Uncoupling of basal body duplication and cell division in crochu, a mutant of Paramecium hypersensitive to nocodazole

Ciliates are unicellular organisms whose polarized and asymmetrical shapes and elaborate surface patterns are best described by the arrangement of their ciliary basal bodies. As is well documented for Paramecium, the cell cortex is a mosaic of cortical domains with distinct morphogenetic potentials. Cell division is accompanied by a complex developmental process that includes a precise spatial and temporal program of basal body duplication (Iftode et al., 1989). As a consequence, mutations affecting basal body duplication result in abnormal cell morphologies and can easily be screened (Jerka-Dziadosz and Beisson, 1990). These mutations provide not only the opportunity to analyze the role of basal bodies in pattern formation, but also the possibility of identifying factors which control basal body duplication. Such factors might not be specific for ciliates. Indeed, ciliary or flagellar basal bodies and centrioles in centrosomes are similar organelles: they share the same basic ultrastructure and the remarkable capacity to duplicate; they also share another physiologically important property: their duplication is generally strictly coupled with cellular division. Mutations affecting these properties in ciliate basal bodies and whose counterpart would be lethal in multicellular organisms (Freeman et al., 1986) may then reveal common conserved mechanisms.

We describe here a mutation of Paramecium which uncouples basal body duplication from the cell cycle. Despite the fact that the duplication of basal bodies and centrioles normally occurs at a fixed stage of the cell cycle, uncoupling of the two processes has been reported to occur under particular physiological, experimental or genetic conditions. In metazoa, many cycles of centriole/centrosome duplication can be triggered in enucleated oocytes of Xenopus (Gard et al., 1990), sea urchin (Sluder and Lewis, 1987) or starfish (Picard et al., 1988), or in Drosophila eggs whose nuclear divisions are blocked by aphidicolin (Raff and Glover, 1988) or by mutation (Freeman et al., 1986) or in mammalian cells arrested in division by hydroxyurea (Balczon et al., 1995). In ciliated epithelia, basal body proliferation takes place as a terminal differentiation step, in non dividing cells (Anderson and Brenner, 1971). In unicellular eukaryotes also, the basal body cycle may proceed independently of the cell cycle, as in Naegleria gruberi (Fulton and Dingle, 1971) or in Physarum (Burland at al., 1993), where flagellar basal bodies assemble in nondividing amoebae; conversely, in Chlamydomonas reinhardtii (Ehler et al., 1995) or in Paramecium (Ruiz et al., 1987), mutational defects in basal body duplication do not preclude cell division or can lead to uncoupling of basal body duplication and the cell cycle (Adams et al., 1985). These various examples demonstrate that specific factors control centriole and basal body duplication and that the generally observed coordination with the cell cycle involves additional factors. However, none have yet been identified.

In Paramecium, we have previously studied two mutations, sm19 (Ruiz et al., 1987) and kin241 (Jerka-Dziadosz et al., 1992) which cause, respectively, reduced or enhanced basal body duplication activity and yield abnormal morphologies, but do not alter the global coupling of basal body duplication with cell division. In contrast, this coupling is impaired by the cro1 mutation described here. In cro1 cells, basal bodies respond to cell division signals but their proliferation does not stop after division and continues throughout the cell cycle, yielding highly abnormal cortical patterns and cell shapes. The mutation also causes hypersensitivity to nocodazole and behaves as an extragenic suppressor of another mutation, noc^r1 which confers resistance to nocodazole (Torrès et al., 1991) and is localized in a β-tubulin gene (Dupuis, 1992; Dupuis- Williams et al., 1997), suggesting a primary effect of the mutation on microtubule properties. The properties of the cro1 mutation provide a promising tool for future analysis of the mechanisms of coupling between basal body assembly and cell division and more specific information on the control of pattern formation in Paramecium.

The strains used here were the wild-type (wt) strain d4-2 of Paramecium tetraurelia, a derivative of stock 51 (Sonneborn, 1974) and the following mutants, all derived from stock d4-2. The mutant cro1, isolated by Beisson and Rossignol (Sonneborn, 1974) is described here. The mutants noc^r1 (Torrès et al., 1991), sm19 (Ruiz et al., 1987), previously studied, are characterized by resistance to nocodazole and defective basal body duplication, respectively. In addition, two mutations were used as genetic markers in crosses: ts111 whose thermosensitive expression causes cell death at 35°C (Beisson and Rossignol, 1969) and nd7-1which prevents trichocyst discharge by blocking the last step (membrane fusion) of the secretory pathway (Cohen and Beisson, 1980).

Cells were grown at 27°C in buffered Wheat Grass Powder (Pines International Co. USA) infusion containing 0.4 μg/ml β-sitosterol and bacterized the day before use with Klebsiella pneumoniae. Nocodazole (from Janssen Life Science Products, Piscataway, NJ) was prepared as described previously (Torrès et al., 1991).

Genetic analysis was carried out according to standard procedures (Sonneborn, 1970). For each cross, 20-30 pairs of conjugants were isolated, the exconjugants separated and the phenotype of the corresponding F₁ clones studied to ascertain heterozygosity. After autogamy, which involves a meiotic reassortment of the parental genes and restores homozygosity at all loci, 30 ex-autogamous F₂ clones from each of two pairs of selected F₁ clones were studied for segregation of the parental genes. The following crosses were studied: cro1 × ts111 : nd7-1 ; cro1 × noc^r1 and cro1 × sm19.

Immunolabeling was performed as previously described (Iftode et al., 1989). Except in some cases when cells had to be deciliated by two successive 5 minute exposures to 2% MnCl₂ (Fukushi and Hiwatashi, 1970) prior to permeabilization, the first step was always permeabilization in PHEM buffer (Schliwa and van Blerkom, 1981) containing 1% Triton X-100 for 5-10 minutes. Cells were then washed in a modified TBST, called TBST^*(0.15 M NaCl in 1 0 mM Tris-HCl pH 7.4, 0.1% Tween 20, 2 mM MgCl₂ and 1 mM EGTA). TBST^* containing 3% BSA (bovine serum albumin) was used at all steps thereafter. Cells were incubated for 10 minutes to 1 hour at room temperature in the primary antibody, washed three times and incubated for 1 hour in the secondary antibody (FITC-labeled goat anti-rabbit from Institut Pasteur production, Paris), or rhodamine- labeled sheep anti-mouse (from Jackson ImmunoResearch Labs,West Grove, Pennsylvania), then washed twice and mounted in Citifluor (Citifluor Ltd, England).

In some experiments, after permeabilization, cells were fixed in 2% paraformaldehyde in PHEM for 10 minutes to 1 hour at room temperature, then they were washed first in TBST^* then in TBST^*BSA and further processed as above. For immunodecoration of the infraciliary lattice, a slightly modified protocol was used: incubation in PHEM-Triton was reduced to 1.5 minutes, followed by fixation as above and all subsequent steps were carried out in standard TBST supplemented with 5 mM CaCl₂.

Immunolabelled cells were observed under a Zeiss epifluorescence microscope and photographed using Kodak TMAX 400 film.

To decorate microtubular structures one of the following antibodies was used, depending on the experiment. (1) A rabbit antiserum raised against Paramecium axonemal tubulin (Cohen et al., 1982) permitted us to visualize most microtubule networks on deciliated cells. (2) A fairly selective decoration of basal bodies was obtained by the polyclonal anti-αGlu-tubulin (aT12) and the monoclonal anti-αTyr- tubulin (a2), respectively directed against the detyrosinated and tyrosinated C-terminal part of α-tubulin (Kreis, 1987) kindly provided by Dr Kreis, as well as by the monoclonal anti-α-tubulin ID5 (Wehland and Weber, 1988), generously supplied by Dr Wehland. (3) With a monoclonal antibody directed against a most conserved sub- C-terminal part of α-tubulin (from Amersham, Bucks, UK), all tubulin-containing structures or only particular subsets could be decorated, depending on the duration of fixation (Beisson, unpublished observations). The striated ciliary rootlets (kinetodesmal fibers) were labeled with a rabbit antiserum raised against the purified Paramecium structure (Sperling et al., 1991). The epiplasm scales were labeled with the polyclonal ‘anti-band 4’ antibody (Keryer et al., 1990) and the infraciliary lattice by a monoclonal anti-centrin antibody, 20H5 (Erabolu et al., 1994), a gift from Dr Salisbury.

Cells were fixed and processed by standard methods as described before (Jerka-Dziadosz et al., 1992) and examined with a JEM 1200 EX electron microscope.

The mutation crochu (cro1) displays pleiotropic effects on cell shape, cell size, generation time and most strikingly on the cortical pattern, characterized by increased and disordered basal body proliferation. The crochu phenotype, expressed at all temperatures (from 13 to 35°C), is variable. Even in small clones, a range of morphologies is observed, from nearly normal to extremely abnormal. Typically the cells present an elongated, thin and often hooked (crochu) anterior part (Fig. 1) and their length averages 1.6 times that of wild type. Their generation time is longer than that of wild type, approx. 7 hours vs approx. 5 hours at 28°C. Micronuclear and macronuclear divisions proceed normally but occasional unequal distribution of nuclei between the daughter cells occur, especially in the most abnormal cells which may yield nonviable progeny with either a defective oral apparatus or a missing macronucleus.

The mutation also causes hypersensitivity to nocodazole, which disassembles microtubules (De Brabander et al., 1976). In wild-type Paramecium, at concentrations above 4 μM, the drug causes immediate disassembly of the most labile microtubule networks, i.e. the intracytoplasmic network in interphase cells and, during division, the microtubule bundles which ensure separation of micronuclei (separation spindles) and elongation of the macronucleus, resulting in cell division arrest (Cohen et al., 1982; Torrès and Delgado, 1989; Torrès et al., 1991). cro1 cells display the same immediate response (not shown) but are also hypersensitive to the drug. Their division is practically arrested at doses (4 μM) which still allow wild-type cells to undergo a few (although abnormal) divisions. Furthermore, cro1 cells soon die, while wild-type cells can survive for 2-3 days after arrest of division. The mutant kin241 (Jerka- Dziadosz et al., 1992) responds like wild type (not shown), indicating that this hypersensitivity is not a common property of slow growing mutants with large cell size and cortical disorders. As cro1 cells respond like wild-type (Ruiz et al., 1976) cells to vinblastine, another microtubule depolymerizing drug, it can be concluded that the mutation specifically affects nocodazole interaction with microtubules.

The cro1 mutation is recessive. The cro1/cro1⁺ heterozygotes recover a normal size, morphology, growth rate and nocodazole-sensitivity within a few divisions after conjugation. Conversely, after autogamy of the cro1/cro1⁺ heterozygotes, cro1/cro1 homozygotes of either mutant or wild-type cytoplasmic origin fully display the crochu phenotype after a few divisions.

The study of the mutation focused on an immuno- cytochemical analysis of the nature and development of the cortical disorders and a genetic analysis of cro1 interactions with other mutations affecting basal body duplication or nocodazole resistance.

The wild-type cortical organization has been fully documented using several antibodies specifically binding to each of the cortical cytoskeletal structures: basal bodies (Fig. 2A), ciliary rootlets, outer lattice, epiplasm and infraciliary lattice (Cohen and Beisson, 1988; Cohen et al., 1987; Garreau de Loubresse et al., 1988; Iftode et al. 1989; Sperling et al., 1991). Cortical units, delineated by the meshes of the outer lattice, are formed by the epiplasmic scales in which basal bodies are anchored. Cortical units contain either one or two basal bodies (1bb and 2bb units respectively). The distribution of these two types of units is regionalized (Sonneborn, 1975; Iftode et al. 1989) into three different fields (Fig. 3, upper panel): a homogeneous 1bb field (A); a homogeneous 2bb field (B); a mixed field where 1bb and 2bb units coexist in apparently random combination (C).

The cro1 mutant has a range of cortical abnormalities, which differ not only among cells of a population but even between sister cells examined soon (40 minutes) after the division of their mother cell (Fig. 2B,B′,C,C′). Despite this variability, four major features define the cro1 phenotype. (1) The presence of 3-4bb units and of dense files of basal bodies in interphase cells (Figs 2 and 4) indicates enhanced basal body proliferation, all the more striking as it occurs also in the anterior left 2bb ‘invariant field’ of wild-type cells (Iftode et al. 1989) in which no basal body duplication takes place. (2) The distribution of the different types of cortical units is abnormal. It differs not only from that of wild type (Fig. 3), but also from that of the mutant kin241 (Jerka-Dziadosz et al., 1992) in which the geography of the 1-bb, 2- bb and mixed fields is altered. In cro1, it is the constitution of the fields which is affected as shown in Fig. 3 (lower panel). (3) The number of ciliary rows and cortical units is increased. This was quantified by counting basal bodies and cortical units on photographs of cells stained with anti-tubulin antibodies, as was done previously for wild-type cells (Iftode et al., 1989). Table 1 presents the data collected from 5 cro1/cro1 and 4 heterozygous cro1/cro1⁺ cells of either cro1 or wild-type cytoplasmic origin. These data point out (i) the increased number of ciliary rows (79-91) as compared to approx. 70 for wild-type cells, in particular along the anterior suture where 31-37 basal bodies is perturbed. The disorders are most apparent in the anterior ventral surface and in the oral apparatus, but occur also at other random cortical locations. In many cells, additional short rows do not extend from pole to pole (Fig. 2B,C). Inverted rows, either single or in patches of 2-3, are most frequently seen on the right and dorsal sides. Most strikingly, nodules of a few to many basal bodies are seen beneath the surface (Fig. 4A,B) and retain the capacity to nucleate new basal bodies (Fig. 4C), as confirmed at the ultrastructural level. Fig. 4 shows fully elongated basal bodies, closed at the distal end by a terminal plate, located outside organized units, parallel to the surface, or in random orientations. Non-ciliated basal bodies were seen docked to the inner alveolar membrane or at other ectopic locations. These abnormally located basal bodies often develop the accessory ciliary rows on the left side abut 18-28 rows on the right side; (ii) the overall increased number of cortical units and of basal bodies; (iii) the frequency of abnormal 1bb, 3bb, 4bb units and (iv) the normalization of the pattern in the heterozygotes cro1/cro1⁺. (4) The alignment of rootlets (not shown). Regardless of their location and orientation, all mature basal bodies show a normal ultrastructure. Thus, unlike sm19 (Ruiz et al., 1987) and kin241 (Jerka-Dziadosz et al., 1992), the cro1 mutation does not affect the organization of basal bodies or ciliary axonemes.

Other cytoskeletal structures were examined: outer lattice, epiplasmic scales, infraciliary lattice. Abnormalities in their organization, whenever observed, colocalized with abnormalities in the basal body pattern.

In wild-type cells, basal body duplication and cortical remodelling follow a precise spatiotemporal program (Iftode et al., 1989). Basal body duplication starts from the oral apparatus and the fission line and progresses towards both anterior and posterior poles as a wave correlated with the elongation of the cell, separation of old and new oral apparatus, elongation of the dividing macronucleus and of the micronuclear separation spindles. Basal body duplication is not uniform but precisely regionalized. There are two ‘invariant’ fields in which no basal body duplication occurs. The anterior ‘invariant’ field is inherited as such by the anterior product of division. The posterior ‘invariant’ field is inherited by the posterior division product. In other equally well defined regions, basal bodies duplicate either once or more than once. For those regions in which basal bodies duplicate more than once, two successive waves occur: by the end of the first wave the number of cortical units in the two daughter cells is fixed; the second wave, which yields a second basal body per unit, is completed before the two daughter cells separate. No further basal body duplication will occur and the subsequent steps of morphogenesis involve only surface growth and reconstruction of the pattern of the other cytoskeletal structures: epiplasmic scales, outer lattice, infraciliary lattice, ciliary rootlets.

Fig. 6 compares wild type and mutant at a similar stage of division and shows that basal body duplication in cro1 cells globally follows the same spatiotemporal pattern as in wild type (A). However, in the mutant, basal body duplication seems to be enhanced in the regions where ‘hyperduplication’ occurs in wild type and in the newly formed oral apparatus whose basal body rows and anarchic field are often enlarged and disorganized (not shown). At other sites, some basal bodies exhibit random orientations and form short or bent stretches, eventually leading to intercalation of additional ciliary rows of either normal or reversed polarity. Despite this enhanced proliferation during division, it is striking that the most conspicuous abnormality, overgrowth of the anterior part of the cell and the presence of 3–4 basal body units in the anterior left field (aLF) does not arise during division. Indeed, this field, neoformed in the posterior division product, was consistently found to be normal at mid and late division stages (Fig. 6B) as well as in posterior division products fixed 40-60 minutes after division (Fig. 2B).

As for the other cortical cytoskeletal structures, their waves of duplication/reorganization do not deviate from the wild-type pattern (Iftode et al. 1989). When present, local disorders (for instance in the infraciliary lattice in the aLF, not shown) colocalize with the abnormal basal body pattern and most likely are secondary effects of the mutation. In particular, the striated ciliary rootlets (cr in Fig. 5) undergo the characteristic regression-regrowth sequence described in wild-type cells (Fernandez-Galiano, 1978; Iftode et al., 1989; Sperling et al., 1991), except that in cro1 short temporary ciliary rootlets also regrow from anterior bbs in units located in the aLF (Fig. 5B), suggestive of abnormal morphogenetic activity of the basal bodies.

If the most conspicuous abnormalities of cro1 cells (their elongated anterior part displaying cortical units with up to 4 bbs) are not due to overproliferation occurring during division, they might be due to basal body duplication occurring in interphase. To ascertain this fact, a kinetic study was carried out. Pools of dividing cells were isolated, then fixed and immunostained 40-60 minutes or 4 hours after completion of division. Fig. 7 shows that the percentage of cells exhibiting abnormal anterior fields increases as a function of time. These quantitative data, along with the fact that, in the posterior division product, overproliferation in the aLF does not occur during division, demonstrate that basal body proliferation continues in interphase. This fact accounts for two observations: firstly, the ‘nodules’ of basal bodies present below the cell surface (Fig. 4), as the insertion in the cortex of newly formed basal bodies is most likely to be dependent upon the longitudinal growth of the cortex that occurs essentially during and soon after division; secondly, the atypical organization of cortical units seen on electron micrographs (Fig. 5A). The invariant posterior field (or at least a large part of it), composed of 1bb units, is the sole cortical territory which escapes the crochu syndrome.

In view of the opposite effects of the cro1 mutation and the sm19 mutation, which causes hypoduplication of basal bodies (Ruiz et al., 1987), we examined their interaction. Analysis of cro1 × sm19 crosses showed that the two mutations complemented each other: the heterozygotes displayed a wild- type phenotype. In the ex-autogamous F₂, an independent segregation of the two genes was observed and the double mutants were easily identified by their crochu phenotype at 28°C and the expression of the sm19 phenotype (smaller and smaller cells and eventually death) at 35°C. Cytological studies of basal body duplication pattern in the double mutant grown at 35°C revealed a coexpression of the two mutations, with hyper or hypoduplication of basal bodies, depending on the cell surface regions. This complex phenotype may indicate subtle differential susceptibilities of the various cortical territories to the expression of either mutation and/or different timing of expression. Nevertheless, it can be concluded that neither of the two mutations is epistatic over the other and therefore that the two mutations correspond to defects along independent pathways in the control of basal body duplication.

The hypersensitivity of cro1 cells to nocodazole led us to examine the interactions of the cro1 mutation with noc^r1, a mutation in a β-tubulin gene (Dupuis-Williams et al., 1997) conferring resistance to nocodazole, i.e. normal growth rate in the presence of 4-8 μM of the drug (Torrès et al., 1991). In the F₁ of cro1 × noc^r1 crosses, the two mutations were found to be complementary. In the ex-autogamous F₂, the two genes segregated independently. 25% of the progeny were wild type and 25% noc^r1 (as judged by cell morphology and comparison of growth in the presence or in the absence of nocodazole). However, among the 50% of progeny expressing the cro1 phenotype, none were resistant to nocodazole, suggesting that the double mutants cro1: noc^r1 did not express the noc^r1 resistance. This was confirmed by back-crosses of F₂ clones of cro1 phenotype to the noc^r1 parental strain. The clones thus identified as double mutant had a sensitivity to nocodazole (growth arrest at 4 μM) similar to that of cro1 clones. The mutation cro1 therefore behaves as an extragenic suppressor of noc^r1.

In Paramecium, the development of cell shape and surface pattern during division depends on a precise spatiotemporal control of basal body duplication (Iftode et al. 1989). In the mutant cro1 described here, both temporal and spatial aspects of basal body duplication are altered. Not only do basal bodies continue to duplicate in interphase, but they also do so in an ‘invariant’ territory where basal bodies never duplicate in the wild type, and at ectopic locations, beneath the cortex. In addition the cro1 mutation causes hypersensitivity to nocodazole and, in the cro1:noc^r1 double mutant, suppresses the resistance to the drug, conferred by the noc^r1 mutation in a β-tubulin gene. These properties of the cro1 mutation provide some new insight into pattern formation in Paramecium and into the coupling of basal body duplication with the cell cycle.

The crochu cells are much enlarged, display an extended anterior part and their basal body pattern is abnormal. However, these defects, which are conspicuous in interphase cells, do not seem to modify significantly the progression of basal body duplication or its spatial modulation during division. The following parameters are not affected: timing and site of development of the new oral apparatus, localization of the fission furrow, retention of normal contours of cortical fields (Fig. 3), succession of first and second waves of basal body duplication and differentiation of cortical units. Therefore, the supernumerary basal bodies formed in interphase neither modify the body plan nor seem to be integrated in the pattern. Their non-integration can be explained by the fact that the differentiation of cortical units, involving precise arrangement of other filamentous structures (epiplasm scales, outer lattice and centrin network) occurs exclusively during the second part of the divisional morphogenesis, in cro1 as in wild type, so that basal bodies inserted into the cortex during interphase may lack their normal environment and be unable to generate new cortical units. If all the supernumerary basal bodies formed during interphase were truly integrated in the cortex then the crochu phenotype would tend to be maintained by cortical inheritance like for instance complete rows of inverted basal bodies which are in fact rows of inverted cortical units (Sonneborn, 1975; Frankel, 1989). Also, their number would increase exponentially over successive cell generations. This is not the case: some homeostasis intervenes, since the average cell size and basal body number are reached after a few divisions in the newly formed cro1/cro1 homozygotes, and since arrest of cell division upon starvation tends to reduce basal body number. In addition, when heterozygotes are formed, one observes a fast return to normal cell size and number of basal bodies (Table 1).

It would appear, as previously discussed, that the program of basal body deployment at division is determined during the previous division and is not modified by changes that occur after division. In the Paramecium mutant kin241 (Jerka- Dziadosz et al., 1990) and in the mutant mlm of Paraurostyla weissei (Dubielecka and Jerka-Dziadosz, 1989; Jerka- Dziadosz and Czupryn, 1997), perturbations in the program of basal body duplication are expressed during division and correlated with a change in the body plan. In sm19, the random loss of basal bodies does not seem to alter either the properties of the territories or the body plan and only leads to a reduced cell size. The notion that positional cues for division are set up at the end of the previous division is well supported by a large corpus of data from yeast (reviewed by Drubin and Nelsen, 1996; Longtine et al., 1996) and also documented in another ciliate, Paraurostyla (Jerka-Dziadosz and Czupryn, 1997).

Previous studies have shown that the mitotic signals, which include a p-34-like kinase (Tang et al., 1994), trigger morphogenetic waves which spread over the cell surface and sequentially activate basal body duplication and reorganization of all cytoskeletal systems accompanied by phosphorylation of at least some of the cortical structures (Sperling et al., 1991; Beisson, 1994). To explain the crochu phenotype (normal timing and propagation of the morphogenetic waves; enhanced basal body duplication during division and its continuation in interphase; hypersensitivity to nocodazole and suppressive effect on a mutation in a β-tubulin gene), the following hypothesis can be proposed. The primary defect of the cro1 mutant is to render constitutive a factor that in wild-type cells is activated or synthesized only in response to the mitotic signals. This defect is independent of the properties of the cortical domains and of the basal bodies, both of which remain normal in the mutant. This factor affects microtubules by decreasing their stability. It is well established that microtubules become much more dynamic during mitosis (Cassimiris, 1993; Bulinski and Gunderson, 1991; Glicksman et al., 1992; Jordan et al., 1992; Hyman and Karsenti, 1996; McNally, 1996), in particular due to the phosphorylation of MAPS by the p34-cdc2 kinase or other kinases such as MARKS (Drewers et al., 1997). Increased microtubule instability also occurs during division in Paramecium, when several networks are disassembled: the internal microtubule network and the post-oral fibers (Cohen and Beisson, 1988) as well as the microtubule bundles called ‘paratenes’ which sustain the anterior invariant field (Fleury and Laurent, 1995). The idea that microtubule instability occurring at division might facilitate basal body/centriole duplication is supported by the observation that centrosome duplication can easily be uncoupled from nuclear division in embryos of Xenopus (Gard et al., 1990), sea urchins (Sluder and Lewis, 1987), starfish (Picard et al., 1988) or in Drosophila (Raff and Glover, 1988), where the microtubular system remain essentially in a mitotic state throughout the early division cycles of the egg.

The hypothesis that the cro1 mutation causes a constitutive instability of at least some microtubule subsets and thus provides a permissive condition for basal body duplication throughout the cell cycle can globally explain the properties of the mutant. Furthermore this hypothesis is consistent with the idea, previously discussed by Beisson (1994), and also considered by Sluder and Rieder (1996) for animal centrosomes, that the temporal coordination of basal body duplication and division may simply rely on the timing of the mitotic signals and on the response of basal bodies which, in Paramecium, is modulated by their location (Iftode et al., 1989). Although our data do not permit us to speculate on the specific nature of the mutation, the effects of the cro1 mutation are compatible with a defect in a tubulin or a MAP (Bré and Karsenti, 1990), as evidenced by its hypersensitivity to nocodazole and interaction with a β-tubulin mutation. It is possible that a kinase or phosphatase is affected since their deficiency was recently shown to uncouple the nuclear and centrosome cycles in mts mutant of Drosophila (Snaith et al., 1996) and kinase and phosphatase inhibitors cause rapid alterations in microtubule dynamic instability in living cells (Howell et al., 1997). The recently developed technology for cloning Paramecium genes by functional complementation (Haynes et al., 1996; Skouri and Cohen, 1997) will hopefully allow us to characterize the cro1 gene.

We are very grateful to Drs T. Kreis, J. Salisbury and J. Wehland for their kind gift of antibodies. We thank Drs Linda Sperling, Jean Cohen and André Adoutte for critical reading of the manuscript and are especially grateful to Dr Joseph Frankel for stimulating discussions and helpful suggestions concerning the manuscript. This work was supported by the CNRS (F. R., J. B.) and by a statute grant (M. J D.) from the Committee of Scientific Research to the Nencki Institute (Warsaw, Poland) and a scientific exchange program between CNRS (Paris, France) and the Polish Academy of Sciences (Warsaw, Poland).