-actin (for reviews, see Refs. 181, 182). For instance, overexpression of RhoV14 and RacV12 causes disruption of the TJ strands due to a chaotic redistribution of ZO-1 and occludin (230). It is known that these GTPases, such as RhoB, regulate the events of actin organization by targeting vesicles to the appropriate sites in the membranous structures within a cell including the sites of TJs and AJs (for reviews, see Refs. 182, 510). More recent studies have shown that both Sertoli and germ cells express small GTPases, such as Cdc42, N-Ras, Rac2, and RhoB (271-273), indicating both Sertoli and germ cells are likely to take part in the reorganization of their cytoskeleton network to facilitate germ cell movement. It is obvious that studies of small GTPases in the testis should be expanded to understand the precise regulatory pathway(s) by which TJ dynamics are regulated utilizing these GTPases (for review, see Ref. 273).

Although major advances were made in the past two decades identifying many constituent components of TJs in different epithelia including the testis (Tables 1 and 2 and Fig. 4), the factors and pathway(s) that regulate TJ dynamics in the testis are poorly understood. Nevertheless, extensive studies have been performed using epithelial cells, such as MDCK cells and keratinocytes, cultured in vitro to identify and investigate the regulatory molecules and signaling pathway(s) that modulate TJ dynamics (for review, see Ref. 120). To date, different signal transduction pathways are implicated in the regulation of TJ dynamics. These include protein kinases, protein phosphatases (26, 87, 262, 264, 317, 319), intracellular Ca²⁺ (318, 434, 435), G proteins (25, 212), calmodulin, cAMP, and phospholipase C (25, 218). On the basis of these earlier in vitro findings, two biochemical models have been proposed attempting to explain how small molecules, such as fatty acids, amino acids, glucose, and IgG, can traverse epithelial TJs during food absorption and inflammatory responses.

This model is based on the observation that depletion of Ca²⁺ from MDCK cells cultured in vitro induces immediate disruption of the TJ barrier, which is manifested by a plunge in the TER across the cell epithelium (for review, see Ref. 120). Upon addition of [Ca²⁺] to the media, cell polarity restores and the TJ barrier reseals. It is also known that the Sertoli cell TJ barrier can be disrupted and resealed by manipulating [Ca²⁺] in the culture medium (169). For instance, depletion of [Ca²⁺] from the spent media of Sertoli cell cultures can perturb the Sertoli cell TJ barrier within 15 min, and its replacement induces the disrupted TJ barrier to reseal within 90 min, making it indistinguishable from control cultures (169). These results unequivocally demonstrate that [Ca²⁺] plays a very critical role in the regulation of TJ dynamics in the testis. Indeed, calcium is a known cell signaling molecule that regulates a variety of cellular events (for review, see Ref. 37).

During TJ assembly or disassembly, the actin cytoskeleton undergoes extensive polymerization and depolymerization, which is an ATP-dependent event per se. For instance, treatment of cells with cytochalasin, a drug that disrupts actin filaments, can perturb the paracellular barrier (430), suggesting the significance of the cytoskeleton network in TJ functionality. This model hypothesizes (for review, see Ref. 120) that when ATP is depleted from the system, ZO-1 becomes associated with cytoskeletal proteins, such as fodrin. This, in turn, pulls ZO-1 away from TJ sites causing TJ leakiness. Upon repletion of ATP, the association between ZO-1 and fodrin becomes disrupted, allowing ZO-1 molecules to move back to TJ sites, resealing the TJ (457). These models, however, apparently can only explain how TJs become leaky in vitro to allow passage of small molecules and ions. This is in contrast to the dynamics of Sertoli cell TJs that constitute the BTB, which must disassemble to allow for the passage of preleptotene and leptotene spermatocytes. These models also do not take into consideration how the recently identified TJ-integral membrane proteins, such as occludins, claudins, and JAM, take part in these processes.

E.  Regulation of TJ Dynamics

1.  Regulation by protein phosphorylation: the interplay of kinases and phosphatases

In mammalian cells, as much as 30% of the cellular proteins are phosphorylated (122), illustrating the regulatory roles of protein phosphorylation. Indeed, tyrosine phosphorylation of junction-associated proteins is known to play a crucial role in junction assembly (373). For instance, tyrosine kinases of the Src family are found at the sites of TJs and AJs (see Tables 1 and 2 as well as Figs. 4 and 7) (460). -Catenin, an AJ-associated protein, becomes highly phosphorylated in Src-transfected cells and is a putative substrate of protein kinases (33). Recent studies from our laboratory have demonstrated the presence of myotubularin, a putative protein tyrosine phosphatase (PTP) in Sertoli and germ cells (262, 264). Its expression is induced when the Sertoli cell TJ barrier is being assembled in vitro (264). Furthermore, a testis-specific serine/threonine protein kinase is also found in the mouse (38). Thus both AJs and TJs apparently consist of the necessary proteins for signaling via tyrosine phosphorylation. With the use of various PTP inhibitors (PTPi), it was shown that both -catenin and ZO-1 are tyrosine phosphorylated and are putative substrates of tyrosine kinases (424). Also, vanadate (a specific PTPi) can induce a rapid increase in TJ permeability in MDCK cells in vitro as revealed by reduced TER and increased permeability to [³H]inulin by increasing the cellular phosphoprotein content (97). This observation is consistent with our recent observations demonstrating that sodium orthovanadate can perturb the Sertoli cell TJ-permeability barrier in vitro (263, 264). These changes in TJ permeability coincided with an increase in phosphotyrosine immunofluorescence at the site of the TJ and with a redistribution of F-actin, E-cadherin, and ZO-1 when examined by confocal microscopy (97). More importantly, these changes can be blocked in MDCK cells and Sertoli cells by the use of a protein tyrosine kinase (PTK) inhibitor (PTKi), such as tyrphostin A25 (263), but to a significantly lesser extent when a serine/threonine protein kinase inhibitor, such as staurosporine, was used (97). Studies in MDCK cells have shown that the assembly, opening, and resealing of TJs correlate with the phosphorylation of occludin on serine/threonine residues (140). Although the physiological significance of these studies remains to be elucidated, they strongly indicate that the assembly and maintenance of TJs are regulated by the phosphorylation status of cellular proteins whose identities remain to be uncovered. Taking these results collectively, it is apparent that a decline in cellular phosphoprotein content favors the assembly and maintenance of the Sertoli cell TJ barrier, whereas an increase in cellular phosphoprotein content perturbs the Sertoli cell TJ-permeability barrier. On the other hand, these results present a biological dilemma: if occludin molecules found in the TJ fibrils are highly phosphorylated (140), why would the use of a PTPi, such as vanadate, which supposedly increases cellular phosphoprotein content, perturb TJ-permeability barrier as seen in MDCK and Sertoli cells (262, 264)? Other studies have shown that tyrosine phosphorylation of AJ-associated proteins, such as -catenin, -catenin, and E-cadherin, can also reduce AJ stability and perturb cell adhesion, which in turn perturbs the TJ-permeability barrier (351). Taking these results collectively, it seems logical to speculate that an increase in the overall phosphoprotein content of epithelial/endothelial cells reduces the stability of TJs and AJs (based on protein phosphatase inhibitor studies). Furthermore, a local mechanism and a compartmentalized microenvironment may exist at the level of cell membrane, such as the one between preleptotene/leptotene spermatocytes and Sertoli cells, where reduced phosphorylation of occludin can open up the Sertoli TJ barrier, and an increase in occludin phosphorylation can reseal the TJ. Also, in studies using a specific PTPi, such as vanadate, to block tyrosine protein dephosphorylation, PTPi can also inhibit and/or activate other signal transduction pathways, which in turn contribute to the observed junction assembly/disassembly events. Also, the assembly of TJs in MDCK cells is regulated by G proteins, phospholipase C, PKC, and calmodulin (25). Figure 5 depicts a hypothetical molecular model by which the Sertoli cell TJ barrier is regulated by changes in phosphorylation of selected target proteins, such as occludin.

View larger version (31K): [in this window] [in a new window]   Fig. 5. A molecular model of the Sertoli cell TJ dynamics in which multiple molecules that interact with each other to regulate the opening and closing of the Sertoli TJ barrier that constitutes the BTB. Left: Sertoli cell TJ barrier that is closed. Right: TJ barrier that is opened. The tight junction dynamics in the testis are regulated by different sets of molecules and/or functional status of proteins, which include GTPases, proteases/protease inhibitors, protein phosphorylation, intracellular calcium and cAMP levels, and cytokines, such as TGF-3. For instance, reduced phosphorylation of occludin causes the dephosphorylated occludin molecules to move to the basolateral region of the membrane, away from the TJ sites, disrupting the TJ-barrier (see right panel). Likewise, an increase in proteases also favors the TJ disruption (see right panel) but an increase in protease inhibitors favors the closing of the TJ barrier (see left panel). This model was prepared based on reports and reviews discussed in section IIIE. Readers are encouraged to seek additional information from published reports and reviews cited in section III, B-E.

The Rho family of small GTPases belongs to the Ras GTPase superfamily of 20- to 30-kDa GTP-binding proteins, which regulates a wide spectrum of cellular functions (465). Like Rho and Cdc42, Rac cycles between a GDP-bound inactive state and a GTP-bound active state (for reviews, see Refs. 440, 465). GTP-bound active GTPases account for only ~0.5-5% of the total GTPases in a mammalian cell (19, 465). These small GTPases also provide a link between growth factor signaling and reorganization of the actin cytoskeleton (277, 483, 515). They are implicated in the signaling pathways that regulate the initiation and turnover of cell-cell adhesion and cell-substratum contact, which are essential for 1) cell movement and 2) junction assembly (181, 182, 196, 320). Although many of the earlier studies were performed in fibroblasts, recent studies reveal that their roles are not limited to fibroblasts and not restricted to the actin filament network (181, 182). GTPases are regulated by at least four types of proteins. These include 1) guanosine nucleotide exchange factors (GEF), which facilitate the activation of GTPases by promoting the binding of GTP onto GTPases (for reviews, see Refs. 231, 237, 465); 2) GTPase-activating proteins (GAP), which stimulate the intrinsic hydrolysis of GTP in GTP-bound GTPases, enabling GTPases to execute the desired biological function, this in turn renders them inactive (for revi