The effects of 6-dimethylaminopurine (6-DMAP) (a putative phosphorylation inhibitor) on the state of assembly of microtubules and intermediate filaments have been studied during the first cell cycle of the sea urchin Strongylocentrotus droebachiensis. Changes in the spatial organization of cytoskeletal structures were studied by indirect immunofluorescence with anti-tubulin and anti-IFa antibodies. The rates and patterns of protein phosphorylation in control and treated eggs were also investigated.
The transfer of fertilized eggs to 600 pM 6-DMAP within 4 min following insemination inhibits pronuclear migration and syngamy. This also prevents male pronuclear decondensation, while chromatin condensation and nuclear envelope breakdown do not occur in the female pronucleus. Immunolabeling with anti-tubulin antibodies reveals the presence of cortical microtubules as early as 15 min after fertilization in both control and treated eggs. However, no sperm astral microtubules could be detected in the treated eggs. At later stages, from syngamy (40 min) up to nuclear envelope breakdown (90 min), 6-DMAP affects neither cortical microtubule organization nor the state of chromatin condensation but it precludes nuclear envelope breakdown and entry into mitosis. Treatment of the fertilized eggs after nuclear envelope breakdown induces permanent chromosome decondensation and premature disappearance of the mitotic apparatus. This last event involves disruption of the spatial organization of both microtubules and putative intermediate filaments.
Quantitative measurements of protein phosphorylation show that 6-DMAP efficiently and reversibly inhibits 32P incorporation into proteins. Qualitative analysis of the autoradiograms of 32P-labeled proteins separated by SDS-PAGE reveals that a major protein band, migrating with an apparent molecular weight of 31 × 103Mr, is specifically dephosphorylated in eggs treated with 6-DMAP.
This study suggests that protein phosphorylation is required for sperm aster microtubule growth and migration, but not for cortical microtubule polymerization. It also strengthens the hypothesis that, in sea urchin eggs, putative intermediate filaments are tightly associated with spindle microtubules. Finally, it confirms that inhibiting protein phosphorylation before nuclear envelope breakdown reversibly prevents the entry into mitosis.
Successful fertilization involves the dynamic reorganization of two major cytoskeletal components, the actin microfilaments, and the tubulin microtubules. Microfilaments are required for sperm incorporation and cell cleavage (Schatten and Schatten, 1981; Schatten et al. 1986; Schroeder, 1975), while microtubules are required for pronuclear migration, centration and fusion, and for chromosomal segregation at mitosis (Bestor and Schatten, 1981; Hamaguchi et al. 1985; Harris et al. 1980; Hollenbeck and Cande, 1985; Salmon et al. 1984; Schatten and Schatten, 1981).
In sea urchin eggs, a third cytoskeletal component has been identified that appears to be related to the intermediate filament (IF) protein family, which has been studied mostly in vertebrate cells. Schatten et al. (1985b) first reported the presence of lamins, on the male and female pronuclei of the early embryo, which are proteins that share similarities with IF proteins (reviewed by Franke, 1987; Steinert and Roop, 1988). Recently, St-Pierre and Dufresne (1990) showed that putative IF proteins were present in spindles and larval cilia of sea urchin embryos. The fact that these putative IF proteins co-distributed with the microtubules both in vivo and in vitro suggested that they could play a structural or physiological role in the organization of the microtubule system.
The dynamic changes in the spatial organization of the cytoskeletal components have been shown to depend upon the intracellular calcium release and the pH increase that follow fertilization and are likely to cause the observed post-translational modifications that affect major cytoskeletal proteins (Begg and Rebhun, 1979; Carron and Longo, 1982; Coffe et al. 1985; Dufresne et al. 1987, 1988; Schatten et al. 1985a;,Suprenant and Marsh, 1987; Suprenant, 1989).
For studying the role of protein phosphorylation in vivo, 6-DMAP, which blocks the cell cycle in Spisula and sea urchin eggs, has proved to be a useful tool (Rebhun et al. 1973). Indeed, this drug has been shown to inhibit reversibly protein phosphorylation in mollusc and starfish oocytes (Dubé et al. 1991; Néant and Guerrier, 1988a, b), as well as in sea urchin eggs (Dufresne et al. 1989; Néant et al. 1989) and mouse oocytes (Rime et al. 1989). Interestingly, it did not affect the rate and patterns of protein synthesis (Néant et al. 1989). The major morphological effects of this drug included chromosome decondensation and the precocious disappearance of spindle fibers, while cortical and pericentriolar microtubules remained unaffected.
Recently, Verde et al. (1990) showed that the shrinking of centrosome-nucleated microtubules observed in vitro, under the influence of cdc2 protein kinase, was completely inhibited by 6-DMAP. Also, 6-DMAP suppressed the labeling of the centrosomes by MPM2, an anti-phosphoprotein antibody (Rao et al. 1989; Vandre et al. 1984,1986). As previously suggested by Kuriyama et al. (1990) and Kuriyama (1989), these results show that phosphorylation of centrosomal components must play an important role in regulating microtubule dynamics. Intermediate filament proteins also exhibit changing levels of phosphorylation but, except for the lamins, the exact effects of these transitions are not well understood (Ben-Ze’ev, 1983; Evans, 1988; Julien and Mushynski, 1982, 1983; Robson, 1989).
The question addressed in this work was to determine the protein phosphorylation requirements for completion of the various steps in microtubule assembly and dissas- sembly. Also, we further investigated the morphological relationships previously described between microtubules and intermediate filaments (St-Pierre and Dufresne, 1990). Our data show that pronuclear migration and centration are reversibly inhibited by the drug 6-DMAP. Applying the treatment at metaphase leads to disruption of the mitotic apparatus, which expands perpendicularly to its original axis. The chromosomes, which remain aligned as originally organized on the metaphase plate, prematurely decondense concomitantly with spindle reorganization. Finally, analysis of autoradiograms of 32P- labeled proteins reveals that one major effect of 6-DMAP is to induce the reversible dephosphorylation of a 31 × 103Mr protein.
MATERIALS AND METHODS
Handling of gametes
Adult specimens of the green sea urchin Strongylocentrotus droebachiensis were collected at Pointe-au-Père (Québec, Canada) and kept in running sea-water tanks at the Station Aquicole of the Institut National de la Recherche Scientifique (INRS, Québec). Shedding was induced by intracoelomic injection of 1 ml of 0.52 M KC1. Spawned eggs were collected in natural filtered sea water (NSW) adjusted to pH 7.8 and to a salinity of 24–26 ‰, and maintained at 10°C. Males were put upside down on Petri dishes on ice while releasing their semen, which was collected and kept dry at 4°C until use. Spawned eggs were filtered through cheesecloth, washed twice with NSW and the final egg suspension was adjusted to 2%, v/v. The fertilizability and synchrony of development of every batch of eggs were tested before carrying out the experiment. Batches showing 95–100% fertilization membrane within 5 min were selected for further experimental manipulation. Sperm dilutions were between 10 000- and 20 000fold.
Chemicals and solutions
6-DMAP was purchased from Sigma (St Louis, Missouri), and 15 mM stock solutions were prepared in distilled water. The DNA stain Hoescht 33258 (Calbiochem, Mississauga, Ontario) was prepared as a stock solution of lmgml−1 in DMSO (dimethyl sulfoxide) and used at 0.5 μgml−1 (see Dufresne et al. 1988). Antitubulin antibodies were purchased from Amersham (Oakville, Ontario) or Sigma, and used as described by St-Pierre and Dufresne (1990). The universal antibody for intermediate filament proteins, called the anti-IFa (Pruss et al. 1981), was kindly provided by Dr Michel Vincent, from the Centre Hospitalier de l’Universita Laval. It has been shown (St-Pierre and Dufresne, 1990) that labeling of the embryos with this antibody reveals a staining pattern similar to that observed after anti-tubulin labeling, and that one band mainly of 117 × 103MT immunoreacted with this antibody after Western immunoblotting of egg or embryo homogenates. Radioactive [32P]orthophosphate was purchased from Amersham.
DNA staining procedures
Staining of DNA was achieved as described by Dufresne et al. (1988). Briefly, the eggs were fixed for lh in 2% formaldehyde prepared in a glucosamine-acetate buffer, washed with fixative-free buffer and stained with 0.5 μg ml−1 of Hoescht 33258. Observations were carried out with a Leitz epifluorescence microscope. Micrographs were taken on Kodak Tri-X film (400 ASA) with Mikrophot Wild MPS 05–12 photo equipment.
The distance between the female and male pronuclei was determined by measuring the length, in micrometers, between the two pronuclei, and the apparent egg diameter for each cell examined. The recorded distance was then corrected for a normal egg diameter of 160 pm. These normalized values were then distributed into four classes of 20 pm each and plotted as frequency histograms.
The same protocols as described by Dufresne et al. (1988) and St-Pierre and Dufresne (1990) were used. Briefly, cells were extracted in KGE buffer (Paweletz et al. 1984) containing 0.8% paraformaldehyde and 1% Triton X-100 for 1 h, washed with the buffer alone, then incubated for 1 h with the first antibody diluted to 1/250 in KGE buffer. After 3 washes, they were incubated for 1 h with a rabbit anti-mouse IgG conjugated to fluorescein (Sigma). The Hoescht 33258 DNA stain was added during this incubation to a final concentration of 0.5 μgml−1. After 3 washes with the buffer, the eggs were mounted in 50% glycerol diluted in KGE. Observations under ultraviolet illumination and photography were performed as described above for DNA.
Protein phosphorylation: quantitative and qualitative analysis
The protein phosphorylation rates and patterns were measured in eggs treated or not within 4 min of insemination with 600 μM 6-DMAP and kept in the presence of the drug until control embryos underwent first cleavage. Also, the effects of washing the embryos after a period of 80 min of incubation were analysed. Finally, the stages of development reached under these various experimental conditions were determined after processing the radioactive samples with Hoescht 33258.
A single batch of eggs was first inseminated at zero min. At 2 min after fertilization (AF), two-thirds of the population was transferred to 600 μM 6-DMAP. At 5 min AF, 200μCiml−1 of [32P]orthophosphate (10mCiml−1) were added to each lot of embryos, which were left until 80 min AF. Control embryos were then washed with NSW. Treated embryos were first divided into two batches, one half being washed with NSW and the other with NSW supplemented with 600 μM 6-DMAP. Starting at 90 min AF, duplicate samples (200 μl) were taken at desired times, transferred into 200 μl of 20 % trichloroacetic acid (TCA) containing 2 % sodium pyrophosphate and left for 30 min on ice. The samples for SDS-PAGE and autoradiography were washed twice with cold acetone and the pellets were dissolved in 100 μl of Laemmli sample buffer (Laemmli, 1970). The other series was washed twice with 1 ml of 5 % trichloroacetic acid (TCA) with 1 % sodium pyrophosphate, and dissolved in 0.3 ml 0.5 M NaOH. The proteins were again precipitated overnight at 4 °C, with 1 ml of 20% TCA containing 2% sodium pyrophosphate, further washed as described above, and dissolved in 0.5 ml 0.5 M NaOH for 30 mm; 400 id were mixed with 5 ml Scintiverse, the remaining 100μl were used to determine the protein content according to the method of Bradford (1976).
Samples processed for electrophoresis were run on 10 % to 15 % gradient SDS-PAGE. After drying, the gels were exposed on Kodak X-Omat films, and revealed with GBX developer (Kodak).
Effects of 6-DMAP on pronuclear migration
In Strongylocentrotus droebachiensis embryos raised at 10 °C, completion of the migration of the male towards the female pronucleus, determined after staining fixed eggs with Hoescht 33258, is observed from 45 to 60 min after fertilization (AF) (Fig. 1A,B). Condensation of the fused pronuclear chromatin is completed by 100 min AF (Fig. 1C), metaphase is observed around 120 min AF (Fig. 1D) and by 180 min AF the first cell division has occurred in 100 % of the egg population.
The various effects of 6-DMAP on the process of pronuclear migration and chromatin events are depicted in Fig. 1 (E to H). In control eggs, male pronucleus migration is completed by 60 min AF. In eggs treated within 4 min of insemination and further incubated in the presence of 300 or 600 μM 6-DMAP, the sperm pronucleus appears to be still located at the cortex (Fig. 1E,F,G). The egg shown in F was treated with 300 μm 6-DMAP, while in G the final concentration was 600 μM 6-DMAP. At the lower dose, despite the inhibition of sperm pronuclear migration, chromatin condensation occurs in the female pronucleus with the same time schedule as in control eggs (compare Fig. 1C and F). At 600 μM 6-DMAP, both female pronuclear centration and chromatin condensation are totally suppressed (Fig. 1G). As illustrated in Fig. 1H, inhibition of migration may be reversed by a simple washing of the cells with NSW. This egg was treated with 600 μM 6-DMAP for 60 min, washed with NSW and fixed 60 min later, at 120 min AF, when the control eggs have reached metaphase (Fig. 1D). Sperm nuclear migration is in progress, although syngamy, which is normally completed by 60 min AF (Fig. 1B), has not yet occurred in this egg.
These qualitative observations were further analysed by measurements of the distance between the two pronuclei at various times after treatment with the drug and after its removal, followed by further incubation of the washed embryos in NSW. The results of such an experiment are illustrated in Fig. 2. In the control embryos (top histograms), completion of pronuclear migration takes place at approximately 60 min AF. In embryos transferred to 600 μM 6-DM1AP within 4 min AF (bottom histograms), pronuclear migration is drastically slowed down. The mean distance (ݲ) separating the pronuclei decreases from an average initial value of 60 μm to 44 μm by 35 min AF. Subsequently, however, embryos raised in the presence of the drug do not show any further pronuclear movements. If the treated embryos are washed and further maintained in NSW, migration resumes, albeit at a slower rate than in control embryos. In the latter, an estimated migration rate of 0.85 μl min−1 is calculated between 25 min and 60 min AF (30 μl in 35 min). Upon removal of the drug, migration proceeds at only 0.55 μm min−1 (33 pm in 60 min).
These quantitative results reveal that 6-DMAP reduces the migration rate during the first 35 min after insemination, then totally inhibits any further motion of the pronuclei. Upon removal of the drug, sperm nuclear migration resumes, albeit at a slower rate than in the control embryos.
Since migration of the pronuclei is a microtubuledependent event requiring sperm astral microtubules, we carried out an immunofluorescence study in control and treated embryos processed for visualization of tubulin- staining structures. In both control and treated embryos, we first observed a cortical array of microtubules (Fig. 3), the intensity of which increases over time. Fig. 3A shows a control embryo processed at 15 min AF, when sperm pronuclear migration is in progress (Fig. 3B). At this time, the cortical meshwork is rather sparse. However, as development proceeds, this meshwork grows in density, and becomes well developed (Fig. 3E and F). Fig. 3C and D illustrate the cortical pattern in embryos maintained in the presence of 600 UM 6-DMAP and processed at 60 min (Fig. 30 and 90 min (Fig. 3G) AF. In this instance, despite inhibition of sperm pronuclear migration (Fig. 3D and H), the cortical meshwork appears as well developed as in the control embryos. The presence of this cortical microtubular array often prevents detection of the sperm astral microtubules. The latter, sometimes observed in untreated embryos, were never seen in the treated ones (data not shown; see St-Pierre and Dufresne, 1990).
We then investigated whether 6-DMAP would block the formation of the mitotic apparatus once fusion has occurred and the mitotic spindle poles have been established.
Morphological effects of 6-DMAP added after syngamy
DNA staining pattern
Fig. 4 shows the chromatin in eggs treated with 6-DMAP at 50 min (A), 70 min (B) and 80 min (C) after insemination, and thus before nuclear envelope breakdown (NEBD), and fixed and processed at 230 min AF. In these instances, chromatin condensation was proceeding before drug addition, and remains blocked in this state. Transfer of eggs to 600,UM 6-DMAP at the time of entry into mitosis, at 90 min (D), 100min (E) and 120 min AF (F) results in the formation of a varying number of resting nuclei, located in the center of the egg. These eggs were all fixed and processed at 230 min AF, a time corresponding to the beginning of the 3rd cleavage in control embryos. Altered states of condensed and decondensed chromatin were not observed at earlier time intervals, suggesting that the cell cycle is interrupted when the drug is added before NEBD and modified when added after NEBD. As mentioned, and described by Néant et al. (1989), these effects may be reversed on washing and raising the embryos in NSW.
Microtubule and intermediate filament staining patterns
Treatment of eggs at metaphase-anaphase stages caused drastic modifications of the spindle organization, which involved concomitant and strongly similar reorganization of the microtubules and putative intermediate filaments. This is illustrated in Fig. 5, which shows in the first two rows the staining patterns of the microtubules of the mitotic apparatus and, in the bottom row, the staining pattern obtained after anti-IFa labeling, with a control egg shown on the left. The anti-tubulin staining reveals that soon after addition of 600 μM 6-DMAP (A, 120 min AF; B and C, 10 min after treatment), the astral fibers shorten from their free ends, concentrating towards the centrosomal area, which flattens (top row). The spindle mid-zone is also brightly stained, except at the site of the chromosomes. Next (D, E and F, 17 min after treatment), outwards astral fibers are no longer detected, whereas the modified mitotic apparatus shows elongation and flattening perpendicular to the original spindle axis. The intermediate filament pattern follows a similar reorganization. In control eggs (G, 120 min AF), anti-IFa labeling reveals a staining pattern covering the polar areas, with discrete fibrillar staining in the mid-zone as well as apparent staining of the chromosomes. Upon 6-DMAP treatment (H, 10 min after treatment; and I, 17 min after treatment), the astral labeling is lost and the staining is concentrated predominantly in the polar regions. Chromosomal staining is still observed as the chromosomes decondense to produce karyomeres at the level of the original equatorial metaphase plate (insets in D and F).
Quantitative and qualitative effects of 6-DMAP on protein phosphorylation
We then analyzed the effects of 6-DMAP on the rates and patterns of protein phosphorylation. Fig. 6 illustrates the cumulative incorporation of 32P into proteins in control fertilized eggs (F, △), in eggs transferred within 4min of insemination in the presence of 600 μM 6-DMAJP (+6D, □) and after removal of the drug at 80 min AF and further culture of the washed cells in NSW (—6D, ○). Addition of 6-DMAP soon after fertilization (+6D, □) inhibits incorporation of radioactive phosphate into proteins. Upon washing the embryos and further incubation in NSW (-6D, O) the level of phosphorylated protein is restored, although to a lower level than in control embryos (F, A) compared to that in the untreated embryos (not shown).
We then compared the patterns of protein phosphorylation of control embryos (Fig-7, lanes 1 to 4), with those of eggs either raised in the continuous presence of the drug (Fig. 7, lanes 5 to 8) or after washing at 80 min AF and returning to NSW (Fig. 7, lanes 9 to 12). For the three conditions, embryos were sampled from 90 min to 270 min AF at 60 min intervals. In this experiment, the amounts of sample from the 6-DMAP-treated embryos and the washed ones loaded onto the gels were adjusted to the ctsmin−1 measured in the control embryos. We thus obtained similar band intensities to those in the control embryos, except for those proteins whose phosphorylation is specifically inhibited by 6-DMAP. Fig. 7 shows that in control embryos (lanes 1 to 4) all the protein bands revealed under our electrophoretic conditions undergo increased phosphorylation over time. In the 6-DMAP-treated embryos (lanes 5 to 8), in which protein phosphorylation is inhibited (Fig. 6), we note that the same protein bands as those depicted in the control embryos are phosphorylated, except for one band, migrating at approximately 31×103Mr. Finally, washing the embryos free of the drug (lanes 9 to 12) results in an electrophoretic pattern similar to the one observed in control embryos. It can be observed that the intensity of the protein band migrating at 31×103Mr is already higher at 90 min in the washed embryos (lane 9) than in the 6-DMAP-treated ones, sampled at 270 min AF (lane 8). The treatment of embryos at 80 or 100 min (before or after NEBD) also resulted in the rapid decrease in intensity of this 31 × 103Mr protein band (not shown).
This study reveals that the phosphorylation inhibitor 6- DMAP has different effects depending whether the drug is added to fertilized eggs before or after NEBD. In the first case, it inhibits pronuclear movements, chromatin condensation and NEBD. However, as soon as NEBD has occurred, 6-DMAP induces premature decondensation of the chromosomes and formation of resting nuclei. Simultaneously, 6-DMAP precludes the establishment of normal mitotic figures. When applied at metaphase, it disrupts the established spatial organization of the microtubules and putative intermediate filaments that usually ensure normal chromosomal segregation.
6-DMAP is thought to exert its biological effect through its capacity to reduce the phosphorylation of some pivotal proteins. This idea is supported by the observation that it does not seem to suppress protein synthesis while it reduces the incorporation of phosphate into proteins in vivo, as shown for eggs or oocytes from diverse species, including echinoderms (Néant et al. 1989; Néant and Guerrier, 1988b), mammals (Rime et al. 1989) and molluscs (Néant and Guerrier, 1988a; Dubé et al. 1991). Furthermore, 6-DMAP does not appear to affect phosphatase activities, and in vitro studies have shown that sea urchin egg histone Hl kinase is efficiently inhibited by this drug (Meijer and Pondaven, 1988). However, sea urchin sperm histone kinase is not inhibited by 6-DMAP (Poccia et al. 1990). This confirms that 6-DMAP shows specificity towards specific kinases. Moreover, in the present work with sea urchin eggs, the main biochemical effect of 6-DMAP consists of a general reduction in the incorporation of [32P]orthophosphate with an apparent preferential effect upon a 31xlO3Mr protein.
Microtubule and spindle dynamics
In our study, cortical microtubules were found in both control and treated eggs, whereas sperm aster microtubules, sometimes observed in control embryos, were never observed in the treated ones. Since the cortical microtubule array might have prevented detection of the sperm astral microtubules, we cannot ascertain whether inhibition of sperm pronuclear migration is due to inhibition of their growth. Nonetheless, our results suggest that phosphorylation is required for completion of microtubuledependent pronuclear migration. Assembly of cortical microtubules would proceed independently of phosphorylation and would not be related to the cell cycle.
Possible kinases sensitive to 6-DMAP may be those phosphorylating centrosomal components and intermediate filament proteins. Components of the centrosome have been shown to be labeled with anti-phosphoprotein antibodies (Kuriyama et al. 1990; Rao et al. 1989; Vandre et al. 1984, 1986), whereas phosphorylation of the centrosomes under the influence of cdc2 kinase is inhibited by 6-DMAP (Verde et al. 1990). At present, we have no direct proof of the involvement of IF phosphorylation as a possible relay mechanism operating in controlling spindle dynamics. However, it is worth noting that the addition of 6-DMAP at metaphase induces a simultaneous and parallel disruption of the IF and the microtubules of the mitotic apparatus (cf. Fig. 5). Moreover, it appears that NEBD, which usually involves lamin phosphorylation (Chou et al. 1989; Dessev et al. 1988, 1989), is efficiently blocked by 6-DMAP, in which case the anti-IFa labeling pattern associated with the nuclear envelope is preserved. Also consistent with this view is the fact that lamin proteins and vimentin are substrates of p34cdc2 kinase (Chou et al. 1990; Peter et al. 1990; Ward and Kirschner, 1990).
When added early after fertilization, 600 μM 6-DMAP inhibits condensation of the female pronuclear chromatin. However, when added after chromatin condensation and before nuclear membrane breakdown, 6-DMAP cannot induce any chromatin decondensation. This would suggest that phosphorylation is required to condense the chromosomes but not to maintain them in this condensed state. Alternatively, 6-DMAP might not interfere with events taking place inside the nucleus. Indeed, once NEBD has occurred, condensed chromosomes rapidly decondense upon addition of 6-DMAP to produce resting micronuclei within 15 min, which will eventually fuse together.
These observations strongly suggest that 6-DMAP causes a rapid inactivation of the M-phase promoting factor (MPF) which is responsible for NEBD, spindle formation and chromosome condensation (Guerrier et al. 1990, for review). The catalytic subunit of MPF, p34cdc2, has been shown to translocate from the cytoplasm into the nucleus before mitosis (Bailly et al. 1989; Booher et al. 1989; Riabowol et al. 1989). One possibility is that 6-DMAP inhibits phosphorylation events taking place in the cytoplasm, and therefore prior to translocation, and once this process is completed 6-DMAP no longer prevents intranuclear chromatin condensation, while still interfering with NEBD.
The analysis of the phosphorylation pattern under the influence of 6-DMAP has revealed that phosphorylation of one protein band of 31×103Mr, which is a major phosphorylated substrate in control embryos, is suppressed. This result suggests that this particular substrate could play a pivotal role in the phosphorylation cascade regulating the cell cycle. Further work (in progress) will characterize this protein in order to understand better the series of events that precede and follow its reversible modification under the influence of 6-DMAP.
The technical assistance of Mrs Louise Coutu and Claude Guerrier was greatly appreciated. We are grateful to B. B. Rudkin for improving the final version of the manuscript. This work was supported by grants from the Centre Jacques Cartier, the Fondation Mérieux (Lyon, France) and the ARC (France) to P.G., and by NSERC grants (Canada) to L.D. and F.D.