Synchronous Exocytosis in Paramecium Cells. Vi. Ultrastructural Analysis of Membrane Resealing and Retrieval

J.CellSd. 77,1-17(1985) Printed in Great Britain © Company of Biologists Limited 1985

SYNCHRONOUS EXOCYTOSIS IN PARAMECIUM CELLS. VI. ULTRASTRUCTURAL ANALYSIS OF MEMBRANE RESEALING AND RETRIEVAL

H. PLATTNER*. R. PAPE, B. HAACKE, K. OLBRICHT, C. WESTPHAL AND H. KERSKEN Faculty of Biology, University ofKonstanz, P.OJi. 5560, D-7750 Konstanz, Federal Republic of Germany

SUMMARY After the synchronous induction of exocytosis of secretory organelles (trichocysts) in Paramecium tetraurelia cells the process of membrane resealing and retrieval could be followed under synchronous conditions. The characteristic aggregates of membrane intercalated particles (MIPs) contained within the freeze-fractured cell membrane (rings and rosettes) and trichocyst membranes (annulus MIPs), in addition to collar striations on the top of trichocyst membranes, served as endogenous ultrastructural markers. This allowed us to follow the re-arrangement of membrane constituents during and after exocytosis with high temporal and spatial precision. Membrane specificity is maintained to a considerable extent (~ 99-5 %), as judged from the rare occurrence of aberrant resealing (according to freeze-fracture data) and from the rather minute shift of glycocalyx components (according to electron staining experiments) during normal membrane resealing. Coated pits are not involved in membrane retrieval (155 ghosts analysed); since the membrane regions involved in exocytotic fusion are backed by apposed materials, probably proteins, this may restrain membrane constituents from intermixing. Another factor for maintaining membrane specificity is the fact that resealing of the exocytotic opening occurs much more rapidly than in most other systems. The retrieval operates with a half-life of 3 (strain 75) to 9min (K401); the involvement of cortical microtubules in the retrieval can be largely excluded, since only two microtubules (of unidentified origin) were seen to approach ghost structures in 4074 cases analysed during this period of intense ghost retrieval. Phalloidtn microinjected at a dose that blocked all cytoplasmic streaming (before synchronous exocytosis was induced) did not abolish membrane resealing and retrieval, which, therefore, may be passive processes.

INTRODUCTION In a variety of cell types exocytosis is followed by retrieval of the secretory organelle membrane, which in turn may undergo recycling via elements of the Golgi apparatus (for reviews, see Farquhar, 1981; Holtzman, 1981). Nevertheless, important aspects of what occurs immediately after the release of secretory contents still await further elucidation. In particular, as recently summarized by Farquhar (1981), the following questions remain to be clarified. (1) What is the precise nature of the membrane removed from the cell surface? (2) What is the nature of the carrier, and are coated vesicles involved? (3) Are there stable, non-perturbing labels available for tracing the retrieval process?

•Author for correspondence. Key words: exocytosis, freeze-fracture, membranes, Paramecium, secretion. 2 H. Plattner and others Although the answers to these questions may be different for different cell types, they appear to be particularly easy to approach in the case of our model system, the ciliated protozoon Paramecium tetraurelia. Each Paramecium cell has more than 1000 secretory organelles (trichocysts), most of which are dischargeable within one or perhaps several seconds by polyamine trigger compounds (Plattner et al. 1984; Plattner, Sturzl & Matt, 1985). This process entails exocytotic membrane fusion, resealing (Olbricht, Plattner & Matt, 1984) and subsequent retrieval (Haacke & Plattner, 1984). Distinct structural elements are present in the fusogenic zones of the cell membrane (Janisch, 1972; Plattner, Miller & Bachmann, 1973) and in or on the trichocyst tip membrane (Allen & Hausmann, 1976; Beisson et al. 1976; Hausmann & Allen, 1976; Plattner et al. 1973), which can be used as endogenous markers. These structural elements are in the form of: (1) membrane-intercalated particles (MIPs) visible after freeze-fracturing, and (2) membrane-associated materials (Westphal & Plattner, 1981a, b) seen in ultrathin sections, when tannic acid is applied as a mordant dye according to Fujiwara & Linck (1982). In contrast, most other exocytotic systems require exogenous markers (e.g. antibodies, lectins or affinity labels) for tracing the fate of certain membrane areas, and these tags may well influence the processes to be analysed (Farquhar, 1981; Helenius & Mellman, 1983; Holtzman, 1981). In the Paramecium cell the resealing of ghost membranes after trichocyst discharge has been described previously by us (Haacke & Plattner, 1984; Olbricht et al. 1984) and others (Allen & Hausmann, 1976; Hausmann & Allen, 1976). However, in none of these publications has the degree of specificity of membrane separation or any possible involvement of clathrin coats, microtubules or microfilaments been analysed in any detail. The present paper sets out to examine these aspects.

MATERIALS AND METHODS Cell cultures Paramecium tetraurelia, strains 7S or K401, which are both equipped with a normal exocytotic apparatus (Plattner et al. 1980) were used as monoxenic cultures in a dry lettuce medium inoculated with Enterobacter aemgenes. 7S cells grew within 3 days, and K401 cells within 4 days, to early stationary phase. Such cells were triggered for trichocyst release with aminoethyldextran (AED) in the presence of exogenous Ca2+ as indicated before (Plattner et al. 1984, 1985).

Electron microscopy For ultrathin sectioning cells were fixed in different ways. For standard preparations we used glutaraldehyde, followed by osmium tetroxide as described previously (Plattneret al. 1984); tannic acid impregnation after saponin permeabilization was according to Haacke & Plattner (1984). In some experiments exogenous Ca^+ was chelated by adding 4mM-EGTA. In other experiments, living cells were exposed for 10s to cationized ferritin (Miles-Yeda, Rehovot, Israel), which could be used as a surface label and simultaneously - like other polycationic compounds (Plattner et al. 1985) - as a trigger agent. Routinely, cell pellets were dehydrated in acetone and embedded in Spurr's (1969) resin. After polymerization at 344 K, ultrathin sections of ~70nm thickness were made and double-stained with uranyl acetate and alkaline lead citrate. Some sections were subjected to polysaccharide staining according to Farragiana & Marinozzi (1979). For an analysis of membrane dynamics cells were quickly frozen in the living state as indicated by Olbricht et al. (1984). A Balzers BAF 300 freeze-fracture machine was used. The designation of fracture faces was as indicated by Plattner & Zingsheim (1983). Membrane reseating and retrieval 3 In addition, 10 cells were microinjected with rhodaminylated phalloidin (a gift from Th. Wieland) under a Zeiss I CM 405 inverted microscope with fluorescence optics and a video-enhanced image-intensifying system, as specified in more detail elsewhere (Kersken, Vilmart-Seuwen, Momayezi & Plattner, unpublished). Phalloidin doses were deliberately chosen high enough (up to 50 ^g/ml cell volume) for any cell motility phenomena to be totally blocked; only then (mostly after 30 min) were cells triggered with AED, and 10 s or 10 min afterwards they were fixed with glutaraldehyde, supplemented with EGTA, tannic acid and saponin (see above). Individual cells were then processed to produce ultrathin sections as outlined above. Sections and replicas were analysed in a Zeiss EM 10 electon microscope (80kV). The different structures presented in Tables 1 and 2 were documented and counted in a random sampling procedure.

RESULTS For the normal ultrastructure of trichocyst docking sites, we refer to Plattner et al. (1973) and Westphal & Plattner (1981a, b). A summary of the salient features is given in the scheme at the end of this paper. For the normal course of ultrastructural changes during and after AED-induced sychronous trichocyst release, concerning aspects not covered in this paper, we also refer to previous work (Haacke & Plattner, 1984; Olbricht et al. 1984; Pape & Platt- ner, 1985). Freeze-fracturing revealed that exocytosis involves the dispersal of rosette MIPs from within the rings to outside the fusion zone, not before but precisely during the formation of the exocytotic opening (Olbricht et al. 1984). Elements of the trichocyst tip membrane, the annulus MIPs and collar striations, become only tran- siently visible from the outside (Figs 1,2); when normal membrane detachment and

Figs 1—4. Resealing stages obtained from AED-triggered freeze-fractured, unfixed cells. Fig. 1 is a PF view of an exocytotic opening; Fig. 2 is a somewhat later stage; an, annulus; co, collar; 'RI presumably indicates the ring, which is difficult to identify due to the expansion of the exocytotic site during discharge, while trichocyst and cell membrane are fused perfectly with each other. Fig. 3 is interpreted as an early {oval ring) and Fig. 4 as a late resealed stage (parenthesis). Bars, 0-1 /an. 4 H. Planner and others

Table 1. Freeze-fracture appearance ofexocytosis sites at different times after triggering ofexocytosis

Time

Stage 0 Is 1 min

Resting stage 71-4 56-7 5-3 (ring + rosette) Oval ring* 1-6 4-2 3-6 (early resealed stage) Parenthesis] 18-0 16-8 82-2 (late resealed stage) Other stagesj 9-0 22-3 8-9

Number of docking sites analysed (n) 1656 978 225

The presence of methylcellulose retarded exocytotic processes in these freeze-fracture samples. Final stages (as indicated here for 1 min after AED addition) may therefore be formed much earlier, but no precise time sequence study was made in this context. Values are % (all stages from all sites = 100%). • Ring shape between a ring of resting stage and a parenthesis and without rosette particles. •f Final stage of membrane resealing, which is maintained until a new trichocyst is docked. j Comprises undergoing membrane fusion processes as well as stages that were unidentifiable (4—8%), e.g. due to technical reasons.

resealing subsequently take place, these structures become clearly detached with the trichocyst ghost membrane. Plasma membrane resealing involves the formation first of an oval ring stage (Fig. 3) and then of a parenthesis stage (Fig. 4), as defined by Beisson et al. (1976). This sequence is also supported by the quantitative data presented in Table 1, although the values indicated for 1 min are probably attained much earlier. (No precise time sequence studies were done in this context.) The dispersal of rosette MIPs (within the cell membrane) and the recovery of trichocyst membrane elements (with the ghosts) are documented in Figs 5 and 6. The number of MIPs composing a ring (which delineates the actual fusion zone) is 69 ± 5 (n = 25) and remains unchanged when a ring is transformed to an oval ring and subsequently to a parenthesis stage after exocytosis. How frequently do exceptions occur from this extremely precise membrane retrieval? In only one out of 178 cells analysed by freeze-fracturing, did we observe that elements of the trichocyst tip membrane were incorporated permanently into the resealed cell membrane (Figs 7—9, for atypical resealing, are all from the same cell). This indicates a very small failure rate of the membrane resealing process. Figs 10—16 are from ultrathin sections. During the normal resealing process the whole trichocyst membrane, including its tip region with the collar, is detached. The lower portion then collapses and vesiculates (Figs 11-13). While the ghost as a whole Membrane reseating and retrieval

Figs 5,6. EF (Fig. 5) and PF (Fig. 6) views of a trichocyst ghost. Unfixed preparations. In Fig. 5 the trichocyst membrane (tm) is still in contact with the cell membrane (cm), whereas it is detached in Fig. 6. Note the dispersal of, presumably, rosette MIPs (arrowheads) in Fig. 5 and the presence of annulus (an) and collar (co) elements in Fig. 6. Bars, 0-1 /tfn.

Figs 7-9. Aberrant resealing, PFview, as found in one out of the 178 cells (unfixed). After AED triggering annulus (an) and collar (co) elements became integrated into the cell membrane. Bars, 0-1 [am. collapses, cytoplasm is intruding all the way up to the tip into the pocket thus formed (Fig. 12). The collar is the last structure to vesiculate (Fig. 15). Are clathrin coatings, microtubules or microfilaments involved in membrane retrieval? We never observed coated pits or vesicles (Figs 10—16), though we analysed 155 ghosts in great detail (Table 2). This holds also for microtubules (Table 2), which were found only very rarely in the proximity of a ghost vesicle (precisely, in only two cases, in which their origin could not be traced in consecutive sections; 4074 ghosts analysed). One could also have expected that cortical actin filaments (Plattner et al. H. Planner and others

Figs 10—16. Morphogenetic series of membrane retrieval steps, as found within 5min after AED-triggered exocytosis. al, alveolar cavities. The ghost of the trichocyst membrane (tm) becomes detached very rapidly from the cell membrane (cm) and collapses. Its basal region vesiculates first (Figs 11-13); finally, the rest (Figs 14-16), representing the collar region, also vesiculates. Bars, 0-1 /tfn. Membrane reseating and retrieval f

Table 2. Absence ofclathrin coats or microtubules from trichocyst ghosts at different times (t) after exocytosis triggering

Microtubules*

Clathrin coat Longitudinal sections Cross-sections

{ JVt nX N\ «X JVf nX i 0 10 10 0 156 285 0 102 739 0 10 s 7 12 0 36 135 1 15 s 165 689 0 20 s 11 60 0 22 174 1 30 s 129 932 0 1 min 5 22 0 34 139 0 5 min 10 12 0 53 58 0 148 775 0 15 min 3 15 0 46 55 0 30 min 5 18 0 44 87 0 lh 8 6 0 43 6 0 Sum 59 155 0 434 939 544 3135 0

Only a fraction of the ghost population was analysed for the individual characteristics. The values do not represent the percentages of ghosts occurring at the different times. • Longitudinal or cross-sections refer to the orientation of docked trichocysts. f Number of cell sections analysed for this particular aspect. j Number of ghosts analysed for this particular aspect. § Number of ghost profiles with clathrin coat (gc) or with microtubules (gm) approaching a ghost.

Fig. 17. Two trichocyst docking sites (A, B) from a cell that was microinjected with phalloidin (50 ^g/ml); after complete termination of cytoplasmic streaming (30 min after injection) the cell was triggered with AED and fixed 15 min later. Ghosts, which are cleared from docking sites in control cells of strain 7S with a half-life of 3 min (Pape & Plattner, 1985), are no longer present in this 'internally immobilized' cell. Small vesicles (u) could possibly be minor remnants of ghosts. Bar, 1 fim. 8 H. Planner and others 1982) could possibly be involved in exocytosis or in the steps thereafter, yet no evidence along these lines was found in the present work. Upon massive trichocyst expulsion cells contract visibly; but this occurred also after microinjection of a very high dose of phalloidin. Is cytoplasmic streaming responsible for the clearing of ghosts? Again this does not seem to be the case. Full immobilization of cytoplasmic bulk transport processes by high phalloidin doses impeded neither trichocyst discharge nor ghost retrieval (Fig. 17), which implies that membrane resealing is also not impaired. In fact, exocytosis sites were resealed at least IS s after AED addition and most ghosts were retrieved 15 min after AED triggering, when the cells had been microinjected with phalloidin 30min before adding AED. The observed cell contraction could thus be a passive phenomenon due to the large loss of volume arising when the majority of trichocysts (representing 18% of the cell volume; Pape & Plattner, 1985) are expelled. Since shrinkage would be predominantly along the long axis of the cell, this would explain the direction of collapse of the ghosts (see below), which in turn could account for the

Figs 18—21. In vivo labelling with cationized ferritin induces exocytosis of trichocysts at many (Figs 19, 20) but not all (Fig. 18) docking sites. Resealing must be quick, as judged from the absence (Fig. 19) or scarcity (Fig. 20) of label within the ghosts, whereas parasomal sacs (coated pits on ciliary bases, i.e. remote from exocytosis sites) are always labelled (Fig. 21). Note the accumulation of ferritin label just over the extrusion site, indicating flow of centripetal material during resealing. Bars, 0-1 /an. Membrane resealing and retrieval f rapid resealing. Once detached, ghosts evidently diffuse from the cortex into deeper regions of the cytoplasm. From these results it becomes evident that membrane resealing and retrieval could take place - at least under our experimental conditions — as merely passive processes. Finally, we tried by ultrastructural cytochemistry to find some additional evidence for selectivity, speed and direction of the movement of membrane components during resealing. The centrifugal dispersal oi rosette MIPs (see above) does not preclude the possibility that other membrane constituents could move in the opposite direction.

Figs 22, 23. Fixed, embedded, sectioned and phosphotungstic acid-stained trichocyst site before and after spontaneous exocytosis. The glycocalyx, the rims of the epiplasm (ep; indicating that some proteins are also stained) and some (glycogen?) aggregates in the cytoplasm are stained. The membrane of the resting trichocyst (tm) (Fig. 22) is unstained, whereas the resealed ghost displays some staining within its uppermost region (down to pointers) due to a centripetal shift of glycocalyx elements. Bars, 0-1 /im. I .'Annulus I Resealing 4- , CoM , I +4- CM : C o0 I0 Resting stage Exocytosis

Fig. 24. Sequence of events derived from Figs 1-23; for more details see the text. The resting atage is characterized by the occurrence of a ring of MIPS (encircling the exocytotic fusion zone) and an aggregate of rosette MIPs in the centre; the trichocyat tip is covered by a collar (atta~hed~roteins) and 'connecting material' ties it to the cell membrane. Upper row, top view; lower row, lateral view. AN, annulrcs; CM, cell membrane; CO, collar; CoM, connecting maten'al;MIPs, membrane-intercalated particlea; TM, trichocyst membrane. Arrows indicate the direction of flow of membrane components. Membrane reseating and retrieval 11 When the surface of living cells was labelled with an excess of polycationized ferritin, this acted simultaneously as a trigger for exocytosis; the resulting ghosts contained very little ferritin (Figs 19, 20). (Parasomal sacs as shown in Fig. 21 were regularly labelled and thus represented a sort of internal control.) This indicates again that resealing must take place very rapidly. Does it also mean that resealing occurs without any substantial incorporation of cell membrane elements into the trichocyst ghost? Glycocalyx staining with aqueous phosphotungstic acid applied to ultrathin sections (Figs 22, 23) appeared to be a more suitable approach to this question, since by this method one could also pinpoint very mobile components, such as glycolipids. Indeed, a very restricted uppermost zone, representing hardly more than 0-5 % of the total estimated amount of ghost membrane, was stained (Fig. 23); this region was not reactive in the resting trichocyst (Fig. 22). The accumulation of ferritin label on the resealing site (Figs 19, 20) is also an indication of a centripetal movement of some membrane constituents during resealing. It is not as yet clear how the different behaviour of these components with regard to the opposite movement of rosette MIPs is brought about. A possible clue to this problem could be that rosettes are anchored by connecting material (Plattner, et al. 1980; Westphal & Plattner, 1981a, b) in the cell membrane above the trichocyst tip and that this connection can no longer be recognized over resealed ghosts (Fig. 12).The centrifugal movement of rosette MIPs could thus be guided by such underlying connecting material, whereas the elements depicted in Fig. 23 could undergo free lateral diffusion from the cell membrane into the tip of the ghost at the same time. There is no evidence for the movement of elements of the trichocyst membrane into the cell membrane (for exceptions see above); this could be accounted for by materials {collar) that remain attached to the cytoplasmic side of the trichocyst membrane even after exocytosis (Figs 10—16). Our finding are summerized in Fig. 24. Marginal observations pertinent to this subject are as follows. One can clearly see the lateral compression of ghosts perpendicular to the cell axis in Fig. 14 (see also Allen & Hausmann, 1976), whereas for unknown reasons parentheses are always compressed in the other direction. An additional observation is the occurrence of ghosts in 'reserve' docking positions, i.e. on the corners of the regular fields ('kineties') of the cell surface relief, while regular docking sites are located in the middle of perpendicular ridges. Similar situations can also be seen in fig. 6 of Hausmann & Allen (1976), and suggest that trichocysts installed at these sites are extrudable (see Pape & Plattner, 1985, for more details).

DISCUSSION General considerations Exocytosis is generally coupled to endocytotic membrane retrieval for coping with the requirements of membrane economy and specificity. In paramecia we could take advantage of endogenous markers to determine the degree of precision of membrane retrieval. Therefore, we did not have to be concer- ned about the problems inherent in the use of exogenous labels (see Introduction) or 12 H. Plattner and others about contamination, as may occur in cell-fractionation methods (Castle & Palade, 1978). Thus, we assume that our data on the precision of membrane separation after exocytosis in paramecia are as precise as possible. In their analyses of Tetrahymena, Satir, Schooley & Satir (1973) also found that rosette MIPs are dispersed during exocytosis, but they assumed that mucocyst membranes would be permanently incorporated into the cell membrane (Satir, 1974a, b); the opposite was shown for Cyclidium mucocysts (Bardele, 1983), which agrees with our own results with paramecia. For Paramecium the occurence of membrane retrieval was inferred from the occasional occurrence of ghost membranes in untriggered cells (Pitelka, 1965), as well as from the abundance of ghosts after massive triggering of exocytosis by different means (Haacke & Plattner, 1984; Hausmann & Allen, 1976; Plattner, 1976; Plattner et al. 1973). Very recently, Allen & Fok (1984) have also demonstrated that ghost fragments retrieved in the presence of peroxidase as a marker can be detected in close vicinity to the Golgi apparatus.

Mode and speed of resealing The resealing stages that we found, i.e. oval rings and parentheses (Figs 3, 4), are in agreement with previous results from experiments with Ca2+ ionophore-triggered trichocyst expulsion (Plattner, 1974, 1976), where we obtained predominantly oval ring stages. These cannot be due to docking of new trichocysts, since docking occurs only at a later stage after synchronous exocytosis (Pape & Plattner, 1985). Under synchronous conditions we found that these stages are transformed to parentheses (Table 1). Using chemical fixation we estimated that the resealing process would take no longer than seconds (Haacke & Plattner, 1984; Plattner et al. 1984). The actual resealing time could be determined only recently using electrophysiological methods (Deitmer & Plattner, unpublished data); it is 1 s for a whole cell and fractions thereof for the single event.

Fate of membrane-integrated particles (MIPs) and involvement of membrane- associated proteins MIPs are generally considered predominantly as equivalents of membrane- integrated proteins (Plattner & Zingsheim, 1983; the acronym MIP takes into account this dual meaning). The structures that are used here as endogenous markers were also shown to be sensitive to proteolytic enzymes (Vilmart & Plattner, 1982, 1983). In paramecia these MIP aggregates and the collar fibres also appear to be held in place by tannic acid-stainable membrane-associated proteins (Westphal & Plattner, 1981a, b). We have recently shown that during exocytotic membrane fusion, rosette MIPs are laterally dispersed within the cell membrane to the extent that the exocytotic opening expands (Olbrichtef a/. 1984). Fig. 5 shows that rosette MIPs are indeed not shifted down into the trichocyst membrane. Conversely, annular MIPs and collar elements of the trichocyst membrane are not normally incorporated into the cell membrane when a ghost is detached (Fig. 6). Although in paramecia exocytosis—endocytosis coupling was never seen to involve coated pits or vesicles (Figs 10-16), these cells do form such structures at other sites Membrane resealing and retrieval 13 (parasomal sacs; Pitelka, 1965). Other systems perform membrane retrieval with coated or smooth endocytotic profiles (Farquhar, 1981; Holtzmann, 1981; Trifar6 & Poisner, 1982), or with both at the same time (Miller & Heuser, 1984). In general, the mode of retrieval appears not to depend upon the speed of resealing or retrieval, although it is tempting to speculate that the extremely rapid resealing process in paramecia would not need a clathrin coating. One could also argue that this would even impede its formation, since internalized secretory organelle membranes no longer keep their clathrin coat attached (Willingham & Pastan, 1984). Another reason would be that the entire trichocyst tip displays a thick backing by membrane- associated proteins (Westphal & Plattner, 1981a, b), which could restrain membrane components from any significant intermixing. DeCamilli, Peluchetti & Meldolesi (1976) have also reported that in parotid acinar cells MIP populations do not intermix; yet they could use only the relative MIP density as a criterion. We can demonstrate here that the failure rate of resealing is only ~ 0-5 % (1) by the low frequency of aberrant resealing (Figs 7—9) and (2) by the low extent of carbohydrate staining on ghost membranes (Figs 22, 23). The latter most probably indicates a limited amount of centripetal movement of very fluid membrane constituents, perhaps glycolipids.

Clearing of ghost membranes The retrieval speed is also rather high in paramecia. For cells of strain 75 a half-life of 3 min, icxK401 9min was determined (Pape & Plattner, 1985; Haacke & Plattner, 1984). Similarly, Hausmann & Allen (1976) reported a retrieval period of 5-10 min for electrically triggered Paramecium caudatum and within this time-frame recycling vesicles were later observed near the Golgi region (Allen & Fok, 1984). Other systems behave in a quite different way. In nerve terminals resealing and retrieval take both ~ 1 min (Miller & Heuser, 1984). In adrenal chromaffin cells resealing alone takes ~ 30 min (Dowd et al. 1983; Lingg, Fischer-Colbrie, Schmidt & Winkler, 1983; Phillips, Burridge, Wilson & Kirshner, 1983). In parotid acinar cells exocytotic membranes are ingested with two different half-lives, 18 and 85 min (Koike & Meldolesi, 1981).

Are microtubules involved in membrane retrieval? The data in Table 2 suggest that trichocyst ghosts are retrieved without the involvement of mircotubules. When resting trichocysts are analysed the docking sites proper are found to be devoid of persisting microtubules (Plattner, Westphal & Tiggemann, 1982). If microtubules were formed just for the purpose of membrane retrieval, they would have to be detectable in great numbers under the synchronous condition used here. Yet we found only a few microtubules of undetermined origin approaching lon- gitudinally cut trichocyst ghosts and none in cross-sections. Since tubulin polymerization rates invivo are relatively slow (1 /im/min; Berlin, Caron & Oliver, 1979; Fulton & Simpson, 1979; Weber & Osborn, 1979), the assembly period required would take more time than is available for the whole retrieval process after synchronous AED triggering of exocytosis. It is very unlikely that we could have missed this stage. 14 H. Planner and others In other systems endocytosis is reported to be either partly supported by or to be independent of the presence of microtubules (Allison & Davies, 1974; Ose, Ose, Reinertsen & Berg, 1980; Piasek & Thyberg, 1979; Silverstein, Steinman & Cohn, 1977). However, since most cells possess a permanent microtubular system running from the centre to the periphery (Weber & Osborn, 1979), it appears difficult to draw firm conclusions, as drug-induced disruption has not been shown convincingly. In paramecia cortical microtubules are arranged in bundles parallel to the surface (Sedar & Porter, 1955) or they can penetrate from ciliary basal bodies into deeper layers of the cytoplasm (Plattner ef al. 1982). They also display permanently installed microtubular rails along other avenues of intense vesicle transport, e.g. departing from the cytoproct (Allen & Wolf, 1974) or from the cytopharynx (Allen, 1974), along the pathways of disc-shaped vesicles (Allen, 1975). It is not known whether ghosts can associate with microtubule populations in deeper layers of the cell. Though the tubular structures within the collar (collar fibres) are also ~ 25 nm thick and, hence, were considered as true microtubules some time ago (Bannister, 1972), they lack the characteristic protofilament pattern after tannic acid staining (Fig. 14; and Plattner et al. 1982). For this reason and because of lack of antitubulin antibody binding (Cohen et al. 1982) they can no longer be considered as microtubules, all the more so as these structures also undergo retrieval (Fig. 15). In summary, we found no evidence for the involvement of temporary or permanent microtubules in the retrieval of trichocyst ghosts.

Are microfilaments involved in membrane retrieval? The cortex of P. tetraurelia has been shown to bind fluorescentlylabelle d anti-actin antibodies, DNase I, heavy meromyosin (Tiggemann & Plattner, 1981) as well as phalloidin (Kersken et al. unpublished data). This clearly proves the presence of filamentous actin in the cortex. Since phalloidin, which was also used in this study, binds specifically to filamentous actin (Wieland, 1975), we assume that its blocking effect on cytoplasmic streaming is due to the inactivation of the cortical microfilament system; the lack of any blocking effect on membrane resealing and internalization indicates that these filaments are not required for these two processes despite their cortical localization. The precise localization of binding sites of phalloidin and its effects on different cell functions will be documented in more detail elsewhere (Ker- sken et al. unpublished data).

Conclusions on the role of cytoskeletal elements in membrane retrieval There are only a few data on this problem in the literature. For instance, in ovarian granulosa cells endocytosis takes place only partly by a saltatory movement along microtubules (Herman & Albertini, 1984). In chromaffin cells it was inferred from the lack of any effect of anti-microtubule or -microfilament agents, that endocytosis coupled to exocytosis does not need these cytoskeletal elements (Patzak et al. 1984). Only rarely was their structure or function clarified, however, in such experiments. Since we could directly observe the absence of microtubules during synchronous retrieval of trichocyst ghosts, their involvement is ruled out. As to microfilaments we Membrane reseating and retrieval 15 conclude also that they are not involved, because other microfilament-operated functions were visibly stopped after phalloidin injection, when resealing and retrieval of trichocyst ghosts was still proceeding.

We thank Professor Th. Wieland for a generous gift of fluorescently labelled phalloidin, which was used for our microinjection studies. We also gratefully acknowledge the excellent technical assistance provided by Mrs C. Braun, R. Hildebrand, R. Sturzl and C. Wolf throughout this work. The work was supported by grant no. SFB 156-B4.

REFERENCES ALLEN, R. D. (1974). Food vacuole membrane growth with microtubule-associated membrane transport in Paramecium. J. Cell Biol. 63, 904-922. ALLEN, R. D. (1975). Evidence for firm linkages between microrubules and membrane-bounded vesicles. J. Cell Biol. 64, 497-503. ALLEN, R. D. & FOK, A. K. (1984). Membrane behavior of exocytic vesicles. III. Flow of horseradish peroxidase labeled trichocyst membrane remnants in Paramecium. Eur.jf. Cell Biol. 35, 27-34. ALLEN, R. D. & HAUSMANN, K. (1976). Membrane behavior of exocytic vesicles. I. The ultrastructure of Paramecium trichocysts in freeze-fracture preparations. J. Ultrastruct. Res. 54, 224-234. ALLEN, R. D. & WOLF, R. W. (1974). The cytoproct of Paramecium caudatum: Structure and function, microtubules and fate of food vacuole membranes. .7. Cell Sci 14, 611-631. ALLISON, A. C. & DAVIES, P. (1974). Mechanisms of endocytosis and exocytosis. Syntp. Soc. exp. Biol. 28, 419-446. BANNISTER, L. H. (1972). The structure of trichocysts in Paramecium caudatum. J. Cell Sci. 11, 899-929. BARDELE, C. F. (1983). Mapping of highly ordered membrane domains in the plasma membrane of the ciliate Cyclidium glaucoma. J. Cell Sci. 61, 1-30. BEISSON, J., LEFORT-TRAN, M., POUPHILE, M., ROSSIGNOL, M. & SATIR, B. (1976). Genetic analysis of membrane differentiation in Paramecium. Freeze-fracture study of the trichocyst cycle in wildtype and mutant strains. J. Cell Biol. 69, 126—143. BERLIN, R. D., CARON, J. M. & OLIVER, J. M. (1979). Microtubules and the structure and function of cell surfaces. In Microtubules (ed. K. Roberts & J. S. Hyams), pp. 443—485. London: Academic Press. CASTLE, J. D. & PALADE, G. E. (1978). Secretion granules of the rabbit parotid. J. Cell. Biol. 76, 323-340. COHEN, J., ADOUTTE, A., GRANDCHAMP, S., HOUDEBINE, L. M. & BEISSON, J. (1982). Immuno- cytochemical study of microtubular structures throughout the cell cycle of Paramecium. Biol. Cell 44, 35-44. DECAMILLI, P., PELUCHETTI, D. & MELDOLESI, J. (1976). Dynamic changes of the luminal plasmalemma in stimulated parotid acinar cells. J. Cell Biol. 70, 59-74. DOWD, D. J., EDWARDS, C, ENGLERT, D., MAZURKIEWICZ, J. E. & YE, H.Z. (1983). Immuno- fluorescent evidence for exocytosis and internalization of secretory granule membrane in isolated chromaffin cells. Neuroscience 10, 1025-1033. FARQUHAR, M. G. (1981). Membrane recycling in secretory cells: Implications for traffic of products and specialized membranes within the Golgi complex. In Methods in Cell Biology (ed. A. R. Hand & C. Oliver), vol. 23, pp. 399-427. New York: Academic Press. FARRAGIANA, T. & MARJNOZZI, V. (1979). Phosphotungstic acid staining of polysaccharides containing structures on epoxy embedded tissues. J. submicrosc. Cytol. 11, 263—265. FUJIWARA, K. & LINCK, R. W. (1982). The use of tannic acid in microtubule research. lnMethods in Cell Biology (ed. L. Wilson), vol. 24, pp. 217-233. New York: Academic Press. FULTON, C. & SIMPSON, P. A. (1979). Tubulin poola, synthesis and utilization. In Microtubules (ed. K. Roberts & J. S. Hyams), pp. 117-174, London: Academic Press. 16 H. Planner and others HAACKE, B. & PLATTNER, H. (1984). Synchronous exocytosis in Paramecium cells. III. Re- arrangement of membranes and membrane-associated structural elements after exocytosis perfor- mance. Expl Cell Res. 151, 21-28. HAUSMANN, K. & ALLEN, R. D. (1976). Membrane behavior of exocytic vesicles. II. Fate of the trichocyst membranes in Paramecium after induced trichocyst discharge .J. CellBiol. 69, 313—326. HELENIUS, A. & MELLMAN, I. (1983). Valency influences the transport of Fc receptor-bound ligands to lysosomes. jf. CellBiol. 97, 168a. HERMAN, B. & ALBERTINI, D. F. (1984). A time-lapse video image intensification analysis of cytoplasmic organelle movements during endosome translocation. J. CellBiol. 98, 565—576. HOLTZMAN, E. (1981). Membrane circulation: An overview. inMethods in Cell Biology (ed. A. R. Hand & C. Oliver), pp. 379-397. New York: Academic Press. JANISCH, R. (1972). Pellicle of Paramecium caudatum as revealed by freeze etching. J. Protozool. 19, 470-472. KOIKE, H. & MELDOLESI, J. (1981). Post-stimulation retrieval of luminal surface membrane in parotid acinar cells is calcium-dependent. Expl Cell Res. 134, 377-388. LINGG, G., FISCHER-COLBRIE, R., SCHMIDT, W. & WINKLER, H. (1983). Exposure of an antigen of chromaffin granules on cell surface during exocytosis. Nature, Lond. 301, 610—611. MILLER, T. M. & HEUSER, J. E. (1984). Endocytosis of synaptic vesicle membrane at the frog neuromuscular junction. J. CellBiol. 98, 685-698. OLBRICHT, K., PLATTNER, H. & MATT, H. (1984). Synchronous exocytosis in Paramecium cells. II. Intramembranous changes analyzed by freeze-fracturing. Expl Cell Res. 151, 14—20. OSE, L., OSE, T., REINERTSEN, R. & BERG, T. (1980). Fluid endocytosis in isolated rat parenchymal and non-parenchymal liver cells. Expl Cell Res. 126, 109—119. PAPE, R. & PLATTNER, H. (1985). Synchronous exocytosis mParamecium cells. VI. Ultrastructural adaptation phenomena during re-insertion of secretory organelles. Eur.jf. Cell Biol. 36, 38—47. PATZAK, A., BOCK, G., FISCHER-COLBRIE, R., SCHAUENSTEIN, R., SCHMIDT, W., LINGG, G. & WINKLER, H. (1984). Exocytotic exposure and retrieval of membrane antigens of chromaffin granules: Quantitative evaluation of immunofluorescence on the surface of chromaffin cells. J. Cell Biol 98, 1817-1824. PHILLIPS, J. H., BURRIDGE, K., WILSON, S. P. & KIRSHNER, N. (1983). Visualization of the exocytosis/endocytosis secretory cycle in cultured adrenal chromaffin cells. J. Cell Biol. 97, 1906-1917. PIASEK, A. & THYBERG, J. (1979). Effects of colchicine on endocytosis and cellular inactivation of horseradish peroxidase in cultured chondrocytes. J. CellBiol. 81, 426—437. PITELKA, D. R. (1965). New observations on cortical ultrastructure in Paramecium. J'. Microsc, Paris 4, 373-394. PLATTNER, H. (1974). Intramembraneous changes on cationophore-triggered exocytosis in Paramecium. Nature, Lond. 252, 722-724. PLATTNER, H. (1976). Membrane disruption, fusion and resealing in the course of exocytosis in Paramecium cells. Expl Cell Res. 103, 431-435. PLATTNER, H., MATT, H., KERSKEN, H., HAACKE, B. & STORZL, R. (1984). Synchronous exocytosis in Paramecium cells. I. A novel approach. Expl Cell Res. 151, 6-13. PLATTNER, H., MILLER, F. & BACHMANN, L. (1973). Membrane specializations in the form of regular membrane-to-membrane attachment sites in Paramecium. A correlated freeze-etching and ultrathin-sectioning analysis. J. Cell Sci. 13, 687-719. PLATTNER, H., REICHEL, K., MATT, H., BEISSON, J., LEFORT-TRAN, M. & POUPHILE, M. (1980). Genetic dissection of the final exocytosis steps in Paramecium tetraurelia cells: Cytochemical determination of Ca2+-ATPase activity over preformed exocytosis sites. J. Cell Sci. 46, 17-40. PLATTNER, H., STORZL, R. & MATT, H. (1985). Synchronous exocytosis 'mParamecium cells. IV. Polyamino-compounds as potent trigger agents for repeatable trigger-redocking cycles. Eur. J. Cell Biol. 36, 32-37. PLATTNER, H., WESTPHAL, C. & TIGGEMANN, R. (1982). Cytoskeleton-secretory vesicle inter- actions during the docking of secretory vesicles at the cell membrane in Paramecium tetraurelia cells.7. CellBiol. 92, 368-377. PLATTNER, H. & ZINGSHEIM, H. P. (1983). Electron microscopic methods in cellular and molecular biology. In SubcellularBiochemistry (ed. D. R. Roodyn), vol. 9, pp. 1-236. New York: Plenum Press. Membrane reseating and retrieval 17 SATIR, B. (1974