Tight and gap junctions in the intestinal tract of tunicates (urochordata): A freeze-fracture study

The junctional complexes of the digestive tract of mammalian systems have been extensively studied at the fine-structural level; such epithelia are characterized by a series of intercellular specializations along their lateral borders, which include zonula and macula adherens, tight junctions (zonulae occludentes) and gap junctions (Farquhar & Palade, 1963). Although the tissues of the gut of those invertebrates examined to date reveal desmosomal contacts and gap junctions, they also possess the so-called septate junctions (Wood, 1959; Noirot-Timothée & Noirot, 1980) but no tight junctions. Although septate junctions were for some time considered to be the invertebrate equivalent of the vertebrate tight junctions (Satir & Gilula, 1973; Green et al. 1979), it has recently become clear that tight junctions proper do exist in invertebrates, in tissues where permeability barriers are to be found (Lane, 1981a), although they do not generally occur in the intestinal tract of these animals (Lane & Skaer, 1980; Skaer et al. 1980).

The tunicates (Urochordata) (also called ascidians) occupy a unique position phylogenetically, in that they are deemed to be chordates by virtue of the notochord and neural tube that their motile and ephemeral larval form possesses. In the adult, however, after they have undergone a retrogressive metamorphosis, their nervous system is reduced to a single, invertebrate-like ganglion, housed in a sessile, relatively simplified zooid. In this way they appear to occupy an intermediate position between the two major divisions of the animal kingdom - invertebrate and vertebrate.

A few earlier studies on tunicates (Cloney, 1972; Lorber & Rayns, 1972; Georges, 1979; Green & Bergquist, 1982) revealed that tight junctions were present in this group, at least in the tissues of the heart, tail and epidermis; comparable occluding junctions have also recently been discovered, however, in certain tissues of insects (Lane & Treheme, 1972; Lane, 1972a, 1978, 1979, 1981a, b, 1982; Lane et al. 1977) and arachnids (Lane & Chandler, 1980; Lane, 1981c; Lane et al. 1981), arthropods that have quite unequivocal septate junctions (Noirot-Timothée & Noirot, 1980) in their digestive tracts (Lane & Skaer, 1980). It therefore seemed of interest to examine the tissues of the intestinal tract of a range of tunicates, to determine whether they would exhibit tight or septate junctions between their component epithelial cells. Three different orders, with different genera from each, have been investigated and, although there are subtle distinctions in the organization and complexity of their junctions, as well as in the number and size of their associated gap and intermediate junctions, the apical lateral border in all of the species examined possesses unequivocal tight-junctional networks. These exhibit the characteristic features of zonulae occludentes as first described in thin sections by Farquhar & Palade (1963) and later in freeze-fracture replicas by Claude & Goodenough (1973). A short preliminary report on part of this work has been published elsewhere (Lee et al. 1985).

The tissues studied in this investigation were the intestinal tracts of a range of tunicates, both solitary and colonial. The order Aplousobranchiata was represented by the species Diplosoma listerianum and Clavelina lepadiformis, the order Stolidobranchiata by Molgula socialis, Botryllus schlosseri and Botrylloides leachi, and the order Phlebobranchiata by Ciona intestinalis and Ascidiella aspersa. These organisms were collected variously from the lagoon of Venice (Italy) or from Plymouth (England). They were either studied immediately, or maintained in tanks of circulating sea water before use.

In most cases the intestinal tract was dissected out into a number of regions: the large pharynx, the relatively short oesophagus, the stomach and the intestine. This last region was in some cases divided again into three sections, the anterior intestine, the mid-intestine, which in some cases consisted of a loop, and the distal intestine, or rectum. In the case of the small colonial forms, such as Diplosoma, Botryllus and Botrylloides, whole zooids isolated from the colony were studied and the different regions of the gut determined by their morphological differences in the thin sections or replicas. The tissues were treated with one of a variety of fixatives, of which the most successful was 1 ·5% glutaraldehyde in 0 ·2M-cacodylate buffer, pH 7 ·4, plus T5% NaCl. In some cases phosphate buffer was used instead of cacodylate and 2% sucrose instead of NaCl₂ was added to the final fixative solution. The final measured osmolality was slightly hyperosmotic to that of sea water. In some cases, 1% colloidal lanthanum was added to the cacodylate buffer, using the electronopaque tracer lanthanum as an extracellular ‘negative’ stain. Tissues were then embedded in either Epon or Araldite for thin-sectioning, or frozen for the preparation of freeze-fracture replicas. For the former, tissues were washed, post-fixed in osmium tetroxide, in some cases stained en bloc in uranyl acetate, dehydrated through an ascending series of ethanols, and embedded in epoxy resin. For the latter, the tissues (fragments of gut or pellets of small zooids) were washed and cryoprotected in glycerol at 10%, 20% and 30% in buffer, for varying periods of time, before freezing by plunging into Freon 22 cooled in liquid nitrogen. The material was then mounted in a Balzers freeze-fracturing device (BAF301 or BA360M models) and fractured at −100 to — 115°C and at 2 ×10⁻⁶Torr (1 Torr—133 ·3 Pa). The preparations were shadowed with platinum-carbon or tungsten-tantalum and backed with carbon. The tissue was removed with sodium hypochlorite or biological detergent, the replicas washed in water and mounted on copper grids for examination in a Philips EM 300, 420 or Hitachi H600, at 60 or 80 kV.

The pharynx and oesophagus in these species of tunicates are characterized by cells with many cilia and a few microvilli. The main epithelial cells that occur in the stomach are either the columnar vacuolated absorbing cells or the zymogen cells (Burighel & Milanesi, 1973, 1977); in both cases they, in contrast, exhibit many microvilli but no cilia. Ciliated cells may occur, however, in restricted regions of the stomach. In the intestinal tract, mucous cells that exhibit many cilia and few microvilli, and vacuolated cells that possess microvilli and only rarely, cilia, are the main cell types. These and other features can be used in the small colonial forms to determine which part of the gut is being examined.

The tight junctions are present near the luminal surface of the component cells, seen in thin sections as punctate cell-cell appositions between adjacent cells (Figs 1, 3). Associated with their cytoplasmic face may be fibrous material, presumably cytoskeletal (Fig. 1). Below these apical borders occur regions with the reduced intercellular clefts that characterize gap junctions (Fig. 2); in some cases many of these are found, each in close association with the next (Figs 2, 5). At higher magnification, the tight junctions exhibit a fusion of the two adjacent outer half membrane leaflets (Fig. 1), while the gap junctions reveal a 2–4nm cleft between their apposed membranes (Fig. 2). In the specimens incubated with lanthanum, the lanthanum was halted at the punctate tight-junctional appositions (Fig. 3). At the level of the gap junctions the tracer penetrates the intercellular clefts, which then have a cross-striated appearance when they are slightly obliquely sectioned (Fig. 2).

In some cases, other, more desmosomal-like structures are present near the tight junctions. These do not resemble septate junctions, which are not present in these tissues. They are characterized in thin section by an enhanced density along the lateral cell borders (inset, Fig. 6), which appears to consist of cytoplasmic fibrils condensed into a feltwork along the side of the membrane. Sometimes, in fortuitous sections, the fibrils can be seen to have a zonular distribution, where they lie as a bundle running from one intercellular junction to another. They appear similar to the zonula adhaerens or intermediate junctions described by Farquhar & Palade (1963). There is no particular organization of intramembrane particles (IMPs) or any other apparent freeze-fracture profile that can be correlated with these contacts (Figs 4, 6, 12, 16) and so it must be presumed that the fibrils do not insert into the membrane as far as its mid-line, the plane of fracture.

Other than this, the freeze-fracture appearance (Fig. 6) of a particular tissue of any given species can be closely correlated with thin sections of comparable tissue (Fig. 5), in that a network of tight junctions can be seen to be present in the apical region, while macular gap junctions lie beneath. In some cases the population of gap junctions is very extensive (Fig. 6); in others, they may be much fewer in number (Figs 4, 7) or none may seem to be present at all (Figs 8—11). This appears to vary with the cell type or with the particular area under consideration; in general, the intestine and pharynx seem to have fewer gap junctions than does the stomach, while the number in the oesophagus is more variable.

The tight junctions form a circumferential band between the lateral borders at the apices of the cells. In all the regions of the gut examined these junctions feature aligned PF particles ranging around 8–12 nm in diameter, arranged in a reticular network (Figs 7 –16). In some cases the particles appear fused into ridges on the P face (Fig. 7), but this is not always so (Fig. 10) and may be due to the angle of shadowing or the response of the junctional components to the cross-linking effects of glutaraldehyde. Some particles may fracture up onto the EF, where they lie in the grooves to be found there (Figs 10, 11). That these E-face grooves are complementary to the PF ridges can be seen when the fracture plane is in transition from EF to PF (Fig. 12). The two are in coincidence with one another, except that the PF ridges are slightly offset with respect to the EF grooves (small arrows in Fig. 12).

The tight-j unctional belt varies in depth as it is composed of a variable number of strands; it is a fairly broad band in the tissues of the pharynx, oesophagus, stomach and the anterior part of the intestine (Fig. 10). In some cases, however, occasionally in the stomach (Figs 6, 25) or posterior intestine, the network band is less wide and may consist of only a few anastomosing strands, at least insofar as the fractured membrane face reveals. There seems to be no consistent pattern to this variable. In other cases, particularly in the stomach (Figs 15, 16), there is a tendency for the basal component of the tight-junctional network to be in looser strands, which branch out from the reticulum. These may continue as extensive individual ridges, running in wavy alignments parallel to the longitudinal axis of the cells, separately, with few anastomoses; this is particularly striking in Ciona (Fig. 17).

The gap junctions, which are found beneath the tight junctions, may lie very close to them (Figs 6, 15, 18, 19) or may be further down the lateral border. They vary in number from many (Figs 6, 18) to relatively few (Figs 7, 12). They are always composed of P-face connexons ranging from 8 nm to 13 nm in diameter (Figs 15, 19), which are usually clustered together in close association, commonly in circular plaques (Figs 6, 15, 18, 20, 21). That these are the sites of the gap-junctional reduced intercellular cleft seen in thin sections is evident where the fracture plane cleaves from PF to EF (Figs 7, 13, 19, 20, 23). Frequently, fragments of the E-face membrane remain attached to the maculae, and the EF pits can be seen on these remnants (Fig. 22); these exhibit the regular connexon packing rather clearly. In other cases, some of the PF particles may adhere to the EF (Fig. 23) where the junctions are recognizable as clusters of pits. There are also some irregular gap-junctional plaques, which assume angular, polygonal outlines (Figs 22, 23, 24), particularly in the stomach of Molgula. These shapes may be found on cell borders close to those with more normal irregular outlines and in some cases transition forms can be found (Fig. 25). Examination of all the other parts of the gut tract of Molgula reveals that the gap junctions tend not to be polygonal elsewhere.

The size of the gap junctions in the tunicate gut tends to be within the range from 0 ·1 μm to 0 ·6 μm (± a few 0 ·4 μm; see Figs 6, 18, 25); that is, there tend to be no exceptionally large gap junctions. However, in some relatively rare cases they may be extremely small in diameter (Figs 12, 16, 26, 28), giving the impression that they are becoming fragmented or perhaps, disaggregated (Figs 26, 27). This may also be the case when small plaques are found near the end of tight-junctional strands (Fig. 28); this is particularly common in Ciona, where normal-sized gap-junctional plaques are remarkably infrequent.

The occurrence of tight junctions in the tissues of tunicates was first reported between the caudal epidermal cells of the ascidian Distaplia (Cloney, 1972), and in the heart of adult Ascidiella (Lorber & Rayns, 1972) from thin sections. Freezefracture studies on heart tissue (Lorber & Rayns, 1972), and on the epidermis and pharyngeal epithelium (Georges, 1979) of adult tunicates, revealed strands of distinct particles forming belt-like networks at the apical parts of cells; anastomosing patterns of IMPs were also observed in the tight-junctional belts of Asterocarpa (Green & Bergquist, 1982). Since these junctional structures were observed initially only in the tissues of adults, or between the differentiated tail epidermal cells in larvae, it was considered that they were characteristic of cells that were completely differentiated, and that they were capable of forming a tight barrier between two compartments (Georges, 1979). In the tissues of the intestinal tract of the range of ascidian genera examined here, the same interpretation seems justified. Their distribution, in fact, parallels that typically encountered in vertebrate gut (Claude & Goodenough, 1973) and clearly demonstrates the basic similarity of the tight junctions found in urochordates and vertebrates. In thin sections, this likeness is also apparent as regards the features of the punctate appositions between adjoining cells, with apparent fusion of the two external half membrane leaflets. Fibrous cytoplasmic material is in some cases associated not only with the zonula adhaerens, but also with these junctions, as described in earlier studies on tunicates (Cloney, 1972); this too parallels the observations made on vertebrate systems, where cytoskeletal components have been found inserting into tight-junctional contact sites (Bentzel & Hainau, 1979) where they may be involved in modulating their tightness. Regulation of tight-junctional permeability in vertebrate tissues has been considered in detail elsewhere (Schneeberger & Lynch, 1984), but there is very much less relevant information available for urochordates and seemingly none on their gut tract. However, during tail resorption in a tunicate larval tadpole (Cloney, 1972), tight junctions were thought to be involved in the contraction process mediated by cytoskeletal components. In such circumstances the role of the filaments that are also seen to insert into the zonula adhaerens, which may occur near the tight junctions in tunicate tissues, is not clear.

The disposition of the tight-junctional strands within the gut cells of tunicates seen in freeze-fracture replicas is as particle alignments or ridges on the P face and E face grooves, with occasional rows of distinct particles within them. This distribution seems very similar in all the ascidians studied here. In an early study on the epidermis by Georges (1979), the PF and EF fracture faces revealed slightly different arrangements, while Green & Bergquist (1982) found alignments of discrete particles (5–25 nm long) that fracture onto the E face in both fixed and unfixed preparations. There seems to be great disparity in particle diameter, with Georges (1979) reporting the junctional particles in Phallusia to be more variable in diameter (6 ·5–25 nm) than those of Ciona and Morchellium (12 ± 2nm). The number of rows of particles and the degree of their anastomosis is not very consistent in the range of species examined here, although there is a general tendency for the junctional networks to become looser as they are further removed from the cell’s apical border. What variations have been observed in the several segments of the alimentary tract may reflect a different functional state of the cells. It has been suggested that the tight junctions of different tissues can vary their geometric pattern according to the flexibility of the plasma membrane in a given region (Hull & Staehelin, 1976). This morphological flexibility would enable the junctions to maintain the transepithelial sealing capacity even when the tissue is undergoing osmotic stress or mechanical stretching.

The studies with tracers suggest that, in the tunicates, these junctions form the morphological basis of permeability barriers, as they do in higher vertebrates (Claude & Goodenough, 1973; van Deurs, 1980), although they may in certain cases be leaky to lanthanum (Green & Bergquist, 1982). The variability in the disposition of their IMPs as separate particles or as ridges may relate to leakiness or may be a feature that is determined in part by their response to the cross-linking effects of glutaraldehyde (van Deurs & Luft, 1979). The angle of shadowing would also be important, as has been observed in studies on the IMPs of septate junctions (Kacher et al. 1986). Rapid freezing would be required to determine if these structures are cylindrical in vivo and hence possibly formed from inverted lipid micelles, stabilized by protein, as has been suggested for vertebrate tight junctions (Kachar & Reese, 1982; Pinto da Silva & Kachar, 1982). As found, until recently, with the vertebrate tight junctions (Stevenson & Goodenough, 1984), no entirely satisfactory isolation procedure has been devised for tunicates, so pure preparations of tight junctions are unavailable for biochemical analysis. Certainly the large numbers of strands in the tunicate gut would make them admirable material for such studies. Moreover, in the case of both the IMPs and ridges of these tight junctions, the PF structures are offset with regard to the complementary EF grooves, which suggests that they have their structural basis in a double fibril, as has also been observed in vertebrate tissues (Bullivant, 1978). A similar model has been proposed for the arthropod tight junctions (Lane, 1981a) and these also, after fast freezing, appear as continuous cylinders (Lane, 1984). The Arthropoda are the only major invertebrate group that possesses unequivocal tight junctions (Lane, 1981a) and so are the only group without ‘backbones’ with which the tunicates may be compared. Tight junctions, rather than septate junctions, also seem to form the basis of the permeability barriers observed in the arthropods, such as in the blood-brain (Lane, 1972a, 1978, 1981a) and blood-testis (Toshimori et al. 1979) barriers. However, the tight-junctional belts tend to be rather simple in the insects (Lane, 1982) and, although they appear as more extensive networks in the arachnids (Lane & Chandler, 1980; Lane et al. 1981), the tight junctions of tunicates are structurally much more complex in the extent of their networks, like those reported in the tissues of vertebrates. We do not find strict PF particle adherence in tunicate tight junctions but do not consider that such a characteristic, together with a dearth of continuous ridges, would constitute a close relationship to septate junctions phylogenetically, as suggested elsewhere (Green & Bergquist, 1982). Septate junctions, which occur in the intestine of arthropods (Skaer et al. 1980), are not to be found in the gut of either tunicates or vertebrates.

The gap junctions seen here, as also described in other tunicate tissues (Lorber & Rayns, 1972, 1977; Georges, 1979), are characteristic of those in vertebrate tissue in that their component particles all fracture onto the P face, leaving complementary EF pits. This is also the case for some other invertebrate phyla, although not the arthropods, in which the connexons are larger and fracture onto the E face (Lane & Skaer, 1980). The actual size of the individual connexon measured here in the gut is, on average, 8–13 nm, in comparison with the 5 nm diameter of the particles comprising them in tunicate epidermal and pharyngeal cells (Georges, 1979). The junctions do exhibit, however, great variability in their number, since in some cases (e.g. Diplosoma and Molgula), gap junctions are very numerous, while in Ciona (as also noted by Georges, 1979), they are relatively infrequent. The size of the junctional plaques themselves is also somewhat variable, those of Ciona being particularly small. The significance of the polygonal outlines exhibited by some gap junctions is not clear, but one possibility is that this reflects their state of coupling (see Peracchia & Dulhunty, 1976; Peracchia, 1977), the more closely packed arrays tending to be uncoupled. Again, fast freezing without fixation might elucidate this point, although earlier studies on such phenomena in both urochordates (Hanna et al. 1981) and mammalian systems (Raviola et al. 1980), as well as crustaceans (van Deurs et al. 1982) and insects (Swales & Lane, 1983), did not support such a contention.

Certainly these organisms are in an interesting phylogenetic position: in the distribution of tight junctions throughout their gut tract, in the morphological features of these and their associated gap junctions, tunicates appear more chordate-like than invertebrate-like. Further, they appear to lack septate junctions, which are a feature unique to the latter group. It is therefore intriguing that in certain other respects, such as the relative simplicity of their adult nervous system, which lacks a blood-brain barrier (Lane, 1972), as well as any glial ensheathment or myelination (Lane, unpublished observations), tunicates are more invertebrate-like. Moreover, scalariform junctions, which are related to septate junctions (Lane & Skaer, 1980), have recently been found in the gut oiBotryllus (Burighel et al. 1985); the possession of these has hitherto been considered to be a uniquely arthropod-like characteristic. In addition, in previous reports the myofibrillar structures of tunicate striated muscles have been described as being typical of chordates, while their sarcotubular system is more similar to that of invertebrates (Burighel et al. 1977).

Further studies on developing tissues, such as the central nervous system, muscle or gut during tunicate retrogressive metamorphosis, may shed light on this dilemma.

We thank Dr Riccardo Brunetti for the selected specimens, Paolo Salvatici and William Lee for technical assistance with the production of the freeze-fracture replicas, Valentino Miolo for help with sectioning, and Ermanno Malatesta and Vanessa Rule for help with typing the manuscript. We are also indebted to MPI for grants during the course of this research.