Goblet Cell Membrane Differentiations in the Midgut of a Lepidopteran Larva

Jf. Cell Set. ao, 357-375 (1976) 357 Printed in Great Britain

GOBLET CELL MEMBRANE DIFFERENTIATIONS IN THE MIDGUT OF A LEPIDOPTERAN LARVA

N. E. FLOWER Physics and Engineering Laboratory, D.S.I.R., Private Bag, Lower Hutt, New Zealand AND B. K. FILSHIE Division of Entomology, C.S.I.R.O., P.O. Box 1700, Canberra, Australia

SUMMARY So-called goblet cells are present in the midgut of lepidopteran larvae. They are thought to be involved in the active transport of potassium out of the haemolymph and into the gut lumen. A number of plasma membrane differentiations within the goblet cell cavity has been investi- gated using conventional staining, lanthanum tracer and freeze-etch techniques. Of particular interest are junction-like inter- and intra-membrane differentiations found on the villus-like cytoplasmic projections present at the apical tip of the goblet cell cavities. These cytoplasmic projections appear to act as a valve; in some cases they seem to close off the top of the goblet cell cavity, so isolating it from the gut lumen, while in other cases they are spread apart leaving a wide channel from the cavity into the lumen. The junction-like structures on these cytoplasmic projections are different in structure from the septate-type junctions which seal the midgut cells together at their apical borders, and the 2 types are present on the same plasma membrane, often within one micron of each other. The need for a different type of junction may possibly be related to the fact that it occurs between 2 areas of the same plasma membrane. The morph- ology of this unusual junction-like structure is discussed and 2 diagrams are presented to illustrate our interpretation of its structure.

INTRODUCTION The midgut of plant-eating species of lepidopterous and tricopterous insects has been shown to be involved in the active transport of potassium out of the haemolymph and into the gut lumen, a function normally carried out in insects by the Malpighian tubules (Harvey & Nedergaard, 1964). The midgut of these insects contains a third cell type, the goblet cell, as well as the normal columnar and regenerative cells (Shinoda, 1927; Anderson & Harvey, 1966). Anderson & Harvey (1966) suggested that these goblet cells were responsible for the potassium transport. They also found that 3 different types of microvillus-like cytoplasmic projections were present in the goblet cell cavity. Towards the base of the cavity were large numbers of cytoplasmic projections, most of which contained mitochondria. They thus appeared very similar to the mitochondria-containing microvillus-like projections present on the lumen surface of cells in insect Malpighian tubules. Towards the apical end of the goblet cell cavity the cytoplasmic projections were shown to become smaller, not to contain mitochondria, and to be present in lesser numbers. At the tip of the cavity where it opens into the gut lumen, a number of projections, each further subdivided into many 358 N. E. Flower and B. K. Filshie villus-like units, were found. These seemed to act as a valve at the top of the cavity and in many sections appeared almost to close off the cavity from the gut lumen. In a previous paper one of the present authors had shown that in freeze-etch replicas of the light organ of Arachnocampa luminosa non-junctional complementary membrane fracture faces were observed, in the sense that the number of particles on the A faces were matched by the number of pits or holes on the B faces (Flower, 1973). This effect was only present in plasma membranes through which small molecules were being actively transported into the gland lumen. The present investigation was under- taken to see if such a membrane differentiation was also present in the goblet cell and to investigate the difference, if any, between the membranes of the different types of microvillus-like projections present in the goblet cell cavity.

MATERIALS AND METHODS The guts of caterpillars of the two lepidopteran species Ephestia kuhniella Zeller, Pyralidae, and Spodoptera litura (Fabricius), Noctuidae, were dissected and the midgut portions isolated. For conventional electron microscopy, material was fixed with 2-5 % glutaraldehyde in 0-05 M cacodylate buffer at pH 7-2 for 2 h at room temperature, washed in several changes of cacodylate buffer for 2 h, postfixed in 1 % osmium tetroxide for 2 h at room temperature, dehydrated, and embedded in Araldite. Thin sections were stained with uranyl acetate and lead citrate before examination. Some material was block stained in uranyl acetate for 12 h at 50 °C, after which sections were usually post-stained with lead citrate. Lanthanum incorporation into the tissue was achieved by the addition of either 1 % lanthanum nitrate or a lanthanum hydroxide colloidal suspension (Revel & Karnovsky, 1967) to both the glutaraldehyde fixative and the buffer washings. Sections of tissues with lanthanum incorporated were examined without further staining. For freeze etching, midguts were placed for 1 h either in 20-30 % glycerol buffered to pH 7-4 with 007 M phosphate or in this solution to which 2-3 % glutaraldehyde had been added. Small annulae were then cut out from the midgut and placed upright on copper grids before being rapidly frozen in Freon 12 held at —150 °C. The freeze etching was carried out in a Balzers BA500 as described by Moor & Miihlethaler (1963).

RESULTS The structure of goblet cells in the midgut epithelium of Lepidoptera has been described in detail for Hyalophora cecropia by Anderson & Harvey (1966) and in E. kuhniella by Smith et al. (1969). The main features are summarized in Fig. 1. The internal cavity of the cell occupies a considerable proportion of each cell volume, and large numbers of cytoplasmic processes project into the cavity. Most of these processes occur towards the base of the cavity and here they frequently contain long filamentous mitochondria within their cytoplasm. Towards the apex of the cavity these projections become sparse and do not contain mitochondria. At the apical tip, where the cavities open into the midgut lumen, are a number of somewhat different cytoplasmic projections, each of which divides into a number of small villus-like units. These projections appear to act as a valve, being open and not readily visible in some cases, and tightly packed into a regularly organized array so as to close off the cavity in others (Figs. 1, 4). With the conventional fixing and staining procedures we normally used, i.e. glutaraldehyde and osmium fixation followed by staining of sections with uranyl acetate and Lepidopteran membrane differentiations 359 lead citrate, a non-staining gap always existed between neighbouring projections (Fig. 4). This 'gap', which has a minimum width of about 11 nm, actually delineates a complex junctional structure which is most readily observed after lanthanum impregnation, as will be described later. Micrographs of goblet cells taken by B. L. Gupta and published in Berridge & Oschman (1972) show a dense, positively stained region between apposing apical projections, although details of the intermembrane structure cannot be discerned. Unfortunately the staining conditions were not recorded, and although we have been able to delineate the 'septa' by block staining with uranyl acetate followed by lead staining of sections (Fig. 5) we have been unable to obtain such dense staining.

Columnar cell Fig. 1. Diagrammatic representation of a goblet cell in the midgut of a mature lepidopteran larva. The goblet cell cavity occupies a large proportion of the cell volume. a si> g P junction; sj, septate-type junction; mvlt microvillus-like cytoplasmic projections at the basal end of the cavity, many of which contain long thin mitochondria; mvt, microvillus-like cytoplasmic projections towards the apical end of the goblet cell which do not appear to contain mitochondria; mv3, columnar cell microvilli lining the midgut; cp, cytoplasmic projections at the apical tip of the cavity which appear to act as a valve between the cavity and the midgut lumen.

When the membranes of the cytoplasmic projections near the base of the cavity are observed in freeze-etch replicas they show complementary fracture faces (Figs. 6, 7), in that the A faces are covered by a large number of particles and the B faces by an almost equivalent number of holes or depressions. As was evident from sectioning studies, these cytoplasmic projections, unlike microvilli, are not regular in shape, often bifurcating or bulging, the latter possibly occurring where mitochondria are enclosed. Measurement of a number of areas on both A and B faces has indicated that the particle density within the membrane is about 55 per ioo-nm square. The particles appear to be 10-12 nm in diameter. A somewhat different picture emerges when the fracture plane passes through the 360 N. E. Flower and B. K. Filshie cytoplasmic projections present at the apical tip of the cavity. These projections are more regularly arranged and appear to form a valve across the cavity which, as in sectioning studies, is sometimes seen closed (Fig. 8) and sometimes open (Fig. 9). Cytoplasmic filaments can be seen running along the length of some of the cytoplasmic projections (arrowed in Fig. 8). Examination of Figs. 8 and 9 shows that the membranes of some of the cytoplasmic projections are differentiated in that, depending on which face is examined, an array of grooves or an array of very closely packed rows of particles is present. The particles in these rows are so closely apposed that in many replicas they appear like continuous rods. The differentiations can be seen much more clearly in Fig. 10, where almost all the cytoplasmic projections are differentiated. The separation of the 'rods' or grooves is about 10 nm. Where the fracture plane has passed from within one membrane and into the membrane of an adjacent cytoplasmic projection in a suitable way (arrowed in Figs. 9 and 10), it can be seen that the differentiations on the 2 membrane faces appear junction-like, in that when present they always show on both faces. Thus a lower membrane face (face A) which has rods on it will be overlain by a grooved membrane face (face B), while undifferentiated areas of the lower membrane face will be overlain by an undifferentiated membrane face. This suggests that the membranes are differentiated only where in contact. This effect is shown especially clearly in Fig. 11, where a row of side-by-side cytoplasmic projections has been fractured at an angle. The rods and grooves appear to be present only on the membranes in regions where the projections abut each other. Within the cavity of the goblet cell and close to the cytoplasmic projections at the apical tip, areas of the plasma membrane are occasionally differentiated in a way very similar to the cytoplasmic projections at the apical tip of the cavity (Fig. 12). The rods and grooves in these regions are about 10 nm apart and appear to run in whorls around or between cytoplasmic projections. In some regions numbers of large particles are present, mostly in well defined groups (small arrow in Fig. 12). This differentiated plasma membrane does not seem to extend far basally into the cavity, the membrane structure quickly reverting to normal (arrowed in Fig. 12). Freeze-etch replicas of the membranes of the cytoplasmic projections at the top of the goblet cell cavity reveal only that there has been an internal change in the membrane structure. Sections of lanthanum-impregnated tissue show that the differentiations extend between the adjacent projections in a junction-like way. In sections where the cytoplasmic projections have been cut perpendicularly to their long axis, translucent bands can be seen between the membranes. These bands are similar to the appearance of septa in septate junctions (Fig. 13). In favourable circumstances it can be seen that each of these unstained septa actually consists of a pair of unstained septa. Two pairs are arrowed in Fig. 13. The spacing of the 2 translucent septa in each pair is about 4 nm and the centre-to-centre spacing between the pairs is about 9 nm. The separation of the 2 plasma membranes in these regions is about 10 nm. In tangential section the picture can be somewhat confusing as even small changes in the angle of section seem to change the appearance of the intermembranous material (Fig. 14). In many sections, however, small regions occur where twin rows of translucent particles or rods can be identified (arrowed in Fig. 14), the diameter of the rods being about 3 nm and their Lepidopteran membrane differentiations 361 separation along the rows about 7 nm. In these regions, careful examination shows that the rods in one of the twin rows are offset along the rows with respect to the rods in the other row. Furthermore, by examining the arrowed region in Fig. 14 at a glancing angle, lines of translucent rods can be identified at angles of approximately 420, 68°, 1120 and 1380 to the direction of the twin rows (arrowed), indicating a regular packing of the rods not only in the twin rows themselves but between the rods in neighbouring twin rows as well. The dimensions of the rods, diameter 3 nm and length 10 nm, probably explain why relatively little tilt of the section in any direction quickly leads to a very variable appearance.

DISCUSSION The goblet cells in both Ephestia and Spodoptera midguts are almost identical in general structure to those found by Anderson & Harvey (1966) in Cecropia midguts and summarized by them in an excellent diagram. Further studies on the goblet cells of lepidopteran midguts have been published by Smith et al. (1969) and Berridge & Oschman (1972). Thus near the base of the goblet cell cavity large numbers of cytoplasmic projections are present, many of which contain mitochondria. Towards the apical end of the cavity these projections become much fewer in number and cease to contain mitochondria. At the very tip of the cavity are a large number of somewhat different cytoplasmic projections which seem to function as a valve between the cavity and the midgut lumen. Among all the different non-junctional membranes present in the various midgut cells, complementary freeze-etch fracture faces have been found during the present study on plasma membranes only at the basal end of the goblet cell cavity. In this region a sufficient number of holes is present per unit area on B faces to match the number of particles per unit area on A faces. This is true of both the plasma membranes of the cytoplasmic projections and the surrounding cavity plasma membrane. Com- plementary fracture faces therefore exist in the region of the goblet cell cavity where mitochondria are present in large numbers both in juxtaposition to the cavity plasma membrane and within the cytoplasmic projections. Anderson & Harvey (1966) suggested that this mitochondria-rich region was where the potassium-secreting function of the cell occurred. Thus it is probable that this region showing complementary fracture faces can be equated with the region where potassium ions are actively pumped out of the cell through the plasma membrane and into the goblet cell cavity. Such complementary fracturing is not seen in most freeze-fractured membrane systems. However, similar complementary fracture faces of plasma membranes, where the A and B faces show approximately the same density of particles and depressions respectively, have been reported previously by 'Flower (1973) and Ito & Schofield (1974). In all these 3 cases where readily observed com- plementarity of the 2 fracture faces has been reported, the membrane has been involved in the secretion of small molecules out of the cell and into a gland lumen. In all 3 membrane systems, however, both the size of the particles present and their density 23 CEL 20 362 N. E. Flower and B. K. Filshie seem to be different. Thus Flower (1973) showed 12-15 nm particles with a density of only about 20 per 100-nm square in the light organ of the glow-worm Arachnocampa luminosa, while the particles in the freeze-etch replicas of mouse gastric parietal cells as measured from fig. 4 of Ito & Schofield (1974) appear to be about 8-10 nm in diameter and very densely packed at about 70 per 100-nm square. The size and density of the particles found in the present investigation lie between those reported earlier as they have a diameter of 10-12 nm and a density of about 55 per 100-nm square. Although the 50% difference in size between the largest and smallest types of these particles is too large for it not to be real to some extent, the variation in size between these 3 sets of particles should be treated with some suspicion, as it is difficult to estimate accurately the true size of any shadowed object, and in freeze-etching there is also the added possibility of distortion of the particles during the fracturing process. However, the inverse proportionality between particle size and density does indicate that there are significant differences between the 3 sets of particles, possibly associated with the different ions or molecules being secreted. Cytoplasmic filaments have been observed running along the length of some of the specialized cytoplasmic projections present at the apical tip of the cavity. They are also present in the normal microvilli of the midgut columnar cells. In contrast they have not been found in the cytoplasmic projections present within the cavity. It is possible that these filaments are present in the apical cytoplasmic projections as a structural support, required by the 'valve-like' function of the projections; a similar structural role has been suggested for the cytoplasmic filaments present in microvilli (Bonneville & Weinstock, 1970; Mukherjee & Staehelin, 1971). It is also possible that the filaments could have a contractile or motile role associated with the opening and closing of the valve. Such a contractile or motile function has been suggested for the filaments in microvilli (Boyd & Parsons, 1969; Thuneberg & Rostgaard, 1969; Tilney & Cardell, 1970; Mukherjee & Staehelin, 1971; Tilney & Mooseker, 1971). A structural basis for this function has been provided by the reports that the microfilaments in microvilli will bind heavy meromyosin, giving the arrowhead configuration typical of the latter's binding to actin, and have a major soluble protein with a molecular weight and net charge indistinguishable from actin as determined by gel electrophoresis (Ishikawa, Bischoff & Holtzer, 1969; Tilney & Mooseker, 1971). All this evidence indicates that the microfilaments in microvilli are either actin or a very similar protein. Microfilaments in other systems have also been shown to be actin-like, and in many of these systems to have an observable motile or contractile role (see for example Komnick, Stockem & Wohlfarth-Botterman, 1973; McNutt, Culp & Black, 1973; Reaven & Axline, 1973; Tilney, Hatano, Ishikawa & Mooseker, 1973; La Fountain,

The suggestion that the apical cytoplasmic projections could function as a valve was first made by Anderson & Harvey (1966). The present finding of junction-like structures on many of these projections reinforces and provides a more definite basis for this view. Furthermore, the membranes of the apical cytoplasmic projections are not differentiated in all replicas, which suggests that this may be a dynamic pheno- menon, since in general the presence or absence of the differentiations seems to depend Lepidopteran membrane differentiations 363 on whether the cavity is closed or open respectively. However, it is difficult to reach a conclusive view on this, mainly because of the impossibility of knowing the precise state of the valve. Thus valves apparently closed may either have just closed or be about to open, while valves apparently open may be either in the process of opening or about to close. This difficulty is illustrated by Figs. 8 and 9: the closed projections of Fig. 8 being little differentiated while the open projections of Fig. 9 are more differentiated. However, it can be said that differentiated membranes only seem to be present where the plasma membranes of neighbouring cytoplasmic projections are in contact.

Displacement

Twin row

^^ ^^ ^^ ^m ^^ t \ Fig. 2. Reconstruction of the appearance of tangentially sectioned lanthanum- impregnated junctions. This schematic diagram shows that if the rods are drawn with the observed spacings and each twin row of rods is displaced along its length by 3-5 nm then rows of rods can be identified at the same angles as observed in Fig. 14. These rows can best be identified by viewing along the arrows at a glancing angle to the page.

The schematic diagram, Fig. 2, demonstrates that if a lattice is drawn with the relative dimensions found between the rods in Fig. 14, then rows of rods can be seen at the observed angles if succeeding twin rows are displaced along their length by 3-5 nm, i.e. half the spacing of the rods along the rows. This diagram demonstrates that the rows of rods must be regularly packed over quite large areas of membrane within the junction to explain the appearance of lanthanum-impregnated junctions in sectioned tissue. A schematic diagram showing our interpretation of both the internal and intermembrane differentiations of the junction as visualized in sectioning and freeze-etch studies is presented in Fig. 3. The rows of particles and grooves observed in freeze-etch studies are equivalent to the twin rows of rods observed in lanthanum tracer studies. The rods or rows of particles observed on A faces in freeze-etch replicas may actually consist of twin rows of particles which are too close together for them

23-2 364 N. E. Flower and B. K. Filshie to be resolved by the freeze-etch technique, although no evidence exists for this at present. Sealing junctions between 2 different areas of the same plasma membrane have been reported previously. However, these cases have involved the same type of sealing junction as is normally present between different cells of the same tissue. Thus Stuart & Satir (1968) found septate junctions between 2 portions of the same plasma membrane in a cell surrounding a dendrite in an insect. Similarly, structures resembling

Face B Intermembrane

Face A

Fig. 3. Schematic diagram showing our interpretation of both the internal and intermembrane differentiations of the junction as visualized in sectioning and freeze-etch studies. The twin rows of rods observed in sections of lanthanum tracer-impregnated junctions are equivalent to single rows of particles or rods observed in freeze-etch replicas of the junctions. Twin rows, which are too close together to be resolved by the freeze-etch technique, may be present within the membranes, but no evidence for this has been found in the present study. tight junctions have been observed between 2 layers of the same myelin sheath surrounding a vertebrate nerve (Dermietzel, 1974; Mugnaini & Schnapp, 1974). As the smooth septate junctions which seal this tissue (Flower & Filshie, 1974), run right to the top of the goblet cell at its contact with the next cell, they come in places to within less than a micron from the cytoplasmic projections. It is therefore somewhat sur- prising to find that the intercellular sealing junction in this tissue is not used to seal the cytoplasmic projections at the cavity tip. In freeze-etch replicas, a region of differentiated plasma membrane is sometimes found just basal to the apical cytoplasmic projections but not extending very far into the goblet cell cavity. The intermembrane ridges and grooves which characterize these surfaces, although organized in a different pattern, are nevertheless very similar to the differentiations on the apical cytoplasmic projections. The membranes of these projections and the cavity just below are not both invariably differentiated, but when they are, the differentiations sometimes appear to be continuous with each other. Although membrane differentiation of the apical projections seems to be specifically associated with areas of membrane contact (i.e. sealing junctions) this is not true of the differentiations of the membrane lining the cavity just basal to these projections, Lepidopteran membrane differentiations 365 since the membrane in this area is never seen in close juxtaposition with any other area of membrane. It is difficult to speculate, therefore, on what special function the membrane differentiations in this area may perform.

We would like to thank Mrs S. M. Q'Kane and Miss M. Kovacs for technical assistance and Mr L. Marshall for the excellent diagrams.

REFERENCES ANDERSON, E. & HARVEY, W. R. (1966). Active transport by the Cecropia midgut. II. Fine structure of the midgut epithelium. J. Cell Biol. 31, 107-134. BERRIDGE, M. J. & OSCHMAN, J. L. (1972). Transporting Epithelia. New York and London: Academic Pres8. BONNEVILLE, M. A. & WEINSTOCK, M. (1970). Brush border development in the intestinal absorptive cells of Xenopus during metamorphosis. J. Cell Biol. 44, 151-171. BOYD, C. A. R. & PARSONS, D. S. (1969). The fine structure of the microvilli of isolated brush borders of intestinal epithelial cells. J. Cell Biol. 41, 646-651. DERMIETZEL, R. (1974). Junctions in the central nervous system of the cat. I. Membrane fusion in central myelin. Cell Tiss. Res. 148, 565-576. FLOWER, N. E. (1973). Complementary plasma membrane fracture faces in freeze-etch replicas. J. Cell Sci. 12, 445-452. FLOWER, N. E. & FILSHIE, B. K. (1974). Junctional structures in the midgut cells of lepidopteran caterpillars. J. Cell Sci. 17, 221-239. HARVEY, W. R. & NEDERGAARD, S. (1964). Sodium-independent active transport of potassium in the isolated midgut of the cecropia silkworm. Proc. natn. Acad. Sci. U.S.A. 51, 757-765. ISHIKAWA, H., BISCHOFF, R. & HOLTZER, H. (1969). Formation of arrowhead complexes with heavy meromyosin in a variety of cell types. J. Cell Biol. 43, 312-328. ITO, S. & SCHOFIELD, G. C. (1974). Studies on the depletion and accumulation of microvilli and changes in the tubulovesicular compartment of mouse parietal cells in relation to gastric acid secretion. J. Cell Biol. 63, 364-382. KOMNICK, H., STOCKEM, W. & WOHLFARTH-BOTTERMAN, K. E. (1973). Cell motility: mech- anisms in protoplasmic streaming and ameboid movement. Int. Rev. Cytol. 34, 169—249. LA FOUNTAIN, J. R. (1974). Microfilaments in spermatocytes of Nephrotomi suturalis. J. Cell Biol. 6o, 784-789. MCNUTT, N. S., CULP, L. A. & BLACK, P. H. (1973). Contact-inhibited revertant cell lines isolated from SV4O-transformed cells. IV. Microfilament distribution and cell shape in untransformed, transformed and revertant Balb/c 3T3 cells. J. Cell Biol. 56, 412-428. MOOR, H. & MOHLETHALER, K. (1963). Fine structure in frozen etched yeast cells. J. Cell Biol. 17, 609-628. MUGNAINI, E. & SCHNAPP, B. (1974). Possible role of zonula occludens of the myelin sheath in demyelinating conditions. Nature, Lond. 251, 725-727. MUKHERJEE, T. M. & STAEHELIN, L. A. (1971). The fine-structural organization of the brush border of intestinal epithelial cells. J. Cell Sci. 8, 573-599. REAVEN, J. E. P. & AXLINE, S. G. (1973). Subplasmalemmal microfilaments and microtubules in resting and phagocytozing cultivated macrophages.^. Cell Biol. 59, 12-27. REVEL, J. P. & KARNOVSKY, M. J. (1967). Hexagonal array of subunits in intercellular junctions of the mouse heart and liver. J. Cell Biol. 33, C7-C12. SHINODA, O. (1927). Contributions to the knowledge of intestinal secretion in insects. II. A comparative histocytology of the mid-intestine in various orders of insects. Z. Zellforsch. mikrosk. Anat. 5, 278-292. SMITH, D. S., COMPHER, K., JANNERS, M., LIPTON, C. & WITTLE, L. W. (1969). Cellular organisation and ferritin uptake in the mid-gut epithelium of a moth Ephestia kiihniella. jf. Morph. 127, 41-72. STUART, A. M. & SATIR, P. (1968). Morphological and functional aspects of an insect epidermal gland. J. Cell Biol. 36, 527-549. 366 N. E. Flower and B. K. Fihhie THUNEBERG, L. & ROSTGAARD, J. (1969). Motility of microvilli. Fedn Eur. biochem. Soc. Symp., Madrid (ed. T. Goodwin), p. 270. London: Academic Press. TILNEY, L. G. & CARDELL, R. R. (1970). Factors controlling the reassembly of the microvillus border of the small intestine of the salamander. J. Cell Biol. 47, 408-422. TILNEY, L. G., HATANO, S., ISHIKAWA, H. & MOOSEKER, M. S. (1973). The polymerization of actin: its role in the generation of the acrosomal process of certain echinoderm sperm. J. Cell Biol. 59, 109-126. TILNEY, L. G. & MOOSEKER, M. (1971). Actin in the brush-border of epithelial cells of the chicken intestine. Proc. natn. Acad. Sci. U.S.A. 68, 2611-2615.

(Received 17 March 1975 - Revised 10 September 1975)

Fig. 4. Section through the apex of a goblet cell showing the specialized apical cytoplasmic projections {cp). Projections from opposing sides of the cell come close together to seal off the cell cavity from the gut lumen (#/)• A non-staining gap is present between the apposing membranes of the projections. Septate junctions (.*/) are present between cells. The section was stained with both uranyl and lead salts, x 16000. Fig. 5. Section through the apex of a goblet cell showing the specialized apical cytoplasmic projections (cp). The tissue was block stained with uranyl acetate and the section stained with lead citrate. 'Septa' can be identified (arrows) between some of the apical cytoplasmic projections, x 8000c. Fig. 6. Freeze-etch replica of cytoplasmic projections within a goblet cell cavity and near its basal end. A faces (a) are covered in a dense array of particles. B faces (b) have approximately the correct number of holes per unit area to match the particles on A faces. These large, bulbous cytoplasmic projections are typical of those containing mitochondria, xnoooo. Lepidopteran membrane differentiations 368 N. E. Flower and B. K. Filshie

Fig. 7. Another freeze-etch replica of cytoplasmic projections near the basal end of a cavity. Although these projections are more like normal microvilli the number of holes per unit area on the B faces (b) still approximately matches the number of particles per unit area on A faces (a), x 110000. Fig. 8. Freeze-etch replica showing a fracture across the apical cytoplasmic projections of a goblet cell. The 2 sets of projections abut each other, suggesting that the cavity is effectively sealed off from the gut lumen. Some differentiation of the membranes of these projections is apparent. Rows of clostly spaced particles are visible on A faces and corresponding grooves on B faces. Cytoplasmic filaments (arrowed) can be seen running along the length of some of the projections which have been cross-fractured, x 53000. Lepidopteran membrane differentiations 369 370 N. E. Flower and B. K. Filshie

Fig. 9. Another freeze-etch replica showing a fracture across the apical cytoplasmic projections of a goblet cell. In this case the 2 sets of projections do not abut, suggesting that the cavity is open to the gut lumen. Again many of the projections are differentiated. Where overlaying and abutting projections have been fractured so as to reveal membrane faces from both projections (arrowed and inset), the grooves and rows of particles on the 2 faces appear to complement each other, x 30000. Inset, x 55000. Fig. 10. Another freeze-etch replica showing a fracture across the apical cytoplasmic projections of a goblet cell. The differentiations on the 2 membrane faces can be seen more clearly in this micrograph. Once again the grooves and rows of particles on the 2 faces appear to complement each other (arrowed and inset), x 53000. Inset, x 100000. Lepidopteran membrane differentiations 371 372 N.rE. Flower and B. K. Filshie

Fig. II. Freeze-etch replica of a more longitudinal fracture through the apical cytoplasmic projections of a goblet cell. A single row of abutting projections (arrowed) can be seen, the membranes of which only seem to be differentiated where in contact, x 53000. Fig. 12. Freeze-etch replica showing an area of plasma membrane within the goblet cell cavity and close to the apical cytoplasmic projections. This membrane is differentiated in a very similar way to the apical cytoplasmic projections. The rods and grooves on this A face run in whorls around or between cytoplasmic projections. In some regions, numbers of large particles are present between the whorls, mostly in well defined groups (small arrows). Such differentiated membrane does not normally seem to extend far basally from the apical cytoplasmic projections. The line along which the membrane reverts to more typical plasma membrane is indicated by the 2 large arrows, x 53000. Lepidopteran membrane differentiations 374 N. E. Flower and B. K. Fikhie

Fig. 13. Lanthanum-stained apical cytoplasmic projections. Section plane is perpen- dicular to the long axis of the projections and membranes are cut transversely. In favourable areas (arrows) the electron-dense lanthanum precipitate delineates double 'septa'. Section is unstained apart from lanthanum. X 300000. Fig. 14. Similar preparation to Fig. 13, but the plane of section is parallel to the long axis of the cytoplasmic projections so that junctions are frequently cut tangentially. The 'septa' of Fig. 13 can be seen to be composed of double rows of small rod-shaped particles (arrow), x 240000. Lepidopteran membrane differentiations 375