Development of surface pattern during division in Paramecium; I. Mapping of duplication and reorganization of cortical cytoskeletal structures in the wild type

Paramecium displays an elaborate surface pattern formed by the arrangement of several thousand juxtaposed cortical units. These unit territories, centred around one (or two) ciliary basal bodies are polarized and tandemly aligned in longitudinal adjacent rows (Ehret & Powers, 1959; Pitelka, 1969; Allen, 1971; Sonneborn, 1970a). At the level of the whole organism, this elementary organization is tailored into an asymmetrical and polarized pattern whose major centre of asymmetry is a large ventral organellar complex, the oral apparatus. As paramecia multiply by binary fission, their division implies the reconstruction of the pattern in both daughter cells.

A particular feature of this reconstruction stems from the fact that during division, cell growth mainly proceeds along the anteroposterior axis and relies on an intussusceptive addition of new basal bodies and cortical units along the pre-existing longitudinal rows, so that a ciliate can schematically be represented as a cylinder and a ciliate clone as an indefinitely elongating and segmenting cylinder (Tartar, 1967; Frankel & Nelsen, 1981). However, because of the cell asymmetry and dorsoventral and anteroposterior polarities, the reformation of the pattern cannot be accounted for only by this longitudinal unit proliferation. At the onset of division, an equatorial furrow separates two half cells of different shape and pattern which both regenerate a complete cell by distinct and complementary ‘epimor-phoses’ (see Lwoff, 1950; Aufderheide et al. 1980 and Frankel, 1984 for reviews on ciliate morphogenesis).

Therefore, although development in ciliates does not go beyond clonal perpetuation of the cellular pattern, the division of a Paramecium raises some topological and mechanistic problems similar to those encountered in a developing egg: what sequence of events leads a particular territory of the mother cell to generate a defined part of the daughter cells; which factors and mechanisms trigger and control the spatially and temporally ordered morphological changes (see Frankel, 1982)? However, in contrast to a developing embryo, no differential gene activity progressively triggered in different territories occurs in a dividing Paramecium. Therefore, the divergent fates of different parts of the original cell are due either to differences in the developmental potentialities of the parts or to interactions between the parts or to the action of intracellular signals, triggered by division and controlling the global morphogenesis.

As a first step towards understanding these morphogenetic processes, we undertook to map, throughout division, the major cortical cytoskeletal structures, which (whether repeated in each cortical unit or spanning the whole cell) provide regular landmarks over the growing surface. The prominent surface structures are the ciliary basal bodies whose arrangement and general pattern of duplication have long been described (Dippel, 1965, 1968; Gillies & Hanson, 1968; Sonneborn, 1970a; Pitelka, 1969; Suhama, 1971; Kaneda & Hanson, 1974) and also documented by observations of Chen-Shan and Sonneborn (Sonneborn, personal communication). However, none of these data provided a fully comprehensive analysis of basal body duplication throughout division. The only other cortical structure whose dynamics during division had been reported is the kinetodesmal fibre (ciliary rootlet). Fernandez-Galiano (1980) showed that the old fibres degenerated while new ones developed from old and new basal bodies and that this phenomenon progressed along ‘gradients’ originating from the gullet. As for the other cortical structures, the study of their dynamics required specific staining. Immunological probes (reviewed in Cohen & Beisson, 1988) have recently become available and permitted us to observe by immunofluorescence not only microtubular structures (Cohen et al. 1982) but also the kinetodesmal fibres, the epiplasm (a fibrous layer underlying the cortical membranes and in which basal bodies are anchored) and the outer lattice (a fibrous network sustaining the plasma membrane and whose meshes delineate each cortical unit; Cohen et al. 1987). Using these probes, along with cytochemical staining of basal bodies and kinetodesmal fibres, we have followed the evolution of five major cortical structures throughout division.

We present here, in parallel, spatiotemporal maps of the reorganization of each of these different cortical elements. The most striking result is the visualization of the progressive invasion of the cell surface by superimposed morphogenetic waves, which successively trigger the duplication, assembly or reorganization of each structure and which all spread with the same shape, from the same epicentres. These observations provide new insight into the cascade of events involved in the duplication of cortical units and demonstrate that, despite the local structural constraints of the preexisting surface organization, the development of surface pattern during division is controlled in an integrated manner by transcellular signals. The molecular basis and the mode of propagation of these signals can now be addressed.

All the experiments have been carried out on strain d4-2 of Paramecium tetraurelia, a derivative of the wild-type 51S strain (Sonneborn, 1974). The cells were grown at 27°C in buffered Wheat Grass Powder (Pines International Co., USA) infusion containing 0·5μgml^-1 /ksisterol and bacter-ized the day before with Aerobacter aerogenes. The cells were analysed during the exponential phase of growth.

The cells were treated as described by Cohen et al. (1982). Briefly, they were permeabilized in PHEM buffer (Schliwa & Van Blerkom, 1981) containing 1% Triton × 100 for 5min, and incubated 1−2 h at room temperature in the primary antibody. The cells were then washed three times in PHEM buffer containing 3% bovine serum albumin and 01% Tween-20, incubated 1 h in the secondary antibody (FITC-labelled goat anti-rabbit antibody or sheep anti-mouse Ig antibody, both from Pasteur Production, Paris, diluted 1/200), washed two times and mounted in Cityfluor (London). They were observed under a Zeiss or a Leitz epifluorescent microscope and photographed with Agfapan 100 or Kodak Plus ×125 films.

The following antibodies were used. The microtubular structures were visualized by a rabbit antiserum raised against Paramecium axonemal tubulin (Cohen et al. 1982). In some experiments, the kinetodesmal fibres were labelled by a rabbit antiserum raised against the Paramecium purified structures (Sperling, in preparation). As previously described (Cohen et al. 1987), the outer lattice was decorated by the monoclonal antibody CC212 raised against ciliated cortices of quail oviduct and characterized as an anti-myosin of smooth muscle or nonmuscle cells (Klotz et al. 1986). Finally the monoclonal CTR 211, raised against purified human centrosomes and kindly provided by M. Bornens, was observed by G. Keryer and N. Garreau de Loubresse (personal communication) to label specifically the epiplasm and was used to decorate this structure.

Both polyclonals were diluted 1/500 and both monoclonals were used as culture supernatants diluted 1/2.

The classical silver-staining method of Chatton & Lwoff (1930) was performed as described by Chen-Shan (1969).

The ‘protargol’ technique used to stain kinetosomes was that of Tuffrau (1967), using Wiaçkowski’s (1985) fixative. Briefly, the cells were fixed in a 1:1 mixture of saturated mercuric bichloride and Bouin-Allen’s fixative (Langeron, 1949), rinsed twice in distilled water, quickly destained in diluted hypochlorite, rinsed thoroughly in distilled water, mounted on glass slides coated with glycerinated albumin; the slides were placed in 95 % ethanol, rehydrated and placed for at least 30 min at 60°C in a 0·5% or 1% solution of silver proteinate (Prolabo-France). The slides were then rapidly rinsed, developed and fixed. After dehydration in ethanol, the slides were permanently mounted.

The ammoniacal silver carbonate method of Fernandez-Galiano (1976) was used with minor modifications to stain kinetodesmal fibres. After staining, the cells were included in glycerinated albumin on glass slides, dehydrated and permanently mounted.

Figures 1−5 present images of interphase cells decorated by the different available immunological or cytological stains, each revealing a particular cortical structure. In Figs 1A,B and 2, respectively, basal bodies are stained by an anti-tubulin antiserum applied to a deciliated cell and by protargol: each dot corresponds to a ciliary basal body. While other cortical microtubular structures, e.g. microtubule ribbons or cytospindle, are decorated by the antibody, only basal bodies are revealed by protargol. In Fig. 3, the kinetodesmal fibres, nucleated from the proximal end of basal bodies, are stained by the method of Fernandez-Galiano (1976). They extend longitudinally over two or more units and overlap and intertwine. The length of the fibres depends on the region: they are particularly long in the anterior left ventral field and much shorter on the dorsal surface. Fig. 4 shows the ventral face of a cell decorated by the monoclonal CTR211 (see Materials and methods) which binds to the epiplasm, a fibrous layer underlying the innermost cortical membrane and in which basal bodies are anchored: each ‘scale’ corresponds to a cortical unit.

Cortical units are not flat but cup-shaped with basal bodies anchored at the bottom. The edges of the epiplasmic scales meet longitudinally and laterally along ridges which form a checkered surface relief, not apparent in Fig. 3, but visible on scanning electron micrographs (Small & Marszalek, 1969; Ruiz et al. 1987). The units are held together by the plasma membrane and a fibrillar network, the outer lattice visualized in Fig. 5 by the monoclonal CC212 (Cohen et al. 1987). This lattice is continuous over the whole cell surface and its meshes delineate the periphery of cortical units, i.e. the top of the surface ridges. It can be noted that the patterns of epiplasmic scales and of the outer lattice are strictly complementary as if they were the negatives of each other, in agreement with the electron microscopy data (Allen, 1971).

These views of the different cortical structures illustrate the asymmetry and polarization of the cortical pattern at both whole cell and cortical unit levels. First, not all rows are of the same length or curvature and not all merge at the poles. Second, three aligned singular regions, the oral apparatus and the anterior and posterior sutures, define a right and a left, and distinguish the ventral surface from the more regular dorsal one, marked only by the pores of the contractile vacuoles. All over the cell, asymmetry and polarity are manifested in particular by the kinetodesmal fibres which are all oriented towards the anterior and the right of each unit. Finally, it is important to point out that not all cortical units are identical as they may contain one or two basal bodies and that the distribution of the two types of units (1-bb vs 2-bb) is regionalized as first noted by Sonneborn (1975). A schematic although accurate view of this regionalization is provided by Fig. 6. This camera-lucida drawing of the ventral surface of an interphase silver-impregnated cell shows that there are fields composed only of 1-bb or of 2-bb units and mixed fields in which the two types of units seem randomly distributed.

This precise organization nevertheless is subject to a certain degree of quantitative variation in cell size, total number of units, number of 2-bb units within the 2-bb field, number of rows etc, as shown in Table 1 for eight cells from the same small population.

Before division starts, the contractile vacuole system duplicates, a new pore appearing just anterior to the old one and a new oral apparatus developing on the right margin of the pre-existing one (Roque, 1956). When cells enter division, the fission furrow develops as an equatorial break in the continuity of longitudinal rows of cortical units. The newly formed oral apparatus is integrated into the posterior fission product and the increasing length between old and new oral apparatuses provides an index of the elongation of the dividing cell. Figs 7-11 present dividing cells (or parts of dividing cells) decorated by the same probes as in Figs 1-5. They indicate the level of resolution of the different methods and the salient features of the reorganization or dupli-cation of the different studied structures. The following descriptions leave out the development of the new oral apparatus (see Jones, 1976) and are focused only on the ‘somatic’ (as opposed to ‘oral’) morphogenetic processes.

Fig. 7 shows basal body proliferation. At an early stage (Fig. 7A), this proliferation first leads to groups of three or four tightly packed basal bodies in zones, near the fission furrow, where only one or two basal bodies per unit pre-existed in the interphase cell (see Fig. 1). This first wave of proliferation in fact proceeds in two ‘rounds’. First, in both 1-bb and 2-bb units, one new basal body is generated. In a second round, another new basal body is added yielding the observed groups of three or four. Then, as the cell elongates, the packed basal bodies move away from each other and become equally spaced along the longitudinal axis. Both steps are shown in Fig. 7 (A,a), respectively, anterior and posterior to the fission furrow. At a later stage (Fig. 7B), the following successive phenomena take place. First, as the dividing cell elongates, an increased spacing of duplicated basal bodies spreads anteriorly and posteriorly. Second, in certain regions such as the anterior end of the posterior division product, basal bodies which have become individualized as just mentioned duplicate again (Fig. 7bl) and this proceeds as a second wave. The resulting pairs of basal bodies will not separate and will each be included in 2-bb units. Third, in a restricted area of the left anterior ventral field, preexisting 2-bb units elongate and the two previously paired basal bodies separate (Fig. 7b2). Later on (not shown), each of these basal bodies will duplicate to reconstitute 2-bb units. This zone (Fig. 7C, zone 2), intercalated between an invariant field (zone 3) and a ‘normally’ duplicating one (zone 5), can therefore be singled out by its specific mode of duplication. It most probably corresponds to the situation analysed by electron microscopy by Dippel (1968) and to one of the two modalities of unit duplication distinguished by Suhama (1971). This mode of duplication seems to be the rule in certain ciliates whose cortex is composed exclusively of 2-bb units (Bohatier, 1979; De Terra, 1972). Finally, in well-defined zones of both future daughter cells, no basal body duplication at all will take place. A more comprehensive view of these processes is provided in Fig. 7C, a camera-lucida drawing (like that of Fig. 6) of a silver-impregnated cell at the same stage as in Fig. 7B: the localizations of the just-mentioned duplication processes and of the nonduplicating areas are outlined. Fig. 7D recapitulates the different types and steps in development of units according to their localization on the cell and initial 1-bb or 2-bb status.

In Fig. 8A,B, the organization of the epiplasm is seen to evolve first by longitudinal elongation of the initial scales, then transverse indentations appear that will eventually lead to segmentation into smaller scales around individual basal bodies.

The reorganization of the outer lattice, previously described (Cohen et al. 1987), is illustrated in Fig. 9. The network remains continuous all over the cell, but ‘grows’ first by longitudinal elongation of the meshes (Fig. 9A). Meshes will elongate more near the fission furrow and not at all in a sector of the left anterior ventral field (Fig. 9B, arrows). Then the pattern is reconstituted by budding of new transverse partitions within the elongated meshes.

Fig. 10A,B illustrates the reorganization of kineto-desmal fibres, which is most conspicuous in the equatorial region, near the oral apparatus, where the fibres are long and basal bodies proliferate more. As previously described by Fernandez-Gali ano (1978), the reorganization proceeds in two steps, regression then regrowth. (1) In the equatorial region and around the oral apparatus, old fibres become disconnected from the parental basal body (Fig. 10A, arrows) and will progressively disappear. In the rest of the cell, the actual disappearance of old fibres is more difficult to ascertain. However, old fibres certainly regress, and this is particularly conspicuous in the left anterior ventral field where the fibres are the longest. (2) Short fibres of equal length emanating from all basal bodies, whether old or new, are observed (Fig. 10B) and they all elongate as division proceeds. Identical observations (not shown) were made in immunofluorescence on cells labelled with the antiserum prepared against purified kinetodesmal fibres (see Materials and methods).

Finally, in Fig. 11, decoration by the anti-tubulin antiserum reveals (in addition to basal bodies) the microtubule bundles of the cytospindle (Cohen et al. 1982) which run from pole to pole on the right side of basal body rows, above the kinetodesmal fibres, just beneath the epiplasm (Sundararaman & Hanson, 1976). This massive microtubule network develops when cells enter division and will progressively be disassembled about 10 min after separation of the daughter cells. The cytospindle first appears at the anterior end of the presumptive posterior daughter cell and on the right of the anterior suture (Fig. 11 A, arrows). It then extends posteriorly and anteriorly as discrete segments (Fig. 11B) and eventually forms continuous bundles as illustrated in Fig. 11C.

As summarized in Fig. 12 and its legend, the structures we have studied differ by their mode of duplication, growth or reorganization. These differences call for two comments. First, despite their close apposition, each structure follows a distinctive morphogenetic strategy. Second, all of the observed morphogenetic changes proceed along the anteroposterior cell axis and are channelled by the basal body row pattern.

By studying these types of pictures of cells fixed and stained at successive stages of divisions, we have mapped the progression of these morphological changes throughout division and established their chronology on the basis of reliable stage indices: macro- and micronuclear division, distances between old and new oral apparatuses, length and shape of the dividing cell. Our observations are summarized in Fig. 13 which indicates successively, for the ventral and the dorsal surfaces, the evolution of (a) the cortical and nuclear landmarks; (b) basal body multiplication; (c) epiplasm and outer lattice reorganization, represented as a single series of schemes as they are chronologically and topographically identical; (d) kinetodesmal fibre regression and regrowth; (e) cytospindle assembly. These observations, completed by quantitative data on unit proliferation within the different cell territories (Table 2), lead us to establish the fate map of parental territories (Fig. 14) and ascertain their respective contribution to the surface of the daughter cells.

Examination of Fig. 13 and Table 2 reveals the following facts. (1) Counting of units throughout division (Table 2) provides two types of information. First, there is an overall doubling of units which indicates that the changes in surface organization depicted in Fig. 13 precisely account for the formation of two daughter cells identical to the mother (within the limits of variation recorded in Table 1) without hyperduplication followed by subsequent regulatory processes. Second, the regionalized differences in unit proliferation also precisely account for the development of the different parts from their corresponding presumptive territory, without any displacement of units after their multiplication, which suggests that pattern formation is coupled to the ‘program’ of unit proliferation. (2) Although basal body and cortical unit numbers globally double, there are precisely delineated invariant regions in which no basal body duplicates and no change occurs in the epiplasmic scales or outer lattice meshes at any stage of the division process. These regions, left blank on Fig. 13B, represent approx. 10% of the cortical units. The remaining units and basal bodies duplicate once or more than once. (3) Basal body duplication proceeds throughout division, from stage 1 to stage 8. Two successive waves of different morphogenetic significance can be distinguished. The first wave is complex and involves a first round of duplication which generates one new basal body in each 1-bb or 2-bb unit. Before this first round has invaded the whole dividing cell, a second round of duplication (Fig. 13B, stages 3, 4) occurs in units localized in a central zone of the ventral surface and a thin equatorial band on the dorsal surface. This round adds one or two basal bodies. By stage 5, a second wave of duplication begins which mostly affects the anterior part of the posterior daughter cell and a sector of the anterior one and scattered 2-bb units over the rest of the cell (with the exception of the posterior invariant 1-bb zones).

The distinction between the two waves is based on the fact that by the end of the first wave (stage 4), each of the basal bodies in the units which have been affected by this first wave will organize a new unit, defined by one epiplasmic scale and one kinetodesmal fibre. These units will be progressively completed, but no additional unit will be formed. This was demonstrated by counting kinetodesmal fibres and epiplasmic scales whose number, after completion of the first wave, is precisely equal to the adult complement of the two daughter cells. Therefore, further basal body duplication beyond stage 4 only completes the units which, in the adult cell, have two basal bodies (but a single kinetodesmal fibre). It then can be concluded that the first wave of basal body duplication appears instrumental in generating new units, while the second wave only concerns the reconstruction of 2-bb units in the remodelled areas. The existence of two steps in basal body proliferation, the second one resulting in completing 2-bb units, has already been suggested by Gillies & Hanson (1968) in their analysis of silver-stained P. trichum. (4) Although all duplicating units undergo the same events (basal body duplication, epiplasmic scale elongation and fragmentation, outer lattice longitudinal elongation and transverse partitioning, kinetodesmal fibre regression and regrowth), these events do not occur in the same order in all units, and four types of units can be distinguished as schematized in Fig. 15. The distinction takes into account (1) the fact that basal body duplication can precede or follow local reorganization of epiplasm and outer lattice, depending on whether basal body duplication pertains to the first or second wave, and (2) the fact that kinetodesmal fibres are reorganized even in invariant units. Altogether, basal body duplication, kinetodesmal fibre regeneration and epiplasm and outer lattice reorganization proceed independently and only outer lattice and epiplasm development are always strictly coordinated. (5) At the wholecell level, multiplication, reorganization or development of all structures, whether repeated in each unit (basal bodies, kinetodesmal fibres, epiplasmic scales) or spanning the whole cell (outer lattice, cytospindle), progress as waves of similar shape and originate from the same epicentre(s). Starting from around the oral apparatus, the waves first develop essentially posterior to the fission furrow. In addition, at least for basal bodies and cytospindle, an early progression over the anterior right side is observed as shown in Fig. 11A. The waves then progress laterally from both sides of the gullet and extend towards the poles. The different waves arise successively in a precise order. Basal body multiplication starts first (stage 1), then simultaneously (stage 2) cytospindle assembly, epiplasm and outer lattice elongation begin, kinetodesmal fibres start regressing (stage 3) and regrowing at the same time (stage 5) as the second wave of basal body multiplication begins. This latter wave (as well as those of kinetodesmal fibre regrowth, epiplasm fragmentation and outer lattice partitioning) differs from all previous ones in that it has two epicentres, one (the fission furrow) in the posterior fission product, the other one (the anterior pole) in the anterior one. Thus, at this stage, the dividing cell is no longer a single morphogenetic unit but consists of two partially autonomous morphogenetic entities. In each presumptive daughter cell, the final round of basal body duplication progresses towards the posterior pole. (6) Finally the fate map (Fig. 14), established from the data summarized in Fig. 13, illustrates the qualitatively and quantitatively unequal contribution of the different territories of the mother cell to the organization of the daughter cells.

The work reported in this paper is a detailed description of duplication and remodelling events occurring within the surface cytoskeleton (‘cortex’) of Paramecium during binary fission. The cortex is indeed the site of major morphogenetic remodelling during division in ciliates. The extent to which these surface events are correlated, or even perhaps triggered, by internal modifications is not addressed by the present work and will not be discussed. Previous studies on basal body duplication in Paramecium (Dippell, 1965, 1968; Gillies & Hanson, 1968; Sonneborn, 1970a; Suhama, 1971; Kaneda & Hanson, 1974) long ago led to an awareness that the development of surface pattern and cell shape during division involves mechanisms acting both at the level of individual cortical units and at the level of the whole organism, since multiplication of units (1) is channelled along longitudinal rows and therefore depends upon longitudinal and lateral unit-unit interactions and (2) is neither synchronous nor uniform but obeys a spatiotemporal order integrated at the whole-cell level. However, major pending questions, that also apply to the developing metazoan embryo, are to delineate the respective role in morphogenesis of the mechanisms acting at the level of elementary parts (cortical units/ cells) and of those acting at the level of the organism (Paramecium / embryo) and to identify the nature of these mechanisms. An approach to these questions is provided here by the mapping of the development throughout division of five major cortical cytoskeletal components, which establishes the sequence of events both within each unit and at the whole-cell level. By comparison with the most detailed previous analyses by Gillies & Hanson (1968) and Suhama (1971) on P. trichum, the major new data presented here concern the exact delimitation of invariant fields, a more precise chronology of basal body duplication and, because of the possibility of following in parallel the different components of cortical units, a comprehensive description of the relationships between basal body duplication, unit multiplication and pattern formation. After analysing the bearing of this new information on morphogenetic processes at the two levels, units and whole cell, we will discuss the possible nature and interplay of the underlying mechanisms and briefly allude to the significance of our data with respect to developmental processes in other systems.

Except for the fact that they may contain either one or two basal bodies, all units are similarly organized and of roughly the same size. However, units differ in two respects: their contribution to the formation of the daughter cells and the steps of their reorganization. As duplication of basal bodies and units is strictly channelled along the rows, basal body and unit lineages can be traced (see also Appendix). It is therefore possible to address the question of whether basal bodies (and units) have differentiated properties -that would be transmitted to their progeny -with respect to duplication potential and determination of the overall pattern.

The fate map (Fig. 14) shows that the presumptive territories differ quantitatively and qualitatively in their contribution to the surface of the daughter cell. In this differential expansion, the fate of units is independent of their lineage and coordinated at a supraunit level within the presumptive territory. This coordination first affects duplication activity of units: no duplication occurs in the invariant regions whose size and pattern are stable and linearly transmitted to a single fission product; in the other regions, one or more rounds of duplication takes place and an important point is that, conversely, invariant regions are generated from regions of hyperproliferation. Second, with the exception of the invariant fields, presumptive territories generate daughter territories whose overall organization does not resemble theirs. Therefore, whatever mechanisms underlie the shaping of the rows (length, curvature) and the formation of local differentiations (1-bb vs 2-bb units, oral apparatus cytoproct, sutures), the global pattern does not rely on heritable properties of particular units or files of units. The ‘progeny’ of cortical units and their role in the formation of the pattern of the daughter cells then mostly depend on their localization on the cell. This point is further illustrated and discussed in the Appendix. The same conclusion has been drawn from previous observations on Tetrahymena (Nanney, 1967) and Paramecium (Beisson & Sonne-born, 1965) showing that, due to an unexplained ‘slippage’ of rows around the cell, a particular row can contribute to any part of the cellular pattern.

Units also differ in the cascade of morphogenetic events they undergo. As all units are composed of the same closely apposed structures, it could have been expected that, within each unit, the same sequence of morphogenetic processes would occur. This is not the case. As summarized in Fig. 15, the sequence varies with the region of the cortex: basal body duplication can precede or follow epiplasmic and outer lattice elongation, kinetodesmal fibre regression and regrowth take place in nonduplicating units as does the assembly of the cytospindle which proceeds along both invariant or duplicating regions. The only two structures which always grow and rearrange coordinately are the epiplasm and outer lattice meshes (Cohen & Beisson, 1988), as previously observed in both wild-type and a mutant defective for basal body duplication. Therefore, although the organization of these different structures is certainly constrained by local interactions within the unit, their duplication, growth or reorganization are controlled at the whole-cell level.

Altogether, it appears that all cortical units are potentially equivalent and that their different fates are controlled by their position on the cell.

As shown in Fig. 13, for all structures, whether repeated in each unit (basal body, kinetodesmal fibre, epiplasmic scales) or spanning the whole cell (cytospindle, outer lattice), duplication, remodelling or growth progresses, over the dividing cell, as successive waves which all originate from the oral apparatus. This initial epicentre is soon relayed by the fission furrow from which the waves progress towards both anterior and posterior poles. The oral apparatus, whose duplication is the first event in cortical morphogenesis (Kaneda & Hanson, 1974; Berger, 1988) has long been recognized together with the suture line along which it lies as the morphogenetic centre of ciliates (Tartar, 1967; Sonneborn, 1970b), but the role of the fission furrow, which is seen to develop as a secondary inductive zone, had not previously been pointed out. Frankel et al. (1981) have previously shown, however, that in Tetrahymena a wave of ciliation of naked bosal bodies occurs during division, emanating from the fission zone and propagating posteriorly. They have stressed the fact that the fission furrow of ciliates behaves like a segmentation boundary in metazoa.

The different waves differ in the stage at which they start and by their final extension (only the cytospindle and the kinetodesmal fibre waves pervade the whole cell), but they all share the following features: their common epicentres, their continuous progression, the asymmetry of their initial contour and the similarity of these contours. It is quite striking that these features are precisely shared by the reorganization during division of another cytoskeletal network, the infraciliary lattice which spans the whole cell at the ecto-endoplasmic boundary (Garreau de Loubresse et al. 1988). Although this network runs at the level of the proximal end of basal bodies, its meshes are irregular and show no 1:1 relationships with cortical units and no physical connection with basal bodies are visible in electron microscopy (Allen, 1971; Garreau de Loubresse et al. 1988). Nevertheless, its reorganization (by disassembly and reassembly) is triggered by signals whose origin and progression are similar to those that affect the outermost cortical structures described here. It would then seem that whatever the nature of the signals, they can affect physically unrelated structures over an approx. 0·5 μm thickness of the ‘cortex’.

The continuous progression of the waves suggests either the diffusion of a signal or cooperative transduction of a biochemical change of some component of the cortex.

As for the differences between the waves, they provide two types of information. (1) The limited differences in their initial contour and their different final extension confirm the previously discussed relative independence of the different structures. (2) The final extension of the waves leads us to distinguish two types of morphogenetic processes. The assembly of the cytospindle and the reorganization cycle of the kinetodesmal fibres (whose waves extend over the whole cell) seem to play a scaffolding role, while the other waves seem instrumental in the development of the pattern. Indeed, it can be pointed out that kinetodesmal fibres regress mostly after basal body duplication, which suggests a role of the old fibre in the positioning of new basal bodies; the cytospindle is likely to maintain row organization while cell growth is accompanied by a looser unit cohesion. In contrast, the deployment of the other waves represents the growth of the cortex whose pattern will not be remodelled afterward, as if this deployment were precisely programmed in the preexisting spatial organization or as if, in the course of their progression, the morphogenetic signals were interpreted differentially according to the position of the different territories on the cell.

Ciliates that differ from each other by wide evolutionary distances (Lynn & Sogin, 1988; Baroin et al. 1988) display two distinct developmental strategies to reproduce their oral and somatic basal body arrays. While in Paramecium all basal bodies are conserved and new units are individually organized within the frame of the pre-existing ones, other ciliates like Oxytrichids resorb and reorganize most of their basal body arrays at each division. Paramecium develops its new oral apparatus close to the old one and cannot regenerate it when it is lost, while most other ciliates develop their new oral apparatus at a site quite distinct from the old one and can regenerate it. In contrast to Stentor which can reconstruct its whole cellular pattern from a small piece of cortex (Tartar, 1960), amputated paramecia can only recover their shape slowly through several successive divisions (Tartar, 1954; Chen Shan, 1969, 1970; Suhama, 1975). The ciliates serving as major experimental organisms (reviewed in Frankel, 1984) have therefore been used either to illustrate the importance of pre-existing local patterning (‘cytotaxis’ in Paramecium, see Sonneborn, 1963,1970a) or, in contrast, to stress the possibility of long-range global regulation (which is more readily observed in Stentor and Hypo-trichs). A distinction akin to that made in embryology between ‘mosaic’ and ‘regulative’ eggs could therefore be used to characterize ciliate morphogenetic strategies.

In this study, we observe that in Paramecium (an extreme case of mosaicism) the two major factors in the expression of the developmental program are (1) transcellular signals that induce the reorganization, assembly and duplication of structures and (2) differential responses of the various surface territories to the propagation of the signals. Such a differential response according to position may either support the concept of positional information (Wolpert, 1969) as it has been extended to ciliates (Frankel, 1974) or indicate the role of local pre-established differentiations.

In the first hypothesis, the different territories would be structurally identical but would have a means of assessing their position over the cell surface. This assessment is usually thought to rest on the graded distribution of a ‘morphogen’. If this were the case, one would have to account for the highly asymmetric shape of the gradient boundaries. Alternatively, the morphogenetic signals could spread homogeneously and it is the different territories that would respond differently because of pre-established distinctive properties. Some support for this notion can be found in the fact that different areas of Paramecium cortex are known to display different physical properties. For instance, the anterior left surface (which corresponds precisely to a territory responding differentially to the morphogenetic waves, see Fig. 13), is the sturdiest portion of the cortex as seen in a variety of cell-breakage experiments. In both hypotheses, the question of when and how the reference points are established (or the local differentiations installed) is raised. Some direct evidence as well as logical reasoning suggest that the set up for division n is determined during the course of division n—1. Our current work is aimed at clarifying these steps formally and also at understanding the possible molecular basis of the transcellular signals.

We thank our colleagues M. Bornens and C. Klotz for their gift of the monoclonals CC212 and CTR211 and L. Sperling for the antiserum against kinetodesmal fibres. We gratefully acknowledge the assistance of N. Narradon and J. Brizard in the preparation of the manuscript. We thank Dr Linda Sperling for critical reading of the manuscript.

This work was supported by grants from the Centre National de la Recherche Scientifique, the University Paris-Sud and The Ligue Nationale Française contre le Cancer and by a fellowship from the Ministère des Relations Extérieures to A.T.R.

In the preceding account we showed that the cortical morphogenesis of Paramecium mainly relies on a precise pattern and time course of basal body duplication, and our observations lead to two major conclusions. First, although the number of cortical units and basal bodies globally doubles as can be expected (see Table 2), not all regions of the cell participate equally in this doubling as summarized on the fate map (Fig. 14). Second, we demonstrate that basal body duplication proceeds in two waves of different morphogenetic importance: while the first wave provides all the new basal bodies required for the overall doubling of cortical units, that is the reconstruction of the gross pattern in the two daughter cells, the second wave only adds one basal body in those units that display two basal bodies in the adult cell, and therefore represents a fine tuning in the reconstruction of the surface pattern.

These conclusions are based upon an analysis carried out at the whole-cell level. However, as each new basal body is inserted within the same row as its mother, each row constitutes an ‘autonomous’ unit of elongation and cleavage. Furthermore, rows differ by their length, curvature and pattern of 1-bb vs 2-bb units. It then seemed of interest to establish basal body lineages, that is to analyse morphogenetic processes at the level of individual rows.

Fig. 16 and its legend explain how we established basal body lineages. Starting from the actual pattern of a particular row (here row number 4 on the right side of the gullet on the stage-5 cell shown in Fig. 7C), we first deduced the interphase pattern of the row in the mother cell. This deduced pattern, to be compared to its actual counterpart (from Fig. 6), is represented in column A. Then, knowing the position of the fission furrow (visible on Fig. 7C and presented in Fig. 16 column B) and knowing the fact that beyond stage 5 only second wave basal bodies appear, we could not only deduce the pattern of the rows in the two daughter cells (column C) but also identify on the stage-5 row (column B) old and new (first or second wave) basal bodies.

Similar pattern reconstructions were carried out for four additional rows on the right ventral side and six rows on the left. The information is mapped on Fig. 17 which shows how the sites of presumptive new first wave and new second wave basal bodies are located on the mother cell (A,A’) and eventually distributed on the daughter cells (C,C’).

Finally, the calculations presented in Fig. 16 and its legend permitted us to identify the putative position of the fission furrow on interphase cells. This position, known for a dividing cell (Fig. 16B) can be indicated on the deduced mother pattern (Fig. 16A), according to the lineages previously established and on the deduced daughter patterns (Fig. 16C), since they are also interphase ones. By tracing back the basal body lineages from C to A, we can now localize the position of the second generation fission furrows (/3 and y) on the dividing cell and its interphase mother. Fig. 18 represents the positions of these deduced putative furrows on actual cell drawings after extension of the calculation to several rows (see legend).

These mappings first confirm, as would be expected, that alpng each row, as at the whole-cell level, the number of basal bodies globally doubles despite of regionalized differences in duplication activity. Furthermore, the visualization of the precise position of future first wave and second wave basal bodies on the mother cell (Fig. 17C and D) provides a finer view of the fields particularly active in morphogenesis, in which, for instance, molecular mechanisms responsible for these waves should be looked for. More interestingly, this study reveals a fact that could not be detected by the observations at the whole-cell level, namely that the position of the fission furrow and the location of new basal bodies (Figs 17 and 18) yields an asymmetry in basal body transmission to the daughter cells: the anterior cell product receives more old basal bodies (and fewer new ones) than the posterior one, especially in the left ventral half.

In addition, the rules described in the legend of Fig. 16 are easily computerizable thus opening a new way for analysing cell morphogenesis. Indeed, (1) modifications of either the patterns or the rules of duplication can be introduced and their morphogenetic consequences after division analysed; (2) basal body lineages can be followed over many cell generations by the computer; (3) mutants that display abnormal basal body duplication, such as sm!9 (Ruiz et al. 1987), or abnormal cortical patterns could be studied by such a computerized analysis in order to determine the precise level of their mutational defect.