Mapping of highly ordered membrane domains in the plasma membrane of the ciliate Cyclidium glaucoma

Cell fusion experiments and freeze-fracture technique have led to the now popular ‘fluid mosaic’ model of biological membranes (Singer & Nicolson, 1972). It took some time before it was realized that the extreme degree of fluidity observed in various tissue-culture lines and blood cells could lead to the danger of unjustified generalizations. Cell junctions, synapses and the ciliary membrane, to quote only a few examples, show a high degree of order at the supramolecular level, with probably little or no fluidity at all.

The plasma membrane of the ciliate cortex as well as the ciliary membrane display a variety of highly ordered membrane domains. In particular the ‘ciliary necklace’ and the ‘ciliary plaques’ have received much attention (Gilula & Satir, 1972; Plattner, 1975; Byrne & Byrne, 1978). Despite the great interest in intramembranous particles as possible ion channels, the function of the various particle arrays in the ciliate cortex is largely unknown.

This paper has two objectives: (1) to search for a suitable ciliate model system in which the entire cortex could be mapped at the level of intramembranous particle arrays, to get an impression of the local diversification of the plasma membrane; and (2) to see whether a hymenostome ciliate, equipped with cilia of different functions, has the same type of specialized membrane domains such as a double-stranded ciliary necklace and ciliary plaques in all of its cilia. It is obvious that a ciliate like Paramecium, which (depending on the species studied) has 5000 to 8000 cilia, is totally unsuited to such an investigation. Even in Tetrahymena, with some 400 somatic cilia, the effort necessary would be immense (if not in vain, because of the paucity of landmarks in relation to the size of the cell). To satisfy the requirement for positive identification of each particular cilium I have chosen the freshwater protozoon Cyclidiumglaucoma, which shows several easily recognized landmarks such as a non-ciliated apical area, a large oral apparatus, a cytoproct, a contractile vacuole pore and a single stiff caudal cilium, together with an asymmetric distribution of some 200 somatic cilia in 10 kineties only.

Besides the structural details of the ciliary membrane I describe three types of highly organized membrane domains in the plasma membrane. These complex arrays of intramembranous particles (IMPs), termed ‘particle plates’, ‘simple’ and ‘complex pellicular rugs’, occur in precisely fixed positions in relation to the general topography of the cell. The distribution of the pellicular rugs shows no gradient, but a sharp boundary (similar to a segment border) at the latitude of the future fission zone. C. glaucoma is the first ciliate for which almost the entire cell surface is mapped with a hitherto unequalled degree of accuracy.

Several other ciliates have proved to be powerful model systems for developmental biologists interested in cortex-related morphogenetic events (Frankel, 1974; Sonne-born, 1975; Nanney, 1980). It is hoped that further studies with Cyclidium on the mechanisms that govern assembly and positioning of the membrane domains during stomatogenesis and cell division will contribute to a better understanding of membrane-mediated developmental processes.

Cyclidium glaucoma O. F. Müller, 1786 stock LB 1616 from the Cambridge Culture Collection was grown in bacterized ‘Protozoan Pellet’ medium (North Carolina Biological Supply, Burlington, U.S.A.) with a boiled rye grain. For preparation of thin sections cells were concentrated by lowspeed centrifugation and fixed in 3 % glutaraldehyde and 0-5 % OsCL in 0 05 M-phosphate buffer (pH 7·0) for 15 min at room temperature. Following standard procedures the cells were dehydrated in a graded series of ethanol followed by two changes of propylene oxide and embedded in Epon 812. The thin sections were stained with uranyl acetate and lead citrate and examined in a Siemens Elmiskop 102.

For freeze-fracture electron microscopy cells were fixed in 1 % glutaraldehyde in 0·05 M-sodium cacodylate/HCl buffer (pH 7·3) for 15 min at room temperature, washed in the same buffer and finally soaked in 30% glycerol in the same buffer for at least 1 h. The cells were then placed into double-replica specimen holders, rapidly plunged into nitrogen slush and transferred to a Balzers BAF301 apparatus. Platinum/carbon replicas were prepared with an electron beam gun, starting the shadowing process immediately before the actual fracturing was done, thus reducing to a minimum possible contamination by recondensation of water (this procedure excludes an etching step). The replicas were cleaned with sulphuric acid and sodium hypochlorite and mounted on uncoated 200 mesh grids. To ensure better adherence of the replicas to the grids the latter were dipped in 10 ml toluol in which the glue of 1 cm² of Tesa tape had been dissolved. The replicas were examined in a Zeiss EM9.

For comparison, living cells kept at room temperature up to the moment of cryofixation, as well as samples of cells cooled to 5 °C within 4 min were prepared in the same way. All particle arrays described in this paper are thermostable and not affected to any meaningful extent by chemofixation. The major drawback of preparing replicas from living cells is the lower yield of large-sized replicas. This is because unfixed mucocysts (probably through conformational changes of their constituents) tear the replicas into pieces the moment the specimen holder is immersed in water to float the replica.

The photographic prints are made from the original negative, therefore shadows appear white. The direction of shadowing is indicated by the encircled arrowhead. All micrographs were mounted in the proper orientation relative to the axis of the animal, with the anterior end pointing to the upper margin of the page. In general the protoplasmic halves (P-face) of plasma membrane are shown; this means that the observer sees the cell from an outside viewpoint. To facilitate comparison, the thin-section micrographs are also printed as seen from outside the cell. This means that the animal’s right is on the left side of the figures and vice versa, following the general convention that right and left in the ciliate cortex refer to the viewpoint of an observer standing inside the cell, parallel to the cell’s anterior-posterior axis, and looking out.

C. glaucoma is a very common freshwater ciliate, which belongs to the pleuronematine scuticociliates, a subgroup of the Hymenostomata (Small, 1967). Among the better-known ciliates, Tetrahymena can be regarded as a fairly close relative of Cyclidium. Even the most radical challenge of the Corlissian ciliate classification (Corliss, 1979) by Small & Lynn (1981) - which will hardly be acceptable to the majority of ciliatologists - keeps some of the hymenostomes (including Tetrahymena) and the scuticociliates (including Cyclidium) close together. Also ovoid in outline, C. glaucoma is considerably smaller than Tetrahymena: it measures only 20 μm in length and 10 μm in width. The cell surface is striped by 10 or occasionally 11 longitudinal grooves, which bear the somatic cilia. C. glaucoma is an erratic swimmer; periods of straightforward swimming are often suddenly interrupted by periods of cessation of movement of the body ciliature. The swimming and resting behaviour of C. glaucoma is very characteristic and pertinent to the observations described in this paper.

When stationary, C. glaucoma lies with its dorsal face against the bottom of the culture dish or some other object and all somatic cilia are spread upright perpendicular to the cell surface (Fig. 1). Actually, it is not the dorsal body surface that touches the substrate but the tips of its dorsal cilia, on which the ciliate rests as if on stilts. During these periods of rest the paroral membrane, a single row of cilia, which is the most conspicuous feature of the oral ciliature, is held extended like a stiff velum. When feeding, which occurs when the ciliate is at rest, the paroral membrane acts as a wall or filter against which the water current with the food particles (bacteria) is driven by the constant beating of the other oral cilia (Fenchel, 1980). During swimming, when the extended paroral membrane would be a hindrance, it is folded back against the posterior half of the cell.

Before describing the ultrastructure of the somatic cortex, the general topography of C. glaucoma has to be outlined in more detail. The 10 longitudinal ciliary rows or kineties (K1–K10; KI being the first to the right of the paroral membrane and counted in clockwise manner as seen from the anterior pole of the cell) do not reach the anterior end of the cell, which is a naked, slightly swollen cap (Fig. 1). The ventral face is defined by the position of the oral apparatus. The oral ciliature of C. glaucoma is rather complex, but its asymmetry also helps to identify smaller replica pieces of the nearby somatic cortex once the topography of the oral area is fully understood. The paroral membrane (POM) on the right side of the oral cavity consists of 27–29 cilia.

Each cilium except the anterior one is accompanied by a parasomal sac. As Fig. 4 shows, the infraciliature of the paroral membrane consists of two rows of kinetosomes. Only the right row is ciliated. The left row of non-ciliated kinetosomes is not visible in freeze-fracture replicas because these kinetosomes are covered by the alveolar sac system (therefore the non-ciliated kinetosomes of the paroral membrane have been omitted in Fig. 22). The left-side oral cilia do not form ordered membranelies, as in Tetrahymena, but form three irregular tufts of cilia called polykinety numbers 1, 2 and 3 (PKI, PK2, PK3). The apportionment of cilium 1 through 9 to PK1 and of cilium 10 through 31 to PK2, which fits fairly well with the enumeration chosen by light microscopists, is based on the location of these cilia on different levels within the buccal cavity. In fixed cells the cilia of PK1 and PK2 are oriented towards the posterior end of the cell while the rather short cilia of PK3 point towards the cytostome. The cytostome is located in the lower left corner of the buccal cavity (compare Figs 1, 8 and 22; a more detailed description of the oral apparatus will be the subject of a separate communication). Further landmarks are represented by the cytoproct, a4 μm long slit-like transient opening of the cell for defaecation, located just posterior to the buccal cavity, the pore of the contractile vacuole behind the last kinetosome of K2, and the stiff caudal cilium, which emerges from a small depression at the very hind end of the cell.

Once these landmarks are known it is possible to analyse the somatic ciliature in more detail. In the anterior half of the cell the cilia are more closely spaced than in the posterior half. This is partly due to the fact that the cortical units in the anterior half of the cell consist of pairs of kinetosomes while in the posterior half they include single kinetosomes only. In an anterior unit, hereafter called a dikinetid, both kinetosomes may be ciliated. But often, in particular in those dikinetids that are located more posteriorly, the anterior kinetosome of a pair may be barren. Rather than counting the number of cilia, which may vary from cell to cell, the average number of kinetosomes per kinety is given below. Furthermore, it should be obvious that it is impossible to map the entire somatic cortex from a single cell because at best 40–50% of the surface area of a particular cell is exposed by the freeze-fracture technique. But due to the many landmarks it is feasible to piece together the cell surface of an ‘average Cyclidium’. From 52 favourable replicas, similar to Figs 6–8, which showed large identifiable areas of the cortex, the following numbers are derived. Kinety no. 1 has 8 double and 4–5 single kinetosomes (KI : 8d/4—5s) ; the respective data for the other kineties are K2: 8d/4–5s; K3: 6—7d/4s; K4: 6–7d/4s; K5: 5–6d/4–5s; K6: 5–6d/4-5s; K7: 6–7d/4–5s; K9: 7d/6–8s; and K10: 7d/4s. There is something special about K9. This kinety is unique in possessing four monokinetids spaced more closely than the other single cilia in the posterior half of the cell, thus forming another useful landmark.

The knowledge of the ultrastructural organization of the somatic cortex in C. glaucoma as derived from thin sections (Figs 2, 4) is another prerequisite for the understanding of the distribution of the membrane domains described later on. Since no detailed description of the pellicle of C. glaucoma is available, the salient features are given in this section and summarized schematically in Fig. 5.

The term cortex as used in this paper is defined as comprising the pellicle proper and an approximately 1 μm thick layer of cytoplasm immediately underneath the pellicle. In this cortical cytoplasm the infraciliature (the kinetosomes with their auxiliary fibre systems) as well as cytoplasmic organelles are found in precise spatial relation to the overall topography of the cortex. Beginning from the outside, there is the plasma membrane, which covers the cell body and the cilia. In most parts the cell surface is underlain by a single layer of alveolar sacs; these are missing only at particular sites, e.g. at the cytostome, the cytoproct and the pore of the contractile vacuole. Moreover, the alveolar sacs circumvent the sites where cilia emerge, where mucocysts (secretory organelles) are docked to the plasma membrane, and where this latter membrane invaginates to form parasomal sacs, interpreted as sites of pinocytotic activity. Alveolar sacs, bounded by a unit membrane, are flat cisternae that extend from one kinety to the next. Their spatial dimensions are seen most clearly in silver-stained preparations (see Didier & Wilbert, 1981). The anterior naked end of the cell is underlain by a single alveolar sac that extends into the interkinetal space down to the second or third dikinetid of each kinety. Then comes a girdle of comparatively small alveolar sacs between the third and the sixth dikinetid, followed by long alveolar sacs that extend all the way down to the posterior end of the cell.

Plasma membrane and outer alveolar membrane run parallel to each other, separated by a narrow space of constant width (Fig. 4). High-power thin-section micrographs have revealed thin struts between these membranes, similar to those seen in Tetrahymena (Franke, 1971 ; Satir & Wissig, 1982). The lateral borders of adjacent alveolar sacs are closely pressed together, forming the alveolar septa. These septa are perforated by 50 nm wide pores, indicating that the alveolar system in C. glaucoma forms a continuous space, as first described for Paramecium (Allen, 1971). The inner alveolar membrane (which shows a particularly high number of IMPs on its P-face; Figs 9, 15) lies in close contact with the epiplasm. The epiplasm consists of a thin layer of amorphous material and is connected with the terminal plate of the kinetosomes (Fig. 2). Together with the kinetosome-associated fibre systems the epiplasm forms a cortical scaffold or cytoskeleton (Collins, Baker, Wilhelm & Olmstead, 1980).

The structures described so far follow the asymmetric pellicular ridges that accompany the longitudinal kineties. The ridges have a steeper slope on the left side and a less steep slope to the right. Mucocysts (seen in cross-section in Fig. 4) are positioned near the crest on the right slope of the ridge. The cortical cytoplasm of the pellicular ridges is further characterized by one to three longitudinal microtubules, which are embedded in the epiplasm and by a 50 nm wide tube of smooth endoplasmic reticulum accompanied by an irregularly shaped cisterna of rough endoplasmic reticulum (all shown in Figs 2, 5).

In C. glaucoma, as in most ciliates, there are three sets of kinetosome-associated fibres that originate from the proximal end of the basal body (the orientation of these microtubular and fibrillar appendages of the kinetosomes has to be known to evaluate possible long-distance positional information exerted upon the membrane domains described in a later section). The posterior kinetosome of the dikinetids as well as the single kinetosome of the monokinetids have a short kinetodesmal fibre that extends from the proximal anterior to the lateral right portion of the kinetosome towards the cell surface. In C. glaucoma the kinetodesmal fibre is relatively short. Only in the most closely spaced anterior dikinetids does it pass the more anterior pair of kinetosomes to its right. There is no elaborate overlapping of kinetodesmal fibres as in Paramecium. A ribbon of four postciliary microtubules arises near the proximal, posterior right quadrant of the same kinetosomes that bear the kinetodesmal fibre. The postciliary microtubules extend posteriorly towards the cell surface. During their course they cross the kinetodesmal fibre of the more posterior kinetid. Finally, a ribbon of four transverse microtubules originates anterior to the proximal end of the posterior (or single) kinetosome and extends to the left. Gradually, the number of microtubules forming the transverse ribbon increases to six, seven or eight. During most of their course the transverse microtubules are oriented perpendicular to the axis of the kineties and pass underneath the flat slope of the pellicular ridge. As in Tetrahymena (Aufderheide, 1979), there is a precise association of mitochondria with these transverse microtubules. The anterior kinetosome of the dikinetids (though often ciliated) has no fibrillar or microtubular associates. Monokinetids have a parasomal sac to the anterior right of the kinetosome; in dikinetids a single parasomal sac lies in the same position with respect to the posterior kinetosome.

The small size and its many landmarks render C. glaucoma an ideal object for freeze-fracture analysis. Fig. 6 shows the anterior pole of C. glaucoma with the naked apical area. The P-face of the plasma membrane is exposed over most of the surface. The two ‘scars’ in the naked apical area mark the location of two major particle places, PPI and PP2.

Fig. 7 shows the posterior end of the cell with the caudal cilium broken off. The stubs of only two somatic cilia are seen, nonetheless it is possible to identify the kineties. A useful landmark to enumerate the kineties is the cytoproct in the upper left corner of the micrograph. The flat furrows that point to the position of the caudal cilium mark the course of the alveolar septa underneath the plasma membrane. There are always two parasomal sacs near the caudal cilium, located on meridians nos 4 and 9.

Fig. 8 shows a favourable view of the buccal cavity with the paroral membrane seen in its full length. The last five stubs of the paroral cilia are somewhat difficult to see; they have been numbered consecutively. The numbers and positions of the cilia of the polykineties as well as the distribution of parasomal sacs are precisely fixed in C. glaucoma. Fig. 8 shows polykinety no. 1 and the anterior part of polykinety no. 2. Note, for example, the sequence of cilia in PK2, which starts with 1-1-1-2-2-3-5 cilia. Most of the cilia of PK3 lie underneath the cilia of PK2 and are not visible in Fig. 8. C. glaucoma has a ribbed wall in the posterior region of the oral cavity adjacent to the paroral membrane. Eight ridges of IMPs are seen in the freeze-fracture replicas. These ridges correspond to the oral ribs known from thin-section studies of the oral apparatus in other hymenostome ciliates (Allen, 1978a,b; Sattler & Staehelin, 1979). The area of the oral ribs is underlain by slim alveolar sacs, and two sets of microtubules run underneath the particle ridges.

The freeze-fracture view of the cytoproct is shown in Fig. 9. The cytoproct is surrounded by a 100 nm broad band of closely opposed IMPs. The band marks the line along which the free lateral border of the alveolar sacs are closely sealed to the plasma membrane (the contractile vacuole pore (not shown) has the same appearance). The left margin of Fig. 9 gives a three-dimensional view of the pellicle with the alveolar sacs.

The attachment sites of mucocysts to the plasma membrane are inconspicuous in specimens with exposed P-face of the plasma membrane. The typical fusion rosettes (not shown) stay with the E-face of the plasma membrane. Recently discharged mucocysts disclose themselves through particle-lined craters of varying diameter (see examples shown in Figs 7, 18). The membrane that surrounds the mucocysts is not incorporated into the plasma membrane as has been postulated for discharging mucocysts of Tetrahymena (Satir, Schooley & Satir, 1972).

As mentioned in the Introduction, one major objective of this investigation was to check whether the interior architecture of the ciliary membrane looks the same in all cilia in a particular ciliate species. The answer is no. All cilia, the somatic as well as the oral cilia, and the stiff caudal cilium of C. glaucoma, have a double-stranded ciliary necklace (Gilula & Satir, 1972). But there are noteworthy differences in the ciliary plaques of somatic and buccal cilia.

Typical ciliary plaques consisting of three rows of four to seven orthogonally arranged I MPs located adjacent to each peripheral microtubule doublet, as described for Tetrahymena (Wunderlich & Speth, 1972), Paramecium (Plattner, Miller & Bachmann, 1973) and many other hymenostome ciliates (Bárdele, 1980, 1981), are found in the somatic cilia of C. glaucoma (Fig. 11). The overall appearance of the ciliary plaques is the same in all motile somatic cilia, in the anterior pairs of cilia as well as in the single cilia in the posterior half of the cell. The size of the individual plaques varies between 3×2 and 3×8 particles; most frequently there are 3×6 and 3×7 particles per plaque. Cilia with particularly small plaques could not be associated with any specific area of the cortex and in the dikinetids there was no significant difference in plaque size between the anterior and the posterior cilium.

It was particularly time-consuming to search for favourable fractures of the single caudal cilium. Most often the posterior end of the cell was encountered in a view as seen in Fig. 7, in which the caudal cilium is broken off at the necklace level. Fractures with a longitudinal view of the proximal part of the caudal cilium were extremely rare. In order to identify the caudal cilium positively the whole cell had to be fractured longitudinally through the very middle of the cell, otherwise an ordinary somatic cilium could easily be mistaken for the caudal one. Fig. 12 shows one of eight specimens in which the caudal cilium was identified with absolute certainty. It shows a higher number of particles in the plaque area but there are no proper plaques.

All oral cilia, those of the paroral membrane and the polykineties, have ‘modified ciliary plaques’. These consist of four (instead of three) rows of particles, which are not in register but are staggered (Fig. 13). Sometimes one has the impression that the inner rows are more closely spaced than the lateral rows. The sizes of the modified plaques vary considerably and, in contrast to the particles of the ordinary plaques, which are associated mostly with the P-face, particles of the modified plaques are frequently associated with the E-face of the ciliary membrane.

The location of non-ciliated kinetosomes in the somatic cortex of C. glaucoma is easily identified by the ‘fairy rings’ of particles (Fig. 10), a vivid term coined by Hufnagel (1979). As described above, it is most often the anterior kinetosome of a dikinetid that lacks a cilium. The precise position of the parasomal sac adjacent to the upper right corner of the posterior kinetosome further helps to identify the position of the anterior one. The fairy rings often show square-shaped arrangements of four IMPs. The number of squares is hard to estimate owing to their irregularity, but the observed configurations are at least consistent with the possibility that the true number is nine. Admittedly, this needs further inquiry.

Three types of highly ordered membrane domains occur in the P-face of the plasma membrane of C. glaucoma. They have been termed ‘particle plates’ (PP), ‘simple pellicular rugs’ (SPR), and ‘complex pellicular rugs’ (CPR).

These consist of parallel ridges of particles. The number of ridges within a PP is not fixed but there is a plate-specific average size. There are four large PPs, two on the naked apical area and two in the buccal cavity and, in addition, four smaller ones that are less easily recognized. Since most PPs occur in a fixed position, they have been labelled PPI through PP8. The more lateral ridges of a PP may be shorter than the more central ridges, so that the edges of the PP have a rounded appearance. The lateral distance between the ridges is 28 nm. Higher magnification (Fig. 19) shows that each ridge actually consists of 3 rows of particles, a middle row of large particles 8–9 nm in diameter, accompanied on both sides by a row of smaller, 5–6 nm wide particles.

Very often the PPs are broken away in replicas, which otherwise show large areas of the P-face of the plasma membrane (Fig. 6). Under these scars the E-face of the outer alveolar membrane is exposed. The special fracturing behaviour at the sites of the PPs may be regarded as an indication of a close structural connection between the P-face of the plasma membrane and that of the outer alveolar membrane. Sometimes at least a peripheral part of the PP is retained, which allows one to determine the orientation of the particle ridges (see PP2 in Fig. 6 or PP4 in Fig. 16 as examples). It must be noted that not every single PP could be documented in its plasma membrane aspect, for lack of space. The size and position of PPs and the orientation of their ridges as seen in a large number of photographs are summarized in Fig. 22.

PPI and PP2 are located on the naked apical area. PPI is the largest PP, covering a surface area of 0·8–1 μm². Its 20-25 parallel ridges are oriented from K2 to K6. PP2 is located just anterior to the oral apparatus and the anterior end of K10 and K9. The other PPs are located within or near the oral apparatus. PP3 is very prominent, and is located on an elevated shelf to the left of the anterior part of PK2. The position of PP3 is indicated in Fig. 8; here again the plasma membrane that carries the PP is torn off. The particle ridges of PP3 are oriented longitudinally with respect to the main axis of the cell. This is also the case for PP4, which is difficult to see in Fig. 8 but more clearly demonstrated in Fig. 16. It is located between the paroral membrane (usually at the latitude of the fifth to the tenth paroral cilium) and PK2. PP4 measures 0·6–0·9 μm². PP5 is the least regularly occurring PP, located on the naked apical area to the left of PP2, and 0 3 μm²in size. Sometimes it is positioned between the anterior ends of K9 and K8. It is not quite certain whether PP5 is totally independent or whether it should be regarded as a derivative of PP2. PP6 and PP7 are rather small, and are both located on the left border of the buccal cavity (Fig. 8). In PP6 the particle ridges are oriented perpendicular to the main axis of the cell, while those of PP7 are oriented more or less parallel to the main axis. PP8 lies in the dorsal wall of the buccal cavity (Fig. 14). Most often this plate is hidden by the cilia of PK2. The particle ridges of PP8 are oriented perpendicular to the main axis of the cell. The type of membrane specialization represented by PP 1 to PP8 never occurs on the posterior half of the cell.

These are restricted to the ciliated somatic cortex of the anterior half of the cell. They are excluded from the naked apical area and the buccal cavity. The SPR consists of three parallel rows of 10 nm particles, in close but irregular association with the anterior half of KI through K10 (Figs 10, 16). The distance separating the longitudinal rows is 25 nm and the centre-to-centre spacing of the particles in a longitudinal row is 19 nm. The overall length of an SPR is between 0-2 and 0-6μm. The particles of the SPR are never found as single or double rows. Individual particles within a row may be missing but this usually leaves the general order unaffected. It is highly probable that in those cases the missing particle was associated with the other half of the membrane. Occasionally two or three SPRs lie parallel to each other (Fig. 16). The different lengths of the individual groups of three rows again shows that we are dealing with complete and separate membrane domains.

The position and orientation of the SPRs are not fixed with respect to any morphological feature of the kineties or any of the deeper-lying fibrillar or microtubular systems of the cortex (Fig. 3). There is, however, a close association with the kineties themselves, in that one end of the SPR usually lies within a distance of less than 0·2 μm from the distal portion of the somatic kinetosomes. Moreover, the longer SPRs tend to be oriented more or less parallel to the main axis of the kinety. Number, length and orientation of SPRs were mapped for individual kinetosomes of precisely identified kineties in 10 different cells but no repeatable pattern was found.

These structures are a characteristic feature of the posterior half of C. glaucoma. Except for the more complex morphology of the CPRs, most of what has been said for the SPRs also holds for the CPRs. The particles of the CPRs are slightly larger than those of the SPRs. Each CPR (Figs 15, 17, 21) consists of three central rows of 12 nm particles accompanied by two rows of more closely spaced particles on either side. The particles of these outer double rows are close to 11 nm in diameter. The lateral spacing of the central rows and their distance from the double rows is 23 nm; the corresponding spacing in the side rows is 14nm. The centre-to-centre distance of the particles along the three central rows is 21·5 nm, while the corresponding distance in the lateral double rows is 15·5 nm. Particularly complex are multiple arrays of CPRs. For the sake of brevity the normal ordinary CPR will be symbolized by the sequence 232 (3 stands for the 3 central rows, 2 for the lateral more closely spaced particles rows). In particular, near the caudal cilium the following arrays were found: 232232 and 232232232 (lateral associations of complete CPRs), but also 23232, 2323232 (Fig. 18) and 22322. Other theoretically possible combinations like 2332, 232 or 3223 did not occur.

The highly differentiated unicellular eukaryotes such as ciliates or hypermastigote flagellates pose particularly intricate developmental problems. Their complexity shows many parallels to that of multicellular organisms. In the latter the developmental processes are generally dealt with at the cellular level: differential gene activation in spatially separated cells, cell shape changes, cell movement and problems of cell recognition dominate the discussion. In unicellular organisms the spatially and temporally controlled assembly of cell organelles and suborganellar structures, their intracellular movement, their positioning and interaction become more apparent (Aufderheide, Frankel & Williams, 1980). The technical approach to the solution of these problems has to be diverse, as do the model systems. While certain ciliates are particularly suited to experimental manipulation of the cortex by either genetical (Paramecium, Tetrahymena) or microsurgical techniques (Stentor, Stylonychia), its small size and many landmarks make Cyclidium particularly useful for cortical mapping. It is indeed the first ciliate for which almost complete mapping at the level of intramembranous particle arrays within the surface membranes has been done. The close relationship between Cyclidium and Tetrahymena will give some weight to the results obtained with this seemingly exotic ciliate. The tremendous structural diversification of its surface membranes has numerous functional and developmental implications that deserve a broader discussion. I will start with the ciliary membrane, for which speculations on the functional significance of the particle arrays have generated much interest. I will then proceed to the highly ordered domains of the plasma membrane, discuss their possible function and their significance in relation to membrane fluidity and comment on the conclusions to be drawn from their localized distribution with respect to the geography of the ciliate’s cortex.

All cilia in ciliates have a double-stranded ciliary necklace (Bárdele, 1981). It has been postulated that the ciliary necklace acts as an area of Ca²⁺ influx and thereby is involved in the control of ciliary beat (Gilula & Satir, 1972). In this context it is noteworthy that four examples of ‘supposed sensory cilia’, which do not beat (Bárdele, 1981) as well as the stiff caudal cilium of C. glaucoma, do have a double-stranded ciliary necklace. A certain type of necklace - and there is considerable variation in this characteristic among the various protist groups (Bárdele, 1983) - occurs in most cilia and flagella, but so far no correlation between particular beating patterns (or absence of beating) and the morphology of the necklace area has been established.

The same unsatisfactory situation holds for the ciliary plaques. In C. glaucoma two types of ciliary plaques are found: the ordinary ones in the motile cilia of the somatic cortex and the modified ones in the oral cilia. Interestingly enough, the motionless caudal cilium does not show any plaques. But nevertheless numerous IMPs of the same size as the plaque particles are found at the proximal end of the caudal cilium. One is tempted to ask whether in this particular case the lack of ciliary plaques is causally related to the immotility of the cilium. Certainly this suggestion cannot be generalized, since I have recently shown that the vast majority of (motile) ciliates lack ciliary plaques. The plaques are restricted to certain taxonomic groups, which, judged on other independent characters, are thought to be phylogenetically related (Bárdele, 1981). Sattler & Staehelin (1974) observed three types of oral cilia in Tetrahymena pyriformis that showed differences in the particle distribution in the distal part of their ciliary membrane. The authors give no detailed information on the proximal part of the ciliary membrane other than the notion that all oral cilia have a double-stranded ciliary necklace. My own unpublished observations on T. pyriformis (a laboratory strain of unknown origin) have shown the same type of typical ciliary plaques (with three rows of particles) on both the somatic cilia and most of the buccal cilia. In C. glaucoma only the somatic cilia have this type of ‘ordinary’ plaque. Certainly the beating behaviour of somatic and buccal cilia is different, but the same is also true for the buccal cilia themselves. During feeding the cilia of the paroral membrane (which have the same modified ciliary plaques as the cilia of the polykineties) are set upright like a sail, and only occasionally does a metachronal wave pass down the paroral membrane, while at the same time the cilia of the polykineties are in constant motion. Here again, there is no obvious relation between the beating behaviour of particular cilia and a distinct type of ciliary plaque. There is no doubt that ciliary beat is controlled in some way by the intraciliary calcium concentration (Naitoh & Kaneko, 1972). Deposits of calcium (Plattner, 1975) and other bivalent ions (Fisher, Kaneshiro & Peters, 1976) have been localized in the ciliary plaques area of Paramecium by using the Oschman & Wall (1972) technique. But ciliates that lack ciliary plaques (e.g. Spirostomum) likewise show calcium deposits in the proximal region of the cilia when fixed in media containing 5 mM-calcium (Bárdele, unpublished). So far, the behavioural mutants of Paramecium tetraurelia have been of little help in clarifying the function of the ciliary plaques. While the mutant ‘paranoiac’ d4–578 showed an abnormal plaque area, another paranoiac (d4–90) did not differ from the wild-type s51 (Byrne & Byrne, 1978). Elsewhere I have noted that mid-log phase cells of a third mutant with paranoiac phenotype (d4-150) as well as the ‘pawn’ mutant d4–500 have absolutely normal-looking ciliary plaques (Bárdele, 1981). Though these observations do not exclude the possibility that the plaque particles are involved in localized ion transport (for a non-functional plaque may still be a visible one), the restricted occurrence of ciliary plaques among ciliates should limit extensive speculation.

Particle plates very similar to those described for C. glaucoma have been found on the anterior ventral surface of Paramecium caudatum (Allen, 1976, 1978a,b) and T. pyriformis (Hufnagel, 1981a,b). For the latter species the tripartite nature of the particle ridges has likewise been shown. In more than 20 other ciliate species I have seen similar rectangular particle plates with tripartite particle ridges, which had a lateral separation of 25–30 nm. In those species in which the position of the particle plates could be localized, thanks to landmarks, it was the anterior ventral and/or buccal area of the ciliates in which the plates occurred. Thus it looks as if particle plates are typical features of many ciliates. The particle plates seem to have components that span the entire width of the plasma membrane and protrude from the external surface of the membrane, since they have been demonstrated in deep-etched preparations of Tetrahymena (Hufnagel, 1981b).

Allen (19786) has suggested that the particle plates may act as chemoreceptors. Likewise one may speculate that they have a mechanosensory function. There is a good chance that a forward-swimming Cyclidium would bump against an obstacle with its anterior plate-bearing end first. Deciliated Paramecium cells are still sensitive to mechanical stimuli, indicating that it is the body surface rather than the cilia where the mechanosensitive channels are localized (Ogura & Machemer, 1980). The particle plates in the buccal cavity of C. glaucoma may, when touched by food particles, induce specific feeding motion of the oral cilia. It has also been emphasized that prior to and during conjugation a number of membrane-mediated changes take place in which the particle plates might be involved (Hufnagel, 1981a). On the other hand, it has been reported that the area in which the actual membrane fusion takes place during conjugation is smooth (Wolfe & Grimes, 1979). In any case, one must recall that there is no direct evidence for a causal relationship between the particle plates and any of the suggested functions, other than the seemingly appropriate location of the particle plates.

So far it has also been impossible to isolate any of the arrays of IMPs observed in ciliates. No reliable method is even available to separate the plasma membrane from the two alveolar membranes. Thus the ‘pellicular membrane’ subjected to biochemical analysis comprises three different types of membranes, isolated and processed together. The problem becomes even more serious since we now have to accept the great structural diversity within an individual membrane, a fact that in all probability does not hold for Cyclidium only.

The simple pellicular rugs resemble the ordinary ciliary plaques, but there is no indication that they will enter the ciliary membrane during outgrowth of young cilia (or that they are the unused remainders of such a process). This supposition can be rejected even more decisively for the complex pellicular rugs. The position of the pellicular rugs in the close neighbourhood of the proximal ends of cilia might be an expression of some functional relationship between these two structures. Could the rugs be mechanosensitive sites, too? Perhaps the cilia become immobilized when they touch the rugs. Simultaneous ciliary arrest of all somatic cilia is typical behaviour in C. glaucoma. But why are there two different rugs in the two halves of the cell?

Nearly all patterned particle arrays that show a fixed number of particles in one dimension have been described from sites of membrane-to-membrane or membrane-to-microtubule junctions, where the underlying components somehow define the orientation of the particle array. At least for the pellicular rugs in C. glaucoma, where the three-dimensional architecture of the membranes and microtubule system of the somatic cortex is known in sufficient detail, it can be stated that there is no visible structure that might be responsible for the differentiated assembly and positioning of these membrane domains. It is particularly obvious that the microtubules in the cortex cannot serve as guidelines for the assembly of the particle plates of the rugs, because these arrays are separated from the microtubules by the alveolar sacs. A longdistance effect of the microtubules across the alveolar sacs seems very unlikely, since the pellicular rugs do not show an orientation that would match any of the fibre systems.

Different levels of complexity may be recognized in the assembly of membrane domains. (1) A random assembly of spheres in a plane will result in a close-packing or hexagonal array. This can be regarded as the most simple two-dimensional pattern. If the subunits are not spheres but are more complex in outline as, for example, the bacteriorhodopsin of the purple membrane, a more complex pattern will result, but it will still be a hexagonal one with no lateral restriction. (2) Another level of complexity is reached if a preferential orientation of assembly is assumed, as is the case with the particle plates. The number of available subunits will determine the size of the plate. But certainly the process is more complex in C. glaucoma where, for instance, plates of different but more or less fixed size are found close to each other and where, moreover, the orientation of particle ridges is fixed. In this case it is particularly tempting to search for some deeper-lying cortical structures, which define orientation, size and position of the particle plates. But one has to realize that these suggestions only push the problem further into the dark. (3) Membrane domains with specified numbers of constituents exemplify an even higher level of complexity. In the pellicular rugs only the width of the array is specified while the length is variable. To my knowledge the complex pellicular rugs are the most complex membrane domains described so far. It will be very difficult to discover the rules that govern the assembly of the 2-3-2 pattern of the particle rows. Peripheral proteins could be involved but their patterns will be difficult to understand.

The fracturing behaviour of the particle plates - in particular that of the larger ones - can be regarded as strong evidence for structural interconnections between the two protoplasmic leaflets of the plasma membrane and the outer alveolar membrane. In Tetrahymena corresponding crossbridges or struts between these two membranes have been demonstrated by Franke (1971) and more recently by Satir & Wissig (1982). These struts ensure constant spacing between the two membranes but they will also restrict membrane fluidity. Though the outer alveolar membrane is continuous with the inner alveolar membrane, in most ciliates there is a pronounced difference between the numbers of particles in the two membranes. In C. glaucoma, and to my knowledge in all other ciliates checked so far, it is the inner alveolar membrane that has an exceedingly high number of IMPs. The functional significance of this difference is not known, but it certainly can be regarded as another example of restricted membrane fluidity or restrained exchange of components between the two parts of an obviously continuous membrane system. On the other hand it was the alveolar system of Tetrahymena that served as a model system for experimentally induced membrane fluidity (Speth & Wunderlich, 1973). Subsequently, Kitajima & Thompson (1977) and Nozawa & Thompson (1979) have shown that the plasma membrane and the ciliary membrane (which is of course continuous with the former) are the least fluid membranes in Tetrahymena. This is in agreement with the observations in C. glaucoma. Although no extensive studies have been performed with C. glaucoma, preliminary studies have shown that none of the membrane specializations in the ciliary membrane and the plasma membrane showed any alteration after cooling Cyclidium cells down to 5 °C within 4 min and processing them immediately or after storage at 5 °C for 30 min. In cold-treated cells as well as in the controls there was no indication of any lateral displacement of the membrane domains. Simple pellicular rugs never enter the posterior half of the cell, nor do complex rugs cross over to the anterior half. Likewise the particle plates stay at their confined locations. Thus, the plasma membrane of C. glaucoma is a non-fluid membrane, which in connection with other cortical structures, e.g. the cilia and the parasomal sacs, forms a highly ordered mosaic.

Development in multicellular organisms is thought to result from differential gene activation of nuclear genes in the ever growing number of individual cells. In unicellular ciliates the intracellular diversification as demonstrated, for example, by the spatial patterning of the membrane domains, can hardly result from differential gene activation.

Gradients of diffusible substances (morphogens) have been postulated to control development in both multicellular organisms and ciliates. On the basis of calculations by Crick (1970), Lynn (1977) has argued that in ciliates smaller than 100 μm, less than 1 min would be necessary to establish a chemical gradient by diffusion. Cyclosis, the rapid movement of the endoplasm, would have to cease for a short period of time (and it really does prior to division). Or, more likely, the gradient would have to appear in the more stable cortical cytoplasm. What really has prevented a more general acceptance of the morphogen concept for ciliates is the fact that no such substance has yet been identified biochemically. An anterior-to-posterior gradient of morphological features in the plasma membrane of a ciliate would also be of great interest for electrophysiologists who have (e.g. in Paramecium)detected a graded distribution of mechanoreceptor channels (Ogura & Machemer, 1980). Obviously ion channels have a morphological substrate but we cannot yet detect it visually.

Once a more detailed knowledge of the structure of the cortex had been acquired it was tempting to attribute a gradient-generating and gradient-maintaining function to the differential assembly of microtubular and microfibrillar systems of the cortex (Sonneborn, 1975). Frankel (1974) favours the view that the basis for morphogenetic gradients lies in the propagation of conformational transitions over the cell surface, possibly through the plasma membrane itself. The membrane-born gradient contours would then exert their positioning effect upon the structures that assemble underneath the plasma membrane. It is difficult to imagine what the morphological substrate of the gradient contours or their imprints in the plasma membrane would look like.

Interphase cells of Cyclidium show no visible gradient in the plasma membrane but an abrupt boundary in the distribution of two qualitatively different membrane domains just at the level of the future fission zone. This raises the question of what happens when the cell divides, when the anterior half of the cell will become the anterior daughter and the posterior half of the cell will become the posterior daughter. Clearly, the typical distribution of the membrane domains will have to be reconstructed somehow during cell division. We may ask whether the simple pellicular rugs in the posterior half of the future anterior daughter transform into complex pellicular rugs? And is there a reverse transformation in the anterior half of the posterior daughter? Or is there a dedifferentiation of all membrane domains followed by a sitespecific redifferentiation? Do the halves of the daughter become established by a process working simultaneously over the entire surface, or is there a wave of changes moving in one or both directions from the fission zone? If there is a wave, then the zone of discontinuity observed in the interphase cells may represent the point of furthest advance of the wavefront. Perhaps, if two wavefronts with different properties (one inducing the simple pellicular rugs, the other inducing the complex pellicular rugs) started from the fission zone and moved in opposite directions, then the situation would resemble the kick-off of a football game with two balls (one for each half of the field), in which the forwards of each team run towards the goal of the opposing team. In both halves of the field the forwards will be stopped by the backs of the opposing team, thus forming two new boundaries. Admittedly, this match will not satisfy all spectators for we have not yet recorded the first minutes of the game. Only further studies with Cyclidium will prove or disprove the postulated rules of the game.

I thank Mrs Anne Hobt and Mr Horst Schoppmann for excellent technical assistance. I also thank Mr Heiner Bauschert for the drawings. This investigation was supported by a grant from the Deutsche Forschungsgemeinschaft.