Evidence for a role of the α-tubulin C terminus in the regulation of cyclin B synthesis in developing oocytes

The cell cycle progression is controlled through periodic activation and inactivation of cyclin dependent kinases (cdks: see Pines, 1995, for a review). These kinases form dimeric complexes with different cyclins. The various complexes have different functions, principally determined by the nature of the cyclin. The mitotic cyclins, such as cyclins A and B, control the G₂/M transition and thereby cell entry in mitosis. The correct control of the expression of mitotic cyclins is central for the control of cell division. Deregulation of these cyclins is a frequent occurrence in tumor cells where cyclins A and B are generally both prematurely expressed and overexpressed (Keyomarsi and Pardee, 1993). Therefore, understanding the regulation of the mitotic cyclin expression is an important challenge for cell biologists and pathologists.

Mitotic cyclins are expressed during a phase of the cell cycle when there is little gene transcription (White et al., 1995; Yonaha et al., 1995; Leresche et al., 1996; Klein and Grummt, 1999). Furthermore they have a complex pattern of expression (Trembley et al., 1994; Hwang et al., 1998). For these reasons, it is very likely that mitotic cyclin expression is also controlled at the translational level. Deciphering the regulation of mRNA translation is a difficult task in somatic cells where mRNAs are synthesized, used and degraded in a continuous way. The regulation of mRNA translation is much easier to study in oocytes and early embryos, where maternal mRNAs are stocked in a nontranslatable form and are subsequently activated to yield protein synthesis (Vassali and Stutz, 1995). With regard to the regulation of cyclin B synthesis, developing oocytes have the additional advantage that at some stages of development, the cyclin B mRNA is preferentially translated, cyclin B being massively synthesized whereas other maternal mRNAs are little or not translated. This feature of developing oocytes may favor the identification of molecules specifically involved in the translation of mitotic cyclins: such molecules remain largely unknown, with the exception of ERK2, which is emerging as an important regulator of the control of cyclin B translation (de Moor et al., 1997; Ballantyne et al., 1997; Fisher et al., 1999).

Here we have investigated the role of the tubulin molecule as one possible factor involved in the control of cyclin B synthesis in developing oocytes. Our interest in tubulin stemmed from two lines of observations. Firstly, previous observations showed that antibodies directed against the tyrosinated form of the α-tubulin C terminus block cell entry into mitosis in developing sea urchin eggs (Oka et al., 1990). The molecular mechanisms involved in the cell cycle blockage remain unknown, but may clearly involve an inhibition of cyclin B accumulation since this is the major factor responsible for cell entry into mitosis in developing eggs. Secondly, recent observations have shown that the tyrosination cycle is often suppressed during tumor progression, leading to the accumulation of detyrosinated tubulin variants in tumor cells (Lafanechère et al., 1998). Such observations also indicate a possible regulatory role of the α-tubulin C terminus in the regulation of the cell cycle. We used blocking antibodies and competing peptides to probe the role of the α-tubulin C terminus in both starfish and Xenopus oocytes. Our data indicate a direct influence of the recognized epitope on Cdk1 activation. We provide evidence that both antibodies directed against the tyrosinated form of the α-tubulin C terminus and their epitope peptides inhibit cyclin B accumulation and entry into mitosis.

YL1/2 is a rat monoclonal antibody directed against the C terminus of yeast α-tubulin (Kilmartin et al., 1982). It has been biochemically characterized (Wehland et al., 1983; Wehland et al., 1984). 20C6 is a mouse monoclonal antibody directed against a C-terminal peptide of tyrosinated α-tubulin (Wehland and Weber, 1987). L3 and L7 are rabbit polyclonal antibodies, specific for detyrosinated α-tubulin and Δ2-α-tubulin, respectively. These antibodies were obtained using synthetic peptides as immunogens and total IgGs were purified according to McKinney and Parkinson (McKinney and Parkinson, 1987), as described (Paturle-Lafanechère et al., 1994). For microinjections, antibodies were concentrated to 20 mg/ml, free of sodium azide. L3 IgGs were affinity-purified on the corresponding peptide and further concentrated to a protein concentration of 10 mg/ml by ultrafiltration. α-tubulin antibody was a commercial monoclonal mouse antibody (N 356; Amersham France SA). β-tubulin (clone TUB2.1) mouse monoclonal antibody was from Sigma (Sigma-Aldrich Chimie SARL, St Quentin Fallavier, France). IgGs were purified from ascitic fluid on a protein A-Sepharose column (Pharmacia), according to Ey et al. (Ey et al., 1978). Purified IgGs were concentrated to 10 mg/ml by ultrafiltration. Anti-mos antibodies and anti-active-ERK antibodies were from Santa Cruz (sc-94 and sc-7383, respectively; Santa Cruz Biotechnology INC, Santa Cruz, CA, USA). Anti-Xenopus cyclin B1 antibodies were rabbit anti-recombinant Xenopus cyclin B1, affinity purified (Abrieu et al., 1997b). Anti-starfish cyclin B were polyclonal antibodies raised against recombinant starfish cyclin B (Labbé et al., 1989).

Secondary antibodies: sheep anti-mouse Ig, fluorescein linked, were from Amersham France (ref. N 1031); goat anti-rat, alkaline phosphatase-coupled (A 8438), goat anti-mouse (A 9917) and goat anti-rabbit (A 1949) peroxidase-coupled, were from Sigma.

All antibody concentrations indicated in the text are final molar concentrations in the cell, assuming that the molar mass of IgG is 150,000 Da.

Peptides were synthesized by Neosystem company (Strasbourg, France) at over 95% purity. The sequences were the last 17 amino acids of bovine tyrosinated α-tubulin (peptide Y: VGVDSVEGEGEEEGEEY) and the last 16 amino acids of detyrosinated α-tubulin (control peptide E: VGVDSVEGEGEEEGEE). We cloned and sequenced the starfish α-tubulin, and found the C-terminal sequence is almost identical: VGIDSVDGEGEDEGEEY.

Concentrations indicated are final molar concentrations in injected cells.

The starfish Astropecten aranciacus was collected by diving from October to May in the ‘Baie de Banyuls’, near the laboratoire Arago (France), and kept in running sea water (SW). Pieces of ovaries and testes were taken through an incision in the dorsal wall of the animals. Prophase-arrested oocytes were prepared free of follicle cells by several rinses in artificial calcium-free SW (475 mM NaCl, 10 mM KCl, 50 mM MgCl₂, 10 mM Hepes-NaOH, pH 8.2). The hormone 1-methyladenine (1-MeAde) was from Sigma. Microinjection was performed according to Hiramoto (Hiramoto, 1974).

For western blots, ten A. aranciacus oocytes in 5 μl SW were added to 15 μl loading buffer (Laemmli, 1970). Synchronization was not necessary, since starfish oocytes reinitiate meiosis and enter mitosis with a naturally high synchronism. For in-gel MBP kinase activity, ten oocytes in 5 μl SW were homogenized in 15 μl EB (80 mM β-glycerophosphate, 20 mM EGTA, 15 mM MgCl₂, pH 7.3) and frozen in liquid nitrogen. After thawing, homogenates were centrifuged for 15 minutes at 13,000 rpm (10,000 g), and 15 μl of the supernatant were added to 5 μl of 4× loading buffer and boiled for 3 minutes.

Pieces of ovaries were manually dissected with watchmaker forceps, then follicle-free oocytes were kept in Ringer’s Modified Medium (MMR: 100 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 0.1 mM EDTA, 5 mM Hepes, pH 7.8). Progesterone (Sigma-Aldrich SARL), was dissolved to 2 mM in methanol, then adjusted to 2 μM in MMR.

For western blots and in-gel MBP kinase activity, one oocyte was resuspended in 20 μl of EB, homogenized and frozen in liquid nitrogen. After thawing, the homogenate was centrifuged at 13,000 rpm at 4°C for 15 minutes, and 15 μl of the supernatant were added to 15 μl of 2× loading buffer, then boiled for 3 minutes. 10 μl were loaded for each lane.

For synchronization and microinjection, oocytes in 2 μM progesterone were collected just at the beginning of germinal vesicle breakdown (GVBD: appearance of a dark circle near the animal pole), usually 2-6 hours after hormone addition (p.h.a.), every 15 minutes. Oocytes collected within 15 minutes were considered as synchronous; half of the collected oocytes were microinjected and kept apart until homogenization. While several experiments were performed, all the results given in this paper came frome the same complete experiment, except for those in Fig. 7D.

Histone H1 kinase activity was measured from homogenates in phosphorylation mix (1 mM MgCl₂, 50 μM ATP, 200 μg/ml histone H1, 40 mM Hepes-NaOH, pH 7.4) plus 1 μl [γ-³²P]dATP at 110 TBq/mmol, 370 MBq/ml (AA 0068, Amersham France), in a final volume of 100 μl.

In-gel MBP kinase detection was performed exactly according to Shibuya et al. (Shibuya et al., 1992).

³⁵S-labelled amino acids were from Amersham-France (redivue Pro-mix L [³⁵S] in vitro cell labelling mix, 37 TBq/mmol, 530 MBq/ml).

Previous work indicated that the inactivation of the tyrosinated α-tubulin C terminus by mAb YL1/2 could block sand dollar eggs either in interphase or in mitosis (Oka et al., 1990). We tested whether such a blockage also occurred in starfish and/or in Xenopus cells.

First, we wanted to identify the proteins recognized by mAb YL1/2 in starfish and Xenopus oocytes. The reactivity of the antibody was tested on immunoblots of oocyte extracts. The YL1/2 antibody reacted with a single band migrating at the expected molecular mass of α-tubulin (Fig. 1). The intensity of the signal remained constant during meiosis and mitosis (not shown). Immunofluorescence examination of oocytes stained with mAb YL1/2 showed bright staining of both interphasic and mitotic microtubular networks (Fig. 2), confirming that the major reactive protein was indeed tubulin. Furthermore, we confirmed by western blotting that α-tubulin was the only protein recognized by mAb YL1/2 in mAb YL1/2 immunoprecipitates (not shown). These data indicate that tyrosinated tubulin is present in starfish oocytes and that it is the only protein strongly recognized by the YL1/2 antibody.

We then microinjected fertilized starfish eggs with mAb YL1/2 (3.3 μM final concentration) following the second polar body emission. Injected eggs never entered M phase (Fig. 3Aa), whereas control eggs showed a normal cleavage pattern (Fig. 3Ab). A similar cell cycle arrest was observed in mAb YL1/2 injected blastomeres (not shown). Injection of eggs with antibody concentrations lower than 3.3 μM did not induce complete cell cycle arrest (Table 1).

To test whether the interaction of mAb YL1/2 with a cellular epitope was indeed essential to the observed effects of the antibody we microinjected a mixture containing 5 mg/ml peptide Y (a synthetic peptide which mimics the sequence of the last 17 amino acids of bovine tyrosinated α-tubulin) and 10 mg/ml mAb YL1/2 (final concentration 6.7 μM). Fig. 4A,B and Table 1 show that the saturation of mAb YL1/2 by its epitope peptide suppressed the effect of the antibody on the cell cycle. This result strongly suggests that the interaction of mAb YL1/2 with specific epitopes is indeed essential for interphase arrest and does not reflect a nonspecific side effect of the antibody.

A series of control experiments was run to further assess the specificity of the observed cell cycle blockage induced by mAb YL1/2.

We microinjected eggs with tubulin C terminus antibodies having different epitope specificities from YL1/2. Eggs were injected either with L3 polyclonal antibody (an antibody selective for the detyrosinated α-tubulin), or with L7 polyclonal antibody (an antibody selective for the Δ2 α-tubulin, lacking both tyrosine and the last glutamate), at similar IgG concentrations to those used for mAb YL1/2. These antibodies were ineffective in arresting cell cycles (Fig. 3Ad-f). In another series of experiments, we microinjected affinity-purified L3 antibody IgGs at the same final concentration with similar results (not shown). Starfish fertilized eggs were then injected with a commercial antibody directed against the β-tubulin C terminus (mAb TUB 2.1) at a final concentration of 13.3 μM, which was unable to arrest the cell cycle (Fig. 4C). In previous work, the only protein that has been found to cross-react significantly with mAb YL1/2, besides α-tubulin, is the small subunit of ribonucleotide reductase (rnr: Standart et al., 1985). As this subunit is abundant in early embryogenesis, we wanted to test the hypothesis that the observed effect of mAb YL1/2 microinjection was due to the activation of some checkpoint mechanism, as a consequence of rnr inhibition. We confimed that mAb YL1/2 immunoprecipitates from homogenates of sea urchin embryo the [³⁵S]methionine-labeled 41 kDa protein identified as rnr small subunit (Fig. 3B, lanes 1, 3 and 4). Interestingly, another monoclonal antibody that recognizes the tyrosinated α-tubulin C terminus, mAb 20C6, was much less efficient in immunoprecipitating rnr under the same conditions (Fig. 3B, lanes 2, 5 and 6), probably reflecting a difference in epitope recognition. Yet, mAb 20C6 microinjection resulted in interphase arrest, as shown in Fig. 3Ac, with similar sensitivity to mAb YL1/2 (Table 1). This observation means that the biological effect of mAbs YL1/2 and 20C6 microinjection is unlikely to be the result of antibody-triggered rnr inhibition.

Taken together, these results suggested a specific blockage of entry into M phase in eggs and blastomeres injected with antibodies that bind to the tyrosinated α-tubulin C terminus. Most of the experiments reported below were done both with mAb YL1/2 and with mAb 20C6 with identical results. We chose to show the mAb YL1/2 experiments in the following sections of this paper.

The results reported above indicate that the tyrosinated α-tubulin C terminus epitope has a crucial role in cell cycle progression in starfish oocytes. This raised the possibility that peptide Y may function as a competitive inhibitor of the relevant epitope in this system. A test of this possibility is shown in Fig. 4D. Peptide Y was injected in starfish eggs at a final concentration of 1 mM. The injected peptide had the same blocking effect as mAb YL1/2 or mAb 20C6. A dose-effect study for peptide Y is shown in Table 1. Interestingly, when peptide E, differing from peptide Y by deletion of the C-terminal tyrosyl residue and corresponding to the detyrosinated α-tubulin C terminus, was microinjected, eggs developed at least until the swimming gastrula stage (Fig. 4E,F). We also injected eggs with several non α-tubulin peptides having a terminal tyrosine residue. Such peptides were unable to mimic the mAb YL1/2 effect (data not shown).

In addition to the results observed with mAb YL1/2 injection, these results provide independent evidence for a specific involvement of the α-tubulin C terminus epitope in cell cycle regulation in starfish oocytes. Furthermore, the peptide microinjection experiments also strongly suggest that only tyrosinated epitopes are involved in the observed cell cycle blockage whereas detyrosinated tubulin molecules are not concerned.

M phase entry is under the control of a major cycling histone H1 kinase, which has been identified in starfish as MPF (M-phase promoting factor), the active cyclin B-Cdk1 dimer (Labbé et al., 1989). An obvious possibility was that the inactivation of the tyrosinated α-tubulin C terminus by mAb YL1/2 interfered with MPF activation. We used meiotic oocytes to test this possibility. Meiosis and mitosis differ in many ways but have in common a requirement for proper MPF activation. Furthermore, meiotic oocytes are easier to synchronize and yield faster results than mitotic eggs. Fertilized oocytes were injected with mAb YL1/2 either during the first meiotic division (M1), or after emission of the first polar body. Cytological observations showed that starfish oocytes microinjected with mAb YL1/2 during M1 arrested in interphase, with an interphasic microtubule array and dispersed procaryons (Fig. 5Aa,b). Furthermore, oocytes were arrested in interphase without DNA replication (Fig. 5Ac). mAb injection per se did not induce inactivation of H1 kinase activity (Fig. 5Ba), which was at least equal, and often slightly higher in microinjected oocytes than in controls. H1 kinase dropped in schedule with uninjected oocytes, but it did not rise again for the second meiotic cell cycle and this explained why oocytes were blocked in interphase. The same experiment was performed in synchronized Xenopus maturing oocytes, with essentially the same results (Fig. 5Bb). When fertilized starfish oocytes were microinjected after the first polar body emission, they were also blocked in interphase but the cell cycle arrested after DNA replication (Fig. 5Ad).

These results indicate a specific effect of mAb YL1/2 injection on Cdk1 activation at the interphase to mitosis transition. Apparently, the antibody does not affect other phases of the cell cycle: the occurrence or nonoccurrence of DNA synthesis in microinjected oocytes remains in schedule with the normal meiotic cell cycle.

The activation of Cdk1 depends both on Cdk1 association with the mitotic cyclin B and on several post-translational regulations. To be active, Cdk1 must be phosphorylated on threonine 161 by Cdk activating kinase (CAK). Furthermore, threonine 14 (T14) and tyrosine 15 (Y15) must be dephosphorylated by the dual-specificity phosphatase Cdc25. Thus, the observed inhibition of H1 kinase activity in fertilized oocytes injected with mAb YL1/2 could be due to several different factors. Cyclin B accumulation could be inhibited, Cdk1-cyclin B association could be inhibited (or reversed); finally, the inhibitory phosphorylation of T14 and Y15 could be maintained by activation of the kinases Wee1 or Myt1, and inhibition of the phosphatase Cdc25.

During the prophase of the first meiotic division (GV, or germinal vesicle stage), Cdk1 is already associated with cyclin B in a complex maintained inactive by T14-Y15 phosphorylation. We microinjected YL1/2 in starfish and Xenopus GV-stage oocytes, and induced meiosis reinitiation 1 hour later with the natural hormones. Fig. 6 shows that the time course of meiosis reinitiation in Xenopus oocytes was not altered by YL1/2 microinjection (Fig. 6A), and that the dose-response to 1-MeAde was not displaced by microinjection into starfish oocytes (Fig. 6B).

These results suggest that mAb YL1/2 interferes with the Cdk1 activation cascade upstream of Cdc25 activation and that mAb YL1/2 effect does not involve overactivation of either Wee1 or Myt 1.

One possibility therefore was that mAb YL1/2 injection interfered with cyclin B accumulation. The variations of the cyclin B concentration in starfish oocytes stimulated with the natural hormone are shown in Fig. 7A. The Fig. shows immunoblots of whole extracts of oocytes fixed at various time points following hormonal stimulation, probed with cyclin B antibody. Cyclin B concentration is high in GV oocytes (lane 1). The cyclin is degraded in prometaphase and the cyclin concentration reaches a minimum at the time of the first polar body emission (80 minutes after hormone treatment, lane 3). Then the cyclin accumulates again to reach a new maximum during meiosis II (90 minutes after hormonal treatment, lane 4). Following the completion of meiosis, the cyclin is degraded to reach a new minimum (150 minutes after hormonal treatment, lane 6) and then accumulates again (360 minutes after hormonal treatment, lane 7).

To test the effect of YL1/2 injection, GV oocytes were stimulated with the physiological hormone and injected during meiosis I, 40 minutes after hormonal stimulation. The injected oocytes were then fixed for immunoblot analysis 90 minutes (lane 5) or 360 minutes (lane 8) after hormonal stimulation.

The 90 minute time point corresponds approximately to the maximum cyclin B concentration reached during meiosis 2. In injected oocytes such a maximum was completely suppressed (lane 5): the cyclin B concentration at 90 minutes was similar to that observed at 80 minutes in controls. This suppression of cyclin B accumulation was apparently irreversible: the accumulation of cyclin B normally observed 360 minutes after hormonal treatment was completely absent in injected oocytes (lane 8). In another series of experiments we found that YL1/2 microinjection also strongly inhibits cyclin B1 accumulation in maturing Xenopus oocytes (Fig. 7B).

The lack of cyclin B accumulation in injected oocytes could be due either to an inhibition of cyclin B synthesis or to a stimulation of the cyclin B degradation. To test which possibility was correct, oocytes injected with mAb YL1/2 330 minutes after hormonal treatment (a time point at which the cyclin B concentration is high) were assayed for cyclin B concentration at time point 360 minutes (Fig. 7A, lane 9). Injected oocytes were indistinguishable from controls with regard to cyclin B concentration. Thus the mAb injection did not trigger cyclin B degradation.

These results strongly suggest that mAb YL1/2 injection blocks cyclin B synthesis without affecting cyclin B degradation. To further assess this possibility we tested whether the new synthesis of cyclin B was indeed inhibited in injected eggs. For this, control and mAb YL1/2 injected oocytes were pulse labeled with ³⁵S-labelled amino acids and assayed for content in labeled proteins at various time points. Cyclin B was prominently synthesized at 90 minutes in control oocytes (Fig. 7C, 90 minutes, lane 2) and labeling was elevated also after meiosis completion (180 minutes, lane 3; note that at that time, cyclin B was by far the major newly translated protein. This result was also observed in fully mature oocytes of the starfish Marthasterias glacialis). In injected oocytes, cyclin B labeling remained low (lane 4), as in GV oocytes (lane 1).

We also tested whether the inhibition of protein synthesis was at least in part specific for cyclin B synthesis. In general, cyclin B-dependent H1 kinase activity may regulate mitotic protein synthesis (Kanki and Newport, 1991; Galas et al., 1993). To test the effect of mAb YL1/2 microinjection on mitotic protein synthesis independently of cyclin B synthesis, we used Xenopus prophase-arrested oocytes, in which H1 kinase is totally repressed. Fig. 7D shows that overall mitotic protein synthesis was not inhibited by mAbYL1/2 microinjection.

The observed inhibition of cyclin B accumulation in eggs and oocytes injected with mAb YL1/2 could be a major factor responsible for the concomitant cell cycle arrest or it could be one perturbation among many. To test which of these possibilities was true we used a system in which cyclin B proteolysis does not occur and assessed whether or not such inhibition was sufficient to suppress the mAb YL1/2 effects on cell cycle.

When starfish oocytes are enucleated, meiotic and mitotic cell cycles can still be induced by hormonal treatment: we have previously shown that maturing enucleated starfish oocytes undergo meiotic and mitotic cycles almost in schedule with control oocytes (Picard et al., 1988). Furthermore, the periodic oscillations of H1 kinase activity occur in the absence of cyclin B degradation. This is shown in Fig. 8A for A. aranciacus oocytes, from which the whole germinal vesicle content had been removed. Such oocytes still showed periodic oscillations of H1 kinase activity. However, the drop of H1 kinase was delayed by 10-20 minutes compared to controls (Fig. 8A, red line). In addition, the levels of H1 kinase activity were by far higher before the drop and after the rise in enucleated than in whole oocytes (compare red and black lines). These high levels of H1 kinase activity are accounted for by an inhibition of cyclin B degradation both during meiosis I and meiosis II: the cyclin B levels remained constant during meiosis, as shown by immunoblot assay of cyclin B levels in the cycling enucleated oocytes (Fig. 8B). Interestingly, mAb YL1/2 microinjection into maturing enucleated oocytes did not inhibit H1 kinase reactivation after its drop at the end of meiosis I (Fig. 8A, blue line), whereas a normal inhibitory effect of the antibody was observed in control oocytes (green line). Thus mAbYL1/2 does not affect H1 kinase activity in the presence of maintained high cyclin B concentrations.

These experiments indicate that the observed inhibition of cyclin B synthesis in starfish oocytes injected by mAb YL1/2 is the major factor responsible for the concomitant cell cycle arrest.

Microinjection of mAb YL1/2 into cells can induce a disorganization of the microtubule network (Wehland and Willingham, 1983; Warn et al., 1987) and this raised an important question: is the C terminus of α-tubulin directly involved in the regulation of the cell cycle or are the observed effects of mAb YL1/2 injection secondary to the activation of microtubule-dependent checkpoints?

We used nocodazole-treated oocytes to test whether or not the mAb YL1/2 effects on the cell cycle were microtubule dependent. We and others have previously shown that the succession of meiotic and cleavage cell cycles is not inhibited by complete depolymerization of microtubules (Picard et al., 1987a; Picard et al., 1987b; Gerhart et al., 1984). Thus, maturing starfish oocytes still show both successive cycles of appearance and disappearance of the nuclear envelope (Fig. 9, upper panel) and cycles of H1 kinase activity (not shown) in the presence of 20 μg/ml nocodazole. In contrast, nocodazole-treated oocytes microinjected with mAb YL1/2 undergo complete cell cycle arrest, as shown by the persistent presence of prokaryons (Fig. 9, lower panel). This observation indicates that the interphase arrest induced by YL1/2 microinjection is microtubule-independent. Further support for this conclusion came from microinjection experiments with mAb TUB2.1. mAb TUB2.1 has the same effect as mAb YL1/2 on microtubule organization in cycling cells (data not shown) and is nevertheless ineffective in blocking the cell cycle in starfish oocytes (see above). These data strongly suggest that microtubules are not involved in mAbYL1/2-induced cell cycle blockade, although free tubulin dimers remain likely targets for YL1/2.

Besides Cdk1, another major player in the orchestration of the cell cycle is ERK2/MAP kinase. ERK2 is the major MAP kinase activated in meiosis. Furthermore, there is strong evidence that ERK2 activity is involved in the regulation of cyclin B synthesis (De Moor and Richter, 1997; Ballantyne et al., 1997; Fisher et al., 1999). In oocytes of both vertebrates (Sagata et al., 1988) and invertebrates (Tachibana et al., 2000), ERK2 activity is regulated by c-mos, a protein whose synthesis and accumulation could conceivably be affected by mAb YL1/2 injection. Therefore we tested whether or not perturbations of c-mos accumulation and of ERK2 activation could account for the observed inhibition of cyclin B synthesis by mAb YL1/2 in injected cells.

In all models, both invertebrate and vertebrate, ERK2 is active during the prometaphase and metaphase of the first meiotic division. ERK2 activation is thought to be implicated in the regulation of microtubule dynamic instability (Gotoh et al., 1991) and in the inhibition of DNA replication between meiosis I and meiosis II (Furuno et al., 1994; Ohsumi et al., 1994). In starfish, ERK2 is activated after GVBD. In starfish and in Xenopus, ERK2 remains active until the second meiotic cell cycle (Sagata et al., 1989; Picard et al., 1996), as shown by probing in-gel MBP kinase activity (Fig. 10A,B). In the present study, when YL1/2 was microinjected after GVBD in either starfish or Xenopus oocytes, ERK2 activity dropped rapidly and remained low at subsequent time points (Fig. 10). Additionally, Fig. 10 shows that there is also a strong inhibition of c-mos accumulation in injected Xenopus oocytes (Fig. 10B), correlative to loss of MAP kinase activity.

These results may suggest that mAb YL1/2 injection primarily affects c-mos translation, with subsequent effects on cyclin B synthesis. However, a careful comparison of Figs 7B and 10B suggests a different sequence of events: apparently cyclin B1 disappears rapidly after mAb YL1/2 microinjection in maturing Xenopus oocytes, well before ERK2 inactivation. Therefore, the effect of mAb YL1/2 on cyclin B synthesis is probably not due either to the observed inhibition of c-mos accumulation or to ERK2 inactivation.

The present work establishes that microinjection of antibodies specific for the tyrosinated C terminus of α-tubulin induces interphase arrest in meiotic starfish and Xenopus oocytes, and in mitotic cells of starfish early embryos. These observations fit well with previous experiments reporting that mAb YL1/2 microinjection into fertilized sea urchin eggs either impedes the G₂-M transition or induces a return to interphase, depending on the time of microinjection (Oka et al., 1990). Furthermore, our observation of similar mAb YL1/2 effects in Xenopus oocytes to those in starfish and sea urchin oocytes excludes the possibility that the interphase arrest caused by YL1/2 microinjection is the result of a singularity in echinoderm cells. However, in Drosophila early embryos and in some somatic mammalian cells, mAb YL1/2 injection induces a mitotic arrest instead of an interphase arrest (Warn et al., 1987; Wehland and Willingham, 1983). Apparently, in Drosophila, the antibody does not diffuse far from the injection area. The discrepancy between the responses of early embryos compared to mitotic cells may be accounted for by prominent differences in regulation of cyclin B synthesis, as discussed below.

mAb YL1/2 microinjection during the first meiotic cell cycle induces interphase arrest between meiosis I and meiosis II, without DNA replication in the reformed prokaryons. In contrast, microinjection during the second meiotic cell cycle or just after second polar body emission, both in fertilized (Fig. 5Ad) and unfertilized (not shown) eggs, induces arrest in G₂ after DNA replication. This indicates that the arrest is not an unspecific arrest of the cell cycle progression, but rather corresponds to an inhibition of the mitotic H1 kinase activation. mAb YL1/2 microinjection does not inhibit meiosis reinitiation in either starfish or in Xenopus. Therefore the possibility that it counteracts the post-translational activation of Cdk1 caused by dephosphorylation of T14 and Y15 is excluded. Thus, the observed inhibition of H1 kinase can be due either to inhibition of cyclin B accumulation, or to inhibition of cyclin B binding to Cdk1. We show that mAb YL1/2 microinjection results in an inhibition of cyclin B accumulation. Moreover, it has no detectable effect in enucleated maturing oocytes, which fail to degrade cyclin B. These data strongly suggest that keeping a low concentration of cyclin B is the only way by which mAb YL1/2 induces interphase arrest. This low cyclin B concentration can be obtained by inhibiting cyclin B synthesis, or by maintaining active the machinery responsible for cyclin B degradation. We show by pulse ³⁵S incorporation into proteins, that mAb YL1/2 strongly inhibits cyclin B synthesis (Fig. 7C), and by western blotting that it is unable by itself to open the cyclin B degradation window (Fig. 7A).

A peculiarity of early embryogenesis is that RNA transcription is not required for cell cycles to succeed: while progression in G₁ phase is essentially under transcriptional control in somatic cells, complete inhibition of ARN production by Actinomycin D or α-amanitin inhibits only the transition from early to late embryogenesis (MBT: Mid Blastula Transition) after several unperturbed cell cycles (Newport and Kirschner, 1984). In starfish, as well as in sea urchin, cleavage cycles occur normally despite complete removal of genetic material by enucleation or aphidicolin inhibition of DNA-polymerase α (Nagano et al., 1981; Picard et al., 1988). This shows that during early embryogenesis, protein synthesis is regulated mainly at the translational level, not the transcriptional level. The proteins whose synthesis is required for progression into M phase are translated from a pre-existing pool of stable maternal messenger RNAs.

The use of Xenopus acellular extracts has allowed a more precise identification of the proteins required for the cycling of early embryonic cells. Cycling extracts are able to reproduce with a correct schedule some biochemical and cytological features of cell cycling: cycling of H1 kinase activity, cycling of nuclear envelope breakdown and re-formation, building and disappearance of mitotic spindle. In such extracts, Murray and Kirschner showed that synthesis of cyclin B alone is sufficient for cycling (Murray and Kirschner, 1989).

Taken together, these results show that the whole cell cycle oscillation of early embryogenesis is regulated essentially through modulation of the translation of stockpiled cyclin B messenger RNAs and through periodic destruction of cyclin B. In contrast, in somatic cells, cyclin B mRNAs are not stockpiled, and even the periodic production of cyclin B protein is also under transcriptional control. This probably explains the discrepancy between the effects of mAb YL1/2 microinjection that we observed in meiotic or cleavage cells and the results obtained in somatic cells. This also strongly suggests that the tubulin C-terminal epitope is mainly implicated in the particular regulation of cyclin B translation during early embryogenesis.

Our results show that mAbYL1/2 inhibits translation of at least one additional protein, c-mos or its starfish homolog (see Fig. 10). On the other hand, Fig. 7D shows that mAbYL1/2 injection does not trigger general, nonspecific, inhibition of protein synthesis, at least in systems where protein synthesis does not depend on H1 kinase activity. This shows that the translational regulation of cyclin B, and accumulation of a few additional proteins is oversensitive to inhibition induced by mAbYL1/2 microinjection: presumably, the binding of the antibody to its cellular epitope disrupts a multimolecular complex involved in the specific translation of these molecules.

In previous works, the observed effects of mAb YL1/2 on the cell cycle have been interpreted as a result of interaction of the antibody with the α-tubulin C terminus (Wehland and Willingham, 1983; Warn et al., 1987; Oka et al., 1990), not with a crossreacting protein. Our data strongly support this view. We find that both mAb YL1/2 and another independent and highly specific α-tubulin C terminus antibody have similar effects on the cell cycle and that these effects are observed in widely divergent species (starfish and Xenopus). Further indication that mAb YL1/2 action results from interaction with tubulin, not with a less abundant crossreacting protein, comes from the examination of dose-effect curves. The tubulin concentration in the cytoplasm of non-neuronal cells is thought to be about 20 μM (Hiller and Weber, 1978). Cell cycle blockage occurs when mAb YL1/2 concentration in eggs reaches approx. 3.3 μM. As there are two tubulin binding sites per molecule of antibody, this corresponds to a concentration of tubulin binding sites of approx. 6.6 μM. Since a sizeable part of oocytes is occupied by the nucleus and other compartments like the vitellus, from which the antibody is excluded, the antibody concentration in the soluble fraction of oocytes is higher than 6.6 μM. In routine experiments, the volume of the soluble fraction obtained following high speed centrifugation of Xenopus oocytes corresponds to about one third of the total egg volume. Applying this conservative estimate to starfish oocytes one obtains an estimate for the mAb YL1/2 concentration in the cytoplasm of about 20 μM, corresponding to saturating levels with regard to tubulin. Thus, our results are compatible with tubulin being the target of the antibody. In previous works, the only protein that has been shown to be efficiently immunoprecipitated by mAb YL1/2 is the small subunit of ribonucleotide reductase (rnr: Standart et al., 1985; Thelander et al., 1985). We confirmed this result with ³⁵S-methionine-labelled sea urchin extracts in the conditions described by Standart et al., but we showed additionally that mAb20C6, another monoclonal antibody that also recognizes the tyrosinated α-tubulin C-terminal peptide, and immunoprecipitates α-tubulin with a similar efficiency, failed to immunoprecipitate labelled rnr (Fig. 3B). Yet, mAb20C6 microinjection was found to inhibit cell cycle progression and cyclin B synthesis exactly like mAbYL1/2 mincroinjection, which is an unexpected result if the biological effect of mAb 20C6 is due to rnr inhibition. Moreover we confirmed in our two usual starfish species that 1 mM hydroxyurea, a potent inhibitor of rnr, fails to induce cell cycle arrest (not shown), as previously published (Yamada et al., 1988).

Additionally we find that, when injected in oocytes, synthetic peptides corresponding to the tyrosinated α-tubulin C terminus have a similar effect on cell division to mAb YL1/2, whereas detyrosinated peptides are ineffective. The simplest explanation is that tyrosinated peptides act as competitive inhibitors of tyrosinated-free tubulin dimers in the regulation of cyclin B synthesis. While they cannot completely rule out the possibility of a crossreaction with another regulatory protein, such observations strongly support a specific involvement of the tyrosinated α-tubulin C terminus in the cell cycle regulation.

The mechanisms by which tyrosinated α-tubulin could regulate cyclin B synthesis remain elusive. It may be the case that specific interactions of the tyrosinated α-tubulin C terminus with effectors of protein translation are involved. Grallert and coworkers have recently shown that the partial inactivation of a yeast general translation factor, the RNA helicase Ded1, was able to arrest cell cycle by impeding specifically cyclin B synthesis (Grallert et al., 2000). These authors showed additionally that only the full-length cyclin B mRNA, containing both 5′ and 3′ UTRs, required Ded1 for being translated, suggesting that Ded1 relieves an inhibition that is mediated by 5′ and 3′ cis-elements. Interestingly, in preliminary experiments, we affinity-purified an RNA helicase belonging to the same family on immobilized peptide Y, whereas the same protein did not bind peptide E (not shown). The ineffectiveness of detyrosinated peptides indicates a crucial importance of the presence or absence of the C-terminal tyrosyl residue on the properties of the α tubulin C terminus. This is compatible with an important role of the tubulin tyrosination-detyrosination cycle in the regulation of cell division during oocyte maturation. However, molecular tools are presently lacking with which to test directly the role of the tubulin tyrosine ligase, the enzyme that catalyzes α-tubulin tyrosination (Ersfeld et al., 1993), and of the tubulin carboxypeptidase, in starfish. The mammalian tubulin tyrosine ligase can be inhibited by mAb ID3. However, we have injected this antibody into starfish eggs without detectable alteration of tubulin tyrosination (data not shown). It will obviously be of great interest to test whether or not the tubulin tyrosination cycle is truly essential to cell cycle regulation in developing oocytes. Finally, a series of recent results indicate an important role for the tubulin C terminus. Tubulin tyrosination is systematically suppressed during tumor growth in animal models, tyrosinated tubulin is apparently required for cyclin B synthesis in developing oocytes and recent results show that invalidation of the ligase gene has fatal consequences in mice (C. Erck, unpublished results). These results cannot be put together in a single simple figure at this time but should eventually unravel unsuspected functions of the tubulin molecule, namely of the α-tubulin C terminus.

Our results may also help shed light on poorly understood regulations concerning mRNA translation in meiotic and cleavage cell cycles. The progression of the meiotic and cleavage cell cycles is essentially ensured by the regulated translation of maternal messenger RNAs. Among them, cyclin B mRNAs have been shown to play a leading role, and previous works have popularized the view that early embryogenesis cell cycle oscillation was governed by a sustained translation of cyclin B mRNAs and a periodic destruction of cyclin B protein, itself under the control of cyclin B-cdk1 activity. While the regulation of the ubiquitin-dependent cyclin B proteolysis is now well documented, very little is known on how the translation machinery favors cyclin B production. Tubulin is already known to regulate translation, at least its own translation by an autoregulatory mechanism. This complex mechanism involves both translation-related mRNA destruction (Cleveland, 1989) and repression of the translation itself (Gonzalez-Garay and Cabral, 1996). A stimulation of α-tubulin translation by β-tubulin overexpression has also been described (Gonzalez-Garay and Cabral, 1995). Our present results suggest an extension of the function of α-tubulin in the regulation of translation beyond the autoregulatory processes, at least during early embryogenesis. Additionally, our results indicate that the tyrosinated form of α-tubulin may play an essential role in this process and that this role involves the free nonpolymerized form of the protein.

We thank warmly Yasmina Saoudi, for originating the story, and Marcel Dorée, for giving us the opportunity of this fruitful collaboration. This work was supported by grants from ARC (Association pour la Recherche sur le Cancer), la Ligue Nationale contre le Cancer, and the Ligues Départementales de l’Aude, de l’Ardèche et des Pyrénées Orientales.