J. Cell Sci. 32, 293-305 (1978) 293Printed in Great Britain © Company of Biologists Limited

DEVELOPMENTAL STAGES IN THE

FORMATION OF INVERTED GAP JUNCTIONS

DURING TURNOVER IN THE ADULT

HORSESHOE CRAB, LIMULUS

NANCY J. LANE

A.R.C. Unit of Invertebrate Chemistry and Physiology, Department of Zoology,University of Cambridge, Doioning Street, Cambridge, England

SUMMARY

Stages leading to the formation of inverted gap junctions between certain basal replacementor interstitial cells in the mid-gut of adult Limuhis can be followed by freeze-fracturing.Free, 13-nm EF intramembranous particles first appear to be organized into short lineararrays or small clusters of particles, which then become transformed into anastomosingparticulate networks covering a considerable surface area. These subsequently become con-centrated into smaller, more nearly circular, macular plaques of EF particles or PF pits.These EF particles, both when free or assembled into macular arrays, possess a centralchannel or pore. Numerous formed gap junctions are present in Limulus mid-gut, whichsuggests that cell-to-cell communication is an important feature of the mature tissue. Theresults show that arthropod tissues can be used to study the development of gap junctionsnot only in differentiating systems but also in adult tissues during normal cell turnover.

INTRODUCTION

A number of studies have been made on vertebrate tissues to determine the changesin particle distribution leading to the formation of gap junctions; these includeinvestigations on embryonic tissues (Revel, Yip & Chang, 1973; Revel, 1974; Decker& Friend, 1974; Hasty & Hay, 1977), regenerating or maturing systems (Yee, 1972;Benedetti, Dunia & Bloemendal, 1974; Albertini & Anderson, 1974), experimentallystimulated material (Decker, 19766) and, in vitro, on cells in culture (Pinto da Silva& Gilula, 1972; Johnson, Hammer, Sheridan & Revel, 1974; Elias & Friend, 1976).

These experiments have primarily made use of freeze-fracturing, since this tech-nique permits the analysis of any changes in the number, pattern or distribution ofthe intramembranous particles which comprise the gap junctions. When fully formedin vertebrates, mature gap junctions consist of macular aggregations of PF particles,about 10 nm in diameter, usually packed fairly tightly in hexagonal arrays (Staehelin,1974). There are some exceptions to this tight packing in mature junctions, such asin mesangial cells (Pricam, Humbert, Perrelet & Orci, 1974), retina (Raviola &Gilula, 1973), mesenteric arteries (Simionescu, Simionescu & Palade, 1976) andfrog ventricle (Kensler, Brink & Dewey, 1977), but they are in the minority.

In invertebrate tissues, however, with the exception of those of molluscs andtunicates, whose junctional particles fracture on to the PF (Flower, 1971, 1977;

294 N- J- Lane

Gilula & Satir, 1971; Lorber & Rayns, 1972, 1977), gap junctions consist of maculararrays of loosely packed EF particles, about 12-13 nm in diameter, often referred toas 'inverted' gap junctions (Flower, 1972). Although mature gap junctions havebeen described in a variety of invertebrate cell types, ranging from tissues such asepithelia in platyhelminthes and annelids (Flower, 1977), molluscs (Flower, 1971,1977; Gilula & Satir, 1971), Hydra (Hand & Gobel, 1972; Filshie & Flower, 1977),tunicates (Lorber & Rayns, 1972, 1977) and arthropods (Hudspeth & Revel, 1971;Flower, 1972, 1977; Satir & Fong, 1973; Satir & Gilula, 1973; Johnson, Herman &Preus, 1973; Noirot-Timothe'e & Noirot, 1974; Baerwald, 1975; Gilula, 1975; Skaer,Berridge & Lee, 1975; Dallai, 1975), to the central nervous system of Crustacea andinsects (Peracchia, 1973a, b; Skaer & Lane, 1974; Lane, Skaer & Swales, 1975,1977; Lane & Swales, 1976), few studies have so far been made of their development.Those which have been made include experiments to investigate the effect of hormoneson chelicerate arthropods. Very loose gap junctions in the horseshoe crab wereinterpreted as being in process of formation, as an apparent response to ecdysterone(Johnson, Quick, Johnson & Herman, 1974); this study, however, has been reportedin abstract form only. More extensive studies on the development of gap junctionshave been carried out on insects in vivo; these investigations on differentiating larvaland pupal tissues of the blow-fly Calliphora (Lane & Swales, 19776, c) have shownthat the gap junctions in the perineurium and glia of the CNS form in early larvalstages, apparently disaggregate in early pupae and reform in late pupae before theemergence of adult flies. In the moth Manduca sexta, embryonic tissues also displaystages in the formation of gap junctions (Lane & Swales, 1977 a). In Calliphora, thestages in the development of the junctions show some similarities to those observedin vertebrates, except that the junctions are inverted, so that the component particlesfracture on to the EF not the PF, and the particles are usually of larger diameterthan those of vertebrates. In addition, they tend not to display clear 'formationplaque' areas (see Decker, 1976a) nor the much larger precursor particles (as inDecker & Friend, 1974, and Revel, 1974), and they do not present linear arraysordered with the same precision as in vertebrates (see Lane & Swales, 19776, c).However, the junctional particles apparently change from being scattered, free13-nm EF particles to small clusters and linear arrays which gradually aggregate toform the mature macular junctions (Lane & Swales, 19776, c). The present studyon the mid-gut of Limulus suggests that stages in the formation of gap junctionsmay occur normally in adult tissues, presumably as the interstitial or replacementcells move up to take their place in, or make contact with, the mature columnarepithelial cells. A similar system of continuous regeneration occurs in insect gut(Smith, 1968) and, interestingly, has been found to be associated with the presenceof continuous junctions (Satir & Gilula, 1973) which also are present in Limulusmid-gut (Lane & Harrison, 1977).

The present report indicates that arthropod tissues can be used to study thestages in development of gap junctions, not only in maturing systems during growthand differentiation, but also in adult tissues during normal cell turnover or whenmature cells begin to make a fresh contact with another cell to establish new inter-

Formation of gap junctions in Limulus 295

cellular junctions. Earlier studies on the mid-gut of Limulus showed that the gapjunctions ordinarily are circular in outline with a fairly loose packing of their com-ponent particles (Johnson et al. 1973; Lane & Harrison, 1977), as is typical ofmature arthropod gap junctions generally.

MATERIALS AND METHODS

The tissue used was the mid-gut of adult specimens of the horseshoe-crab, Limulus poly-phentus; hepatopancreas was also examined. The animals were obtained from Woods Hole,Mass., U.S.A. and maintained in large aerated tanks of sea water at around 16 °C. The materialwas fixed in one of a variety of ways, optimal preservation being obtained with fixation at4 °C in 0-75% glutaraldehyde in o-i M cacodylate buffer, pH 7-4, plus 5 % formalin, 1%acrolein, 3 % NaCl and 35 % sucrose (Fahrenbach, 1976). Tissues to be embedded werewashed, post-osmicated, en bloc stained with uranyl acetate, dehydrated, and embedded inAraldite. Thin sections were stained with lead citrate and uranyl acetate. The tissues forfreeze-cleaving were left in fixative for about 0-5 h, washed briefly in several changes ofo-i M cacodylate buffer, pH 7-4, plus 8 % sucrose, and treated with 2 0 % glycerol in the caco-dylate buffer rinse for 20-30 min before mounting and rapid freezing in Freon 22 cooled withliquid nitrogen. Unfixed tissues were also treated with buffered 20 % glycerol prior to freezing.Freeze-fracturing took place in a Balzers BA 360M freeze-etching device at about 133 x io~*N m"1 (15 x io~' torr) at — 100 °C and the fracture face was shadowed with tungsten-tantalumor platinum-carbon and backed with carbon. The replicas were cleaned by brief treatmentwith concentrated H2SO4 followed by concentrated sodium hypochlorite and then weretreated with distilled water washes after rinsing in dimethyl formamide. Replicas were mountedon coated grids and both they and the thin sections were examined in a Philips EM300.

The freeze-fracture micrographs are mounted so that the direction of shadow is from thebottom or side.

RESULTS

The mid-gut of Limulus is lined by a simple columnar epithelium with microvillion the luminal surface (Fig. 1). The adjacent lateral cell membranes are associatedby desmosomes at the very edge of the lumen, and below that, by 'stacked' continuousjunctions (see Fig. 1; Lane & Harrison, 1977). In the areas of these zonulae continuae,gap junctions {maculae communicantes) occur; the latter are characterized in thinsections by a reduced intercellular space between the apposed plasma membranes(Fig. 2) and occur singly (Fig. 2A) or in groups (Fig. 2B). The continuous junctionsare more conspicious when closer to the apical border of microvilli (Figs. 1, 2A).In the more basal areas of the cells the gap junctions sometimes appear less lengthy,at times almost punctate (Fig. 2 c).

In freeze-fracture preparations, although the particles composing the continuousjunctions are often of very low profile and barely distinguishable (for example, seeFig. 10, cj in inset), the mature gap junctions are prominent maculae, consistingof round to oval arrays of loosely packed EF particles (Fig. 4). The componentparticles range from 10 to 15 nm in diameter (most commonly found to be 12-14 nm,and averaging 13 nm), with complementary pits in the PF (Fig. 4).

An examination of cell membranes closer to the basal border reveals a complexsystem; here interstitial cells are being inserted and replacement cells are differentiatingto reinstate the mid-gut cells lost by degeneration (Fig. 3). In such areas the lateral

296 N. J. Lane

Formation of gap junctions in Limulus 297

membranes may display, in some freeze-fractured replicas, elongate, irregularlyshaped or dispersed gap junctions (Fig. 5). In some cases these EF particle clustersare relatively small, suggesting that coalescence of such small aggregates may haveoccurred to form the larger irregular plaques. On other membrane faces closer stillto the basal area, the intramembranous EF particles sometimes seem to be groupedtogether, with 13-nm particle-free membrane ridges between them, but the particlesthemselves are only very loosely associated (Fig. 6). Small clusters of particles, somein the form of linear strands, may be seen in adjacent membranes, aggregated atrandom (Figs. 7, 8). In all cases, these areas of loose aggregates, when present, arerestricted to certain regions of the membrane face; they are bounded by areas ofmembrane which are free of such large intramembranous particles (see * in Fig. 7,inset, dotted line in Figs. 8, 10). Other deep membrane faces may show linear arraysof particles seemingly only beginning to be arranged in a loose network (Fig. 9); thismay nevertheless be very extensive, covering a considerable patch, but distinctlybordered by a relatively particle-free area of membrane face. In regions close tothese, small clusters (Fig. 9, inset) or short linear arrays of individual 13-nm particlescan be seen lying scattered at random on the EF, sometimes near free 13-nm particles(Fig. io), but no conspicuously larger particles have been observed. These loosepatterns of intramembranous particles have been found close to the basal border inboth fixed and unfixed preparations, but they occur only in a proportion of thepreparations examined, presumably those where the membrane of a new cell isbeing inserted and is seeking to establish low-resistance pathways between itself andthe other (i.e. mature) mid-gut cells. In certain cases, the free 13-nm particles canbe seen lying beside indistinct rows of other particles (Fig. 10, inset) that are part ofthe stacked continuous junctions which occur between adjacent mid-gut cells inLimulus (see Fig. 2A and Lane & Harrison, 1977). The 'formation plaque' area ofmembrane where the particles aggregate is not characterized by any special freedomfrom other particles (Figs. 9, 7, inset), although the PF just beside the formationarea often possesses many smaller ones, similar to normal intramembranous particles

Fig. 1. Thin section through the mid-gut of Limulus showing columnar epithelialcells. The apical border consists of microvilli (vw) and the lateral borders betweenadjacent cells possess desmosomes (d) at the lumen and thereafter continuousjunctions (cj) and gap junctions. Note the cisternae of endoplasmic reticulum lyingparallel to these junctions and the dense secretory granules, m, multivesicularbodies, x 6800.

Fig. 2. Gap junctions as seen in ultrathin sections, between adjacent epithelial cellsof Limulus mid-gut.

A, gap junction (j>) fairly close to the lumen of the gut, lying between continuousjunctions whose faint cross-striations are scarcely visible (arrows), x 115 800.

B, gap junctions (at g) farther removed from the gut lumen, x 91 900.c, gap junctions (#) closer to the basal area of mid-gut cells; several junctions have

undulating membranes in between, and there is one punctate apposition (arrow),perhaps where a gap junction is about to form, x 73700.

Fig. 3. Area near basal region of mid-gut cells showing complex cellular inter-digitations. Note degenerating cell (dc), presumably to be replaced by another.n, nuclei, x 6000.

N. jf. Lane

PF

\

Formation of gap junctions in Limulus 299

(arrows in Fig. 6, p in Figs. 8, 10). In some cases certain of these form PF ridgeswith complementary EF grooves (Fig. 4).

The differing arrangements of 13-nm particles described above suggest the possi-bility that they could be different stages in the process of gap junction formation.If so, several stages in the formation of these junctions may occur on the samemembrane face (as in Fig. 10); particle patterns typical of a particular stage tend tocluster together, often near the next stage in junctional formation (see Figs. 9, 10)as if the tracts of membrane and their component particles had been organized intoa series of developmental steps each of which must occur before they move on tothe next stage.

At higher magnifications the free 13-nm particles sometimes display a centraldepression (Fig. 11) which may represent a channel to be used for cell-to-cellexchanges in formed junctions. This can be found in 13-nm particles in other stagesof junction development as well as in the individual particles within the maturejunctions (Fig. 4).

In adult Limulus hepatopancreas, the gap junctions are sometimes very loose, withcircular arrays of 13-nm particles surrounding a strikingly particle-free interior(Figs. 12, 13). These junctions may at other times be very closely packed in hepato-pancreas and so a structural heterogeneity is to be found in what would appear tobe mature gap junctions; these ring-like particle arrays are also typical of such

Figs. 4-11. Freeze-fracture preparations of mid-gut of Limulus to show changes in13-nm particle distribution that occur apparently when replacement or interstitialcells are establishing contact with other mid-gut cells. The sequence of micrographsin Figs. 4-10 is such as to portray the possible sequence of events that occur duringthe development of the junctions, working from Fig. 10 through to Fig. 4; hence theone-quarter plate micrographs from Figs. 4-10 are printed at the same magnification,x 51000.

Fig. 4. Gap junctions between established mid-gut cells near the apical borderof columnar epithelial cells. EF (EF) shows macular arrays of 13-nm particlesrelatively loosely packed, typical of fully formed gap junctions,-while PF (PF) possessesmacular aggregations (gp) of complementary pits. Note the short ridge-like arrayon the PF (arrow) and a complementary groove on the EF (double arrows) x 51000.Inset, higher magnification of junction cleaved to reveal both EF (EF) and PF (PF).Note central depressions in some of the component EF particles (as at arrow), x 92 250.

Fig. 5. 13-nm EF particle arrays showing irregular outlines of gap junctions duringtheir formation when component particles have apparently begun to become con-centrated in particular areas of membrane. Linear strands (at arrows) are stilljoining some clustered arrays together, x 51000.

Fig. 6. 13-nm particle arrays on the EF (EF) and complementary pits in thePF (PF) of extensive pre-junctional arrays that are just beginning to coalesce orbecome concentrated into separate plaques. Note the other intramembranousparticles that occur, especially in the PF (arrows), x 51000.

Fig. 7. Loose 13-nm particle arrays in the mid-gut EF. Some particles are becomingclustered and some irregular linear arrays occur (arrows) as particles become aligned,x 51000. Inset shows a low-power view of an area with a comparable particledistribution, to illustrate that not all membranes are junction-bearing. 'Smooth'regions with only normal intramembranous particles (•) separate the 13-nm particle-laden areas where junctions are in the process of forming, x 27900.

N. J. Lane

8

PF

P

13

Formation of gap junctions in Limulus 301

vertebrate junctions as the mature frog nexus (Kensler et al. 1977) and so they donot necessarily represent a stage in the development of the junctions. The circulargap junctions seen in hepatopancreas are quite distinct from the scattered 13-nmparticles of the mid-gut, which seem ultimately to aggregate into macular arrays.

DISCUSSION

The different patterns of 13-nm particle distribution observed in the basal areaof Limulus mid-gut epithelium, i.e. in those regions where epithelial regenerationwould occur, suggest that this could be a site of junctional development betweenthe replacement cells and the others. However, it may be that the forming junctionsobserved here are making connexions between the mid-gut epithelial cells and theinterstitial or reserve cells which have been shown to send cytoplasmic extensionsinto the mid-gut in Limulus (Johnson et al. 1973). Heterocellular gap junctions havebeen observed between mid-gut and interstitial cells, although only in thin sections,not in replicas; in lanthanum-impregnated tangential sections the subunit packingof these junctions has been reported to be very loose at times (Johnson et al. 1973),as would be expected in gap junctions which were in process of developing.

Although it is not possible to be certain, the loose particle arrays observed infreeze-fracture replicas and the punctate gap junctions seen in thin sections of themore basal areas suggest the presence of forming junctions and these, together withthe appearance of distinctly mature gap junctions nearer the lumen, and the consistent13-nm diameter of the particles composing the various arrays, make it tempting to

Fig. 8. Loose clusters and linear arrays of 13-nm particles in mid-gut EF (EF)with complementary PF (PF) pits. Note boundary (dotted line) of junctional area.p, smaller particles on the PF. x 51000.

Fig. 9. Short alignments of 13-nm EF particles (EF) with complementary grooveson the PF (PF). Note other smaller intramembranous particles, especially on thePF. Arrow indicates area, lower third of membrane face, where 13-nm particles areclustered more than in the upper part of the membrane. Inset, small clusters of13-nm EF particles (EF) with PF pits (arrows) which sometimes occur instead of, oralong with, the linear arrays, x 38500.

Fig. 10. Free 13-nm EF particles on membrane face above an area (starting atthe big arrow) where linear arrays are apparently beginning to form (arrows). ThePF (PF) shows pits as well as smaller intramembranous particles (p). The dottedlines show the boundary of the presumptive junctional area, x 51000. Inset showsan area with loose, 13-nm particles beginning to line up in short ridges, two particleslong. On the same membrane face, to the right, are particles characteristic of Lrniuluscontinuous junctions (cj) which also occur between mid-gut cells (see Fig. 2A).x 55 100.

Fig. 11. 13-nm EF particles revealing the central channel or pore (arrows), alsocharacteristic of particles comprising the mature gap junctions (see Fig. 4, inset),x 113 600.

Figs. 12, 13. Freeze-cleave preparations from Limiihu hepatopancreas to showthe loose, ring-like gap junctions that occur between the lateral borders of its com-ponent cells. These have a particle-free membrane area (•) within the 13-nm particlerings, but other intramembranous particles occur in the EF of the membranearound them. Fig. 12, x 61900; Fig. 13, x 62700.

20 CKL 32

302 N. J. Lane

speculate that stages in the formation of gap junctions are being observed. Therewould appear to be a sequence of stages, commencing with free 13-nm particles,then linear arrays and clusters of particles (possibly identifiable in thin sections asnear-punctate membrane appositions), which became arranged in an anastomosingpattern and finally the particles become loosely, then closely associated and con-centrated into macular plaques. Rounding up occurs to transform irregular formsinto more circular or elliptical ones. As in insects (Lane & Swales, 1977b, c), sucha sequence of events would exhibit certain dissimilarities to the situation generallydescribed for vertebrate tissues, in that the particles are inserted in the EF not PF,no large precursor particles are to be found (see Revel, 1974), while the 'formationplaque' area is not particularly particle-free (for example, see Decker, 1976a). Thislast feature differs from the situation observed in Limulus hepatopancreas, wheresome gap junctions are very loose, and therefore possibly in the act of forming, andtend to be associated with a remarkably smooth inner membrane area (as in Figs. 12,13). It is possible that the arrays seen in the mid-gut represent junctions that arebreaking down, not forming, but given the nature of the tissue and its function, itseems more reasonable that the gap junctions are in the process of maturing, especiallyas degenerating gaps might here be expected to be disposed of by internalization asoccurs in decidual degeneration (Amsterdam, Bratosin & Lindner, 1976).

The anastomosing network-like arrays of particles observed in Limulus mid-gutare more extensive than the linear arrays of intramembranous particles found indeveloping gap junctions in vertebrates (such as in Decker, 1976a, b; Decker &Friend, 1974; Benedetti et al. 1974) or in insects such as larval Calliphora (Lane &Swales, 19776), larval Manduca (Lane & Swales, 1977a) or late pupal Calliphora(Lane & Swales, 1977c). In these insects as well as in vertebrates, the junctionalparticles seem to line up in relatively short random linear arrays which are scatteredabout, before coalescing into mature macular aggregation plaques, while in Limuluslarge networks of particles are distributed over considerable areas of membraneface and then tend to aggregate. The significance of these differences is as yetobscure, but they may reflect variations in such parameters as the response of themembrane particles to the stimulus to cluster, the chemical nature of the intra-membranous particles, the make-up of the fluid plasma membrane within which theparticles may move translaterally, or differences in the peripheral glycoproteins(see Hasty & Hay, 1977) with which they may be associated.

It is interesting that hormonal induction of gap junctions in horseshoe crabsseemed to produce only irregular or elongated gap junctions with loosely packedcomponent particles (Johnson et al. 1974), not the extensive anastomosing formsobserved here. It is possible that the hormone did not actually induce formation, andthat what was observed were some of the less closely packed, mature junctions.

It would appear that gap junctions are being formed between interstitial or replace-ment cells and established cells in order to maintain cell-to-cell communicationbetween adjacent mid-gut cells in Limulus. Presumably the central pore observedin the junctional particles would correspond to the channel utilized in maturejunctions when two cells are coupled, for exchange of information or small-molecular-

Formation of gap junctions in Limukis 303

weight materials, as suggested elsewhere in other invertebrate tissues (Perrachia,19736; Dallai, 1975; Lorber & Rayns, 1977). The channels found in the free 13-nmparticles would presumably be modified in the configuration of their protein subunitsso as to be effectively closed; at this stage the two cells would not yet be aligned withrespect to their 13-nm particles and hence would be uncoupled.

The short PF ridge-like arrays of particles with complementary EF grooves whichare encountered in mid-gut membranes bear similarities to those described in insectCNS (Skaer & Lane, 1974; Lane et al. 1975; Lane & Swales, 1976, 1977a, b, c;Wood, Pfenninger & Cohen, 1977) and tracheoles (Lane, Skaer & Swales, 1977;Lane & Swales, 1977c) and it has been suggested that they could possibly be axo-glial maculae occludentes in the developing CNS of Calliphora (Lane & Swales, 19776).They have also been seen in insect muscle (Smith & Aldrich, 1971) as well as inother Limulus tissues (Lane, in preparation) and they have been briefly described as'scattered segments of tight junctions' in flea mid-gut (Ito, Vinson & McGuire,1975). Their function, however, remains obscure.

I should like to thank Mr. W. M. Lee for his kind assistance in preparing the photographicplates. I am also grateful to Mr J. B. Harrison for assisting with the embedding and sectioningof the material studied.

REFERENCESALBERTINI, D. F. & ANDERSON, E. (1974). The appearance and structure of intercellular

connections during the ontogeny of the rabbit ovarian follicle with particular reference togap junctions. J. Cell Biol. 63, 234-250.

AMSTERDAM, A., BRATOSIN, S. & LINDNER, H. R. (1976). Gap junctions between deciduaJcells in the rat uterus. In Electron Microscopy 1976 (Proc. 6th Eur. Congr. Electron Microsc).Vol. 2. Biological Sciences (ed. Y. Ben-Shaul), pp. 372-373. Jerusalem: Tal Int. Publ.

BAKRWALD, R. J. (1975). Inverted gap and other cell junctions in cockroach hemocyte capsules:a thin section and freeze-fracture study. Tissue & Cell 7, 575-585.

BENEDETTI, E. L., DUNIA, I. & BLOEMENDAL, H. (1974). Development of junctions duringdifferentiation of lens fibers. Ptoc. natn. Acad. Sci. U.S.A. 71, 5073-5077.

DALLAI, R. (1975). Continuous and gap junction in the mid-gut of Collembola as revealed bylanthanum tracer and freeze-etching techniques, jf. submicrosc. Cytol. 7, 249-257.

DECKER, R. S. (1976a). Hormonal regulation of gap junction differentiation. J. Cell Biol. 69,660-685.

DECKER, R. S. (19766). Adrenocorticotropic hormone (ACTH)-induced formation of gapjunctions between differentiating Y-i tumor cells in vitro. J. Cell Biol. 70, 412a.

DECKER, R. S. & FRIEND, D. (1974). Assembly of gap junctions during amphibian neurulation.J. Cell Biol. 62, 32-47.

ELIAS, P. M. & FRIEND, D. S. (1976). Vitamin-A-induced mucous metaplasia. An in vitrosystem for modulating tight and gap junction differentiation. J. Cell Biol. 68, 173-188.

FAHRENBACH, W. H. (1976). The brain of the horseshoe crab {Limuluspolyphemus). 1. Neuroglia.Tissue & Cell 8, 395-410.

FILSHIE, B. K. & FLOWER, N. E. (1977). Junctional structures in Hydra. J. Cell Sci. 23,151-172.

FLOWER, N. E. (1971). Septate and gap junctions between the epithelial cells of an invertebrate,the mollusc Cominella maculosa. J. Ultrastruct. Res. 37, 259-268.

FLOWER, N. E. (1972). A new junctional structure in the epithelia of insects of the orderDictyoptera. J. Cell Sci. 10, 683-691.

FLOWER, N. E. (1977). Invertebrate gap junctions. J. Cell Sci. 25, 163-171.

304 N. J. Lane

GILULA, N. B. (1975). Junctional membrane structure. In The Nervous System (ed. D. B.Tower). Vol. 1. The Basic Neurosciences. New York: Raven Press.

GILULA, N. B. & SATIR, P. (1971). Septate and gap junctions in molluscan gill epithelium.J. Cell Biol. 51, 869-872.

HAND, A. R. & GOBEL, S. (1972). The structural organisation of the septate and gap junctionsof Hydra. J. Cell Biol. 52, 397-408.

HASTY, D. L. & HAY, E. D. (1977). Freeze-fracture studies of the developing cell surface.1. The plasmalemma of the corneal fibroblast. J. Cell Biol. 72, 667-686.

HUDSPETH, A. J. & REVEL, J. P. (1971). Coexistence of gap and septate junctions in an inverte-brate epithelium. J. Cell Biol. 50, 92-101.

ITO, S., VINSON, ].W. & MCGUIRE, T. ] . (1975). Murine typhus rickettsiae in the orientalrat flea. Ann. N.Y. Acad. Sci. 266, 35-60.

JOHNSON, R., HAMMER, M., SHERIDAN, J. & REVEL, J. P. (1974). Gap junction formationbetween reaggregated Novikoff hepatoma cells. Proc. natn. Acad. Sci. U.S.A. 71, 4536—454O-

JOHNSON, R. G., HERMAN, W. S. & PREUS, D. M. (1973). Homocellular and heterocellulargap junctions in Limulus: a thin section and freeze-fracture study, jf. Ultrastruct. Res. 43,298-312.

JOHNSON, G., QUICK, D., JOHNSON, R. & HERMAN, W. (1974). Influence of hormones ongap junctions in horseshoe crabs. J. Cell Biol. 63, 157A.

KENSLER, R. W., BRINK, P. & DEWEY, M. M. (1977). Nexus of frog ventricle. J. Cell Biol.73,768-781.

LANE, N. J. & HARRISON, J. B. (1977). A new type of continuous junction in Limtihis. J.Ultrastruct. Res. (in Press).

LANE, N. J., SKAER, H. LE B. & SWALES, L. S. (1975). Junctional complexes in insect nervoussystems. J. Cell Biol. 67, 233 a.

LANE, N. J., SKAER, H. LE B. & SWALES, L. S. (1977). Intercellular junctions in the centralnervous system of insects. J. Cell Sci. 26, 175-199.

LANE, N. J. & SWALES, L. S. (1976). Dynamics of cell junctions in the differentiating insectnervous system. J. Cell Biol. 70, 590 A.

LANE, N. J. & SWALES, L. S. (1977a). Differentiation of intercellular junctions in the CNSof developing insects. J. Cell Biol. 75, 56.4.

LANE, N. J. & SWALES, L. S. (19776). Changes in the blood-brain barrier of the centralnervous system in the blowfly during development, with special reference to the formationand disaggregation of gap and tight junctions. 1. Larval development. Devi Biol. 62.

LANE, N. J. & SWALES, L. S. (1977c). Changes in the blood-brain barrier of the central nervoussystem in the blowfly during development, with special reference to the formation and dis-aggregation of gap and tight junctions. II. Pupal development and adult flies. Devi Biol. 62.

LORBER, V. & RAYNS, D. G. (1972). Cellular junctions in the tunicate heart. J. Cell Sci. 10,211-227.

LORBER, V. & RAYNS, D. G. (1977). Fine structure of the gap junction in the tunicate heart.Cell Tiss. Res. 179, 169-175.

NOIROT-TIMOTHEE, C. & NOIROT, C. (1974). Les jonctions 'gap' chez les insectes. Structure,localisation, coexistence avec d'autres types de jonctions. J. Microscopie 20, 76 a.

PERACCHIA, C. (1973a). Low resistance junctions in crayfish. I. Two arrays of globules injunctional membranes. J. Cell Biol. 57, 54-65.

PERACCHIA, C. (19736). Low resistance junctions in crayfish. II. Structural details and furtherevidence for intercellular channels by freeze-fracture and negative staining. J. Cell Biol.57, 66-76.

PINTO DA SILVA, P. & GILULA, N. B. (1972). Gap junctions in normal and transformed fibro-blasts in culture. Expl Cell Res. 71, 393-401.

PRICAM, C , HUMBERT, F., PERRELET, A. & ORCI, L. (1974). Gap junctions in mesangial andlacis cells. J. Cell Biol. 63, 349-354.

RAVIOLA, E. & GILULA, N. B. (1973). Gap junctions between photoreceptor cells in thevertebrate retina. Proc. natn. Acad. Sci. U.S.A. 70, 1677-1681.

REVEL, J. P. (1974). Some aspects of cellular interactions in development. In The Cell Surfacein Development (ed. A. A. Moscona), pp. 51-56. New York: Wiley.

Formation of gap junctions in Limulus 305

REVEL, J. P., YIP, P. & CHANG, L. L. (1973). Cell junctions in the early chick embryo. Afreeze-etch study. Devi Biol. 35, 302-317.

SATIR, P. & FONG, I. (1972). Insect cell junctions. J. Cell Biol. 55, 227a.SATIR, P. & GILULA, N. B. (1973). The fine structure of membranes and intercellular com-

munication in insects. A. Rev. Ent. 18, 143-166.SIMIONESCU, M., SIMIONESCU, N. & PALADE, G. E. (1976). Segmental differentiations of cell

junctions in the vascular endothelium. J. Cell Biol. 68, 705-723.SKAER, H. LE B., BERRIDGE, M. J. & LEE, W. M. (1975). A freeze-fracture study of adult

Calliphora salivary glands. Tissue & Cell 7, 677-688.SKAER, H. LE B. & LANE, N. J. (1974). Junctional complexes, perineurial and glia-axonal

relationships and the ensheathing structures of the insect nervous system. A comparativestudy using conventional and freeze-cleavage techniques. Tissue & Cell 6, 695-718.

SMITH, D. S. (1968). Insect Cells: Their Structure and Function. Edinburgh: Oliver & Boyd.SMITH, D. S. & ALDRICH, H. C. (1971). Membrane systems of freeze-etched striated muscle.

Tissue & Cell 3, 261-281.STAEHELIN, L. A. (1974). Structure and function of intercellular junctions. Int. Rev. Cytol.

39, 191-283.WOOD, M. R., PFENNINCER, K. H. & COHEN, M. J. (1977). Two types of presynaptic con-

figurations in insect central synapses: an ultrastructural analysis. Brain Res. 130, 25-45.YEE, A. G. (1972). Gap junctions between hepatocytes in regenerating liver. J. Cell Biol.

55. 294a-{Received 2 December 1977)