|Home | About | Journals | Submit | Contact Us | Français|
Successful cell duplication requires orderly progression through a succession of dramatic cell-cycle events. Disruption of this precise coupling can compromise genomic integrity. The coordination of cell-cycle events is thought to arise from control by a single master regulator, cyclin:Cdk, whose activity oscillates. However, we still know very little of how individual cell-cycle events are coupled to this oscillator and how the timing of each event is controlled.
We developed an approach with RNA interference (RNAi) and real-time imaging to study cyclin contributions to the rapid syncytial divisions of Drosophila embryos. Simultaneous knockdown of all three mitotic cyclins blocked nuclei from entering mitosis. Despite nuclear arrest, centrosomes and associated myosin cages continued to divide until the midblastula transition. Centrosome division was synchronous throughout the embryo and the period of the uncoupled duplication cycle increased over successive divisions. In contrast to its normal actions, injection of a competitive inhibitor of the anaphase-promoting complex/cyclosome (APC/C) after knockdown of the mitotic cyclins did not interfere with the centrosome-duplication cycles. Finally, we examined how cyclin knockdown affects the onset of cellularization at the midblastula transition and found that nuclear cell-cycle arrest did not advance or delay onset of cellularization.
We show that knockdown of mitotic cyclins allows centrosomes to duplicate in a cycle that is uncoupled from other cell-cycle events. We suggest that high mitotic cyclin normally ensures that the centrosome cycle remains entrained to the nuclear cycle.
The temporal order of cell-cycle events detailed by elegant classical cell biological studies is key to successful cellular reproduction. We now recognize elaborate molecular circuits that create checkpoints to help order and coordinate events. Cyclins, and the associated cyclin-dependent kinase (cyclin:Cdk), have been labeled as the master cell cycle regulator, for they control progression from one cell-cycle phase to the next . Not surprising, cyclin:Cdk is extensively regulated. In most contexts, mitotic entry is controlled by complex inputs that govern the phosphorylation status of preformed cyclin:Cdk complexes, which facilitates abrupt activation of its kinase activity .
The rapid division cycles of early embryos appear to operate by different mechanisms and have been viewed as streamlined cycles stripped of growth inputs and other complexities . Nuclei oscillate directly between DNA replication and mitosis without intervening gap phases. Cdk inhibitory phosphorylation, although it occurs at low level during these cycles [4, 5], does not appear to inactivate a significant proportion of Cdk1 [6, 7]. Instead, cell-cycle transition in the early embryonic cleavage cycles has been postulated to operate by the accumulation of mitotic cyclins . In this model, the rate of cyclin protein synthesized from maternal message defines the time required to reach a critical threshold at which mitosis is triggered. Subsequent mitotic destruction of cyclin, promoted by the anaphase-promoting complex/cyclosome (APC/C), facilitates mitotic exit and is thought to reset the cyclin-accumulation timer.
If accumulation of cyclin to a threshold controls cell-cycle advance, then a failure to accumulate mitotic cyclins to the mitotic threshold should block all mitotic events under cyclin control. At least this is the expectation for events that are under positive regulation by a unique cyclin:Cdk threshold. However, some treatments of embryos with drugs affecting cellular processes differentially affect various mitotic events. For example, centrosome overduplication has been observed in Xenopus egg extracts and Drosophila embryos after inhibition of DNA synthesis [9, 10]. Additional centrosome duplication was also observed after inhibition of protein synthesis in Sea urchin eggs and Xenopus blastulae [11, 12]. In these cases, it was not determined how the treatments uncoupled centrosome and cell-cycle events.
The biology of Drosophila embryos has presented several challenges to the idea that cyclin accumulation times embryonic cell-cycle transition. The first 13 mitotic cycles in fly embryos are synchronous and occur in a common syncytial cytoplasm . During the earliest cycles (cycles 2–9), there is no oscillation in cyclin levels, no changes in the status of cyclin:Cdk phosphorylation, and no bulk oscillation in cyclin:Cdk activity . Nonetheless, cyclins do turn over, and stabilized versions of cyclins block cell-cycle progression . It has been suggested that localized cyclin destruction might contribute to the cycle , but localized destruction, although it can provide a means to exit mitosis, leaves a cyclin pool that would seemingly disturb a timing mechanism that is based on cyclin accumulation. Mitotic cyclin destruction is first detected at cycle 9, but only a minor fraction is destroyed . During cycles 10–13, the proportion of mitotic cyclin that is destroyed increases progressively, until 80% of the mitotic cyclin is degraded in anaphase of cycle 13. The lengths of these cell cycles can be influenced by changes in the dose of cyclin genes as well as inhibition of DNA synthesis, which most likely reflects activation of a checkpoint [6, 16, 17].
In the 14th division, these rapid maternally driven cycles end, and zygotic regulation begins . This transition, referred to as the midblastula transition or maternal-zygotic transition (hereafter referred to as MBT), coordinates a multitude of cellular events [18, 19]. In cycle 14, S phase is dramatically prolonged to 45 min, nuclei are cellularized, maternal messages are degraded, and the first gap phase (G2) is inserted. Onset of the MBT was originally attributed to the nuclear:cytoplasmic ratio, where the exponential expansion of nuclei titrates a maternal component required for cell-cycle progression [19, 20]. The titrated factor has remained largely unknown. Notably, genetic manipulation of cyclin:Cdk and its regulators affect onset of the MBT and have led to a model where the gradual decrease in cyclin:Cdk1 activity during the blastoderm divisions, with likely inputs through checkpoint mechanisms, initiates the MBT at some critical threshold [5, 16, 21, 22].
Here, we directly explore the contribution of the mitotic cyclins in coordinating the blastoderm divisions and onset of the MBT. Simultaneous RNAi of all three mitotic cyclins during the blastoderm divisions induced nuclear cell-cycle arrest, indicating that indeed cyclin synthesis is required for entry into the blastoderm mitoses. Despite the absence of threshold levels of mitotic cyclin, centrosomes and associated myosin cages divided around arrested nuclei. Centrosome divisions and movements remained relatively periodic and coordinated despite treatments that disrupt the normal rise-and-fall mechanism of cyclin concentration. We suggest that cyclin:Cdk above a certain threshold normally entrains centrosome division to the nuclear cycle until the anaphase decline in cyclin:Cdk1 activity. Additionally, we find that knockdown of mitotic cyclins and subsequent nuclear cell-cycle arrest did not advance or delay onset of cellularization, suggesting that neither a decline in cyclin:cdk activity nor a prolonged interphase is insufficient to trigger this marker of the MBT.
We coinjected double-stranded RNAs (dsRNAs) targeting the three mitotic cyclins, cyclin A, cyclin B, and cyclin B3, into late preblastoderm embryos (~cycles 8–9) and followed the embryos by live imaging. Rapid imaging allowed 4D analysis of multiple embryos and revealed highly reproducible phenotypes. Cell-cycle defects appeared first near the point of injection about 25–35 min after injection (Figure 1B and Movies S1 and S2 available online; in all cases, the injected pole is to the left). In the affected cycle, mitosis initiated near the uninjected pole close to the normal time, with a mitotic wave spreading toward the injected pole but stopping at a sharp transition (described below) behind which nuclei remained in interphase (Figure 1B and Movies S1 and S2). The difference in time between the first and last nuclei to successfully enter mitosis ranged from 2 to 5 min, a delay representing the maximum extension of interphase that occurred without disturbing mitosis. This interphase extension is less than that normally observed during the progressive slowing of the blastoderm cycles, suggesting that the normal interphase extension involves more than a progressive limitation in cyclin synthesis.
To assess the response to threshold levels of cyclin, we followed nuclei at the boundary between interphase arrest and successful mitosis. In a narrow intervening zone, nuclei entered mitosis but exhibited various mitotic defects (Figure S1; for a complete phenotypic description see Supplemental Results). These results suggest that cyclin levels can meet a threshold for certain mitotic events without providing activities sufficient for its proper execution.
Embryos first affected in cycle 13 initiated gastrulation normally without miscoordination between cycle 13 and cycle 14 regions (Movie S1). There were no subsequent mitoses in the arrested part of the embryo, whereas normally patterned mitosis 14 occurred near the uninjected pole. Thus, RNA interference (RNAi) blocked zygotic cyclin function in the areas in which it was effective against maternal function. When arrest preceded cycle 13, the next wave of mitosis halted before reaching the previously arrested nuclei to produce a striped embryo with different regions arrested in different cycles (Figure 1B and Movie S2). Regions arrested prior to cycle 13 did not appear to make organized movements during gastrulation (as timed by the uninjected end). Thus, many post-MBT events were unaffected by cyclin reduction and arrest in cycle 13 but were defective when arrest occurred in earlier cycles.
The efficacy of RNAi knockdown was assessed by reverse transcriptase-polymerase chain reaction (RT-PCR). We injected both ends of early blastoderm embryos to improve the exposure of the embryo to RNAi. Arrested embryos showed reduction in the messenger RNA (mRNA) of all three mitotic cyclins compared to controls injected with LacI dsRNA (Figure 1C).
To examine protein knockdown, we injected embryos expressing cyclin B-GFP (GFP: green fluorescent protein) with dsRNA targeting all three mitotic cyclins. Injected embryos arrested in cycle 12 or 13 failed to accumulate significant GFP over the next 2 hr, whereas controls became brightly fluorescent (Figure 1D). Quantification of fluorescence showed a 5-fold to 6-fold reduction in GFP signal (n = 3 embryos). We conclude that reduction of cyclin mRNA by RNAi reduces cyclin protein accumulation.
Foci of GFP-histone form during normal blastoderm interphases and become especially prominent approximately 15 min into cycle 14 at the heterochromatic apex of each nucleus . Arrest of earlier cycles by RNAi led to enhanced foci of GFP-histone (Figure 1E) and to regions of more intense staining with DNA dyes (Figure S2). The effect was dramatic in cycle 11, which exhibited hypercondensed interphase chromatin. Arrested nuclei did not stain positive for phospho-histone H3, suggesting that this compacted chromatin is not incomplete mitotic chromosome condensation (Figure S2). We suggest that cell-cycle arrest in blastoderm embryos allows chromatin reorganization and compaction.
Typically, blastoderm mitoses begin at the poles of the embryo and progress toward the equator . To test the importance of the poles to mitotic initiation, we injected dsRNA into both poles. Despite a block to mitosis at the poles, mitosis initiated and progressed properly in the center of the embryo (Figure 1F). Moreover, injection of cyclin dsRNA into the center of the embryo caused local interphase arrest in the injected region (data not shown). These findings suggest that cyclin input into mitotic entry is primarily local in the syncytial cytoplasm.
Downregulation of mitotic cyclin:Cdk in some contexts leads to DNA rereplication in the absence of mitosis [24, 25]. We tested whether continued DNA replication occurs upon cyclin-RNAi-induced interphase arrest. When fluorescently labeled nucleotide (dUTP) was injected after dsRNA injection at one pole, incorporation was confined to unaffected regions distal to the point of dsRNA injection (Figure S3). We conclude that knockdown of the mitotic cyclins in the blastoderm embryo arrests nuclei in interphase without inducing continued DNA replication.
Cellularization accompanies the MBT (Figure 2A). About 5–10 min into interphase 14, cortical folds develop, and these are actively extended toward the embryo interior until each nucleus is separated by membrane . Myosin regulatory light chain fused with GFP (RLC-GFP) marks the leading edge of membrane progression (Figure 2B and Movie S5). Movies in a superficial focal plane reveal orderly perinuclear rings of RLC-GFP that appear to fade as the rings move basally out of focus. In cross-section, a line of RLC-GFP at the membrane front moves toward the embryo interior.
Mitotic cyclin RNAi produced chimeric embryos with regions arrested in interphase 12, 13, or 14. Despite differing regions of nuclear density, cellularization began uniformly and at the correct developmental time in these embryos (Figure 2B and Movies S6 and S7). These data suggest that a decline in cyclin:Cdk1 activity, a prolonged interphase, or the underlying nuclear density is insufficient to initiate the MBT.
Though the schedule of cellularization was unaffected, cyclin knockdown disrupted the normal orderly rings of RLC-GFP in regions of the embryo arrested in cycle 12 (Figure 2B). Higher resolution imaging revealed the nature of this disruption (Figure 3). During blastoderm interphases, RLC forms a cage around each nucleus. These cages disassemble in mitosis and reform around daughter nuclei . After RNAi, the RLC-GFP cages, continued to divide while the nuclei, visualized as dark regions lacking RLC-GFP, remained arrested (Figure 3A and Movie S8). Uncoupled cage divisions near the injected pole occurred in synch with mitosis 11 and 12 in the uninjected end of the embryo. In regions arrested in cycle 13, cages did not divide; however, a thin strip of RLC-GFP appeared over a few nuclei at the time of cellularization (Figure 3B; 1:01:29), suggesting some aspect of cage division.
Because centrosomes contribute to cytoplasmic organization, we used centrosomal GFP-TACC to follow their dynamics. Centrosome duplication is adapted to the speed of the early cycles . The single centrosome (with a centriole pair) at each pole of the spindle splits during anaphase, so that each daughter nucleus has two associated centrosomes (each with singlet centrioles). The centrosomes move apart during interphase (duplicating their centrioles during the process) so that two mature centrosomes lie on opposite sides of the nucleus ready for the next mitosis. When cyclin RNAi blocked mitosis, the two centrosomes split anyway, and the resulting four centrosomes remained associated with the nucleus (Figure 4A and Movie S9). In regions arrested in nuclear cycle 11, centrosomes divided three additional times (Figure 4A), although the third division displayed some asynchrony, and not all centrosomes duplicated a third time. Additional centrosome separation was not observed in embryos arrested in cycle 13 (Figure 4B and Movie S10). We conclude that centrosome duplication does not require a nuclear cycle and that the MBT might influence these cytoplasmic events, because neither centrosomes nor RLC cages duplicated in cycle 13-arrested embryos. Moreover, we suggest uncoupled centrosome division creates a mismatch between nuclei and cytoskeleton organization, leading to the disorderly cellularization observed in embryos arrested prior to cycle 13.
The additional rounds of centrosome division occurred roughly in time with the nuclear cycles in the uninjected end of embryos, perhaps suggesting an influence from distant nuclei. However, dsRNA injection along the embryo length, which arrested the nuclear cycle throughout the embryo, was still followed by uncoupled centrosome division. The timing of centriole cycles was recorded from these embryos (Figure 4C). As in control embryos, the centrosome cycle in cycle 11-arrested embryos grew progressively longer, but at each cycle the uncoupled centrosome cycles were longer (by ~5.5, ~8, and ~10 min) than control centrosome divisions (n = 11 embryos). A later arrest (cycle 12) was associated with a somewhat slower cycle. These data show that multiple cycles of coordinated centrosome divisions occur without a nuclear cycle.
To further characterize the uncoupled centrosome divisions, we examined two additional proteins with unique mitotic localizations, Nuclear-fallout (Nuf) and Polo kinase. GFP-Nuf, a Rab11 effector, accumulates around interphase centrosomes (Figure 5A top and Movie S11; 00:00 and 09:40) and disperses throughout the cytoplasmin mitosis (metaphase 4:00, telophase 06:39) . After RNAi of the mitotic cyclins, GFP-Nuf remained focused at the centrosome even during centriole splitting and gradually became more tightly focused around the centrosome (Figure 5A bottom and MovieS12). These data show centrosome division without one hallmark of a mitotic cytoplasm.
Polo kinase contributes to various mitotic events, including mitotic entry and cytokinesis. GFP-Polo is centrosomal in interphase (Figure 5B top and Movie S13; 00:00), but a fraction of the protein enters the nucleus shortly before nuclear-envelope breakdown (02:20) and associates transiently with kinetochores. It relocalizes to the pseudo cleavage furrow upon mitotic exit (07:06). Even after interphase arrest following cyclin knockdown, GFP-Polo exhibits a cycle of relocalization. Nuclei became fluorescent, and kinetochore-like foci formed, but this fluorescence decayed after about 5 min without mitotic entry (Figure 5B bottom and Movie S14). Shortly thereafter, centrioles divided. There was no parallel to the normal Polo marking of a pseudo cleavage furrow. Prior to the next round of centriole division, GFP-Polo again entered the nucleus. The cyclical localization of Polo suggests that some regulatory oscillation occurs in conjunction with centrosome division, perhaps representing a signal to the nucleus presaging centrosome duplication.
In principle, cytoskeleton transformations might constitute a cytoplasmic cycle that underlies the uncoupled centrosome cycle. To test this, we injected colchicine into embryos that were poised in interphase for their first uncoupled centrosome division. Despite a reduction in microtubules and interphase arrest, centrosomes divided one additional time, suggesting an intact cytoskeleton is not required for centrosome division (Figure S4A; for a complete phenotypic description, see Supplemental Results).
The amino-terminal peptide of sea urchin cyclin B (su-CYCB13–110) is an APC/C substrate  and has previously been utilized as a competitive inhibitor for endogenous APC/C substrates . At high concentrations, it inhibits mitotic exit . Injection of suCYCB13–110 into normal blastoderm embryos arrested the ensuing mitosis and nuclei remained arrested for at least 1 hr (data not shown). To test for possible contribution of the APC/C to the uncoupled centrosome cycle, we injected suCYCB13–110 into embryos poised to begin uncoupled centrosome multiplication. Despite the presence of suCYCB13–110, centrosomes continued to divide around arrested nuclei with dynamics similar to the uncoupled cycle in control embryos (Figure S4B and Movie S16). Embryos whose nuclei arrested in cycle 11 exhibited up to three additional centrosome divisions in the presence of suCYCB13–110 (n = 7). Because suCYCB13–110 arrests wild-type embryos in mitosis without uncoupling centrosome replication cycles (data not shown), the continued centrosome divisions in the presence of suCYCB13–110 when cyclins are knocked down suggests the contribution (if any) of APC/Cto the uncoupled centrosome cycles is less important than it is in a normal cell cycle.
Inhibition of protein synthesis with cycloheximide induces interphase arrest in syncytial embryos within 1–2 cycles and has been presumed to act through blocking new cyclin synthesis. We examined histone-GFP embryos after cycloheximide injection to determine whether the effects paralleled those of cyclin RNAi. Similar to our findings following cyclin RNAi, histone-GFP-expressing embryos arrested after cycloheximide injection developed intense histone-GFP foci in the apical region of the nucleus about 13 min after interphase onset (Figure 6A and Movies S17 and S18). This condensation persisted for the duration of the experiment.
Cycloheximide injection into GFP-TACC-expressing embryos revealed that centrosomes separated at least one additional time after nuclear arrest in cycle 10 or 11 (Figure 6B and Movie S19). Often, embryos arrested in cycle 10 exhibited a second centrosome division, but at this division one daughter centrosome was less intense, suggesting incomplete duplication (Figure 6B; 01:07:49). Similarly, the single round of additional centrosome separation in cycle 11 embryos often produced unequal daughters (data not shown). Centrosome divisions observed after cycloheximide injection were generally slower than those observed after cyclin knockdown, particularly when a second cycle of centrosome division was observed (i.e., ~50 min cycle in Figure 6B). There was no additional centrosome separation in cycle 12- or cycle 13- arrested embryos.
We also tested whether RLC-GFP cages continued to divide after cycloheximide injection. Beginning 10–20 min after arrest, RLC-GFP cages became increasingly disorganized until just patches of RLC-GFP were distributed throughout the embryo. Nonetheless, a fraction of cycle 10-arrested embryos displayed one round of RLC-cage division prior to losing ordered structure (Figure 6C; 14:48 and 44:50). Duplication of centrosomes and RLC cages is consistent with effects seen upon cyclin knockdown; however, the rate of centrosome duplication and RLC disorganization suggests that continued protein synthesis is also needed for proper cytoskeletal function.
We have used RNAi in the early Drosophila embryo to examine the contribution of the mitotic cyclins to cell-cycle progression and to the MBT onset. Knockdown of the mitotic cyclins blocked nuclei from entering mitosis. Despite the failure of nuclei to enter mitosis, centrosomes and myosin cages continued to divide around arrested nuclei. We suggest that cyclin:cdk above a certain threshold normally acts to couple the centrosome cycle to the nuclear cycle until cells enter mitosis. Additionally, we show that knockdown of mitotic cyclins and subsequent cell-cycle arrest is not sufficient to initiate the MBT and suggest modification of previous models describing the MBT onset.
Interestingly, our data show that reduction in mitotic cyclin blocks mitosis without blocking centrosome duplication. We suggest that the uncoupled centrosome cycles represent full duplication cycles, because three cycles of centrosome duplication were observed, which requires more than a division of previously duplicated centrosome components. Furthermore, the appearance and movements of centrosomes is similar to that seen in normal cycles. How might cyclin knockdown bypass the normally tight coordination between centrosome and nuclear division?
After cyclin knockdown, we see reduced accumulation of a GFP-cyclin reporter and reduced cyclin A on Western blots (data not shown). The residual cyclin could retain some function, but it is not adequate to facilitate mitosis. Thus, the level of mitotic cyclin required for centrosome multiplication (if any) is less than the level required for mitosis. Accordingly, in a standard cell-cycle paradigm, cyclin accumulation would first satisfy the threshold for centrosome multiplication and only later reach the mitotic threshold, followed by the resetting of the cycle by mitotic cyclin destruction. A simple interpretation of our results is that we have arrested cyclin accumulation between the two thresholds so that mitosis is blocked but centrosome multiplication proceeds. However, if the centrosome cycle is mechanistically coupled to the mitotic cycle, one might expect that blocking mitosis would also block the centrosome cycle, unless mitotic cyclins are also required for the coupling mechanism. Indeed, there are a number of indications that mitotic cyclins influence the centrosome cycle. Moreover, there are also observations that suggest several points of coupling of the centrosome-cycle and cellcycle progression.
The centrosome-duplication cycle normally occurs in lock-step with progress through the cell cycle. During syncytial mitoses, the centrioles of a centriole pair disjoin at the transition to anaphase, daughter centriole assembly begins in anaphase, and centrosomes move apart during interphase. Subsequent shifts in this coordination occur in parallel with changes in cyclin:Cdk regulation. When a G2 phase appears in cycle 14, completion and maturation of daughter centrioles is held in abeyance until expression of Cdc25stg . Furthermore, when a G1 appears in cycle 17, initiation of daughter centrioles is deferred because of cyclin E downregulation . Apparently, multiple steps of the centrosome cycle are coupled to the cell cycle, and previous work suggests various ways that cyclin:Cdk might couple the centrosome and mitotic cycles.
Conceptually, the once-per-cell-cycle duplication of centrosomes is similar to the regulation of DNA replication . DNA replication is coupled to oscillations in cyclin:Cdk activity because cyclin:Cdk inhibits one step of replication but is required to promote another. However, centrosomes have been found to duplicate in experimental conditions apparently lacking oscillations of cyclin:Cdk1 . In a Xenopus egg extract arrested with low mitotic cyclin:Cdk1 kinase activity by inhibition of DNA synthesis, centrosomes continued replicating in a cyclin E dependent fashion . Thus, cyclin E:Cdk2 makes a positive contribution to the centrosome cycle, but centrosomes multiplied in its continuous presence, indicating that this cyclin:Cdk does not block duplication. Similarly, upon deletion of S. cerevisiae Clb 1–4 (G2 cyclins), uncoupled duplication of the spindle-pole bodies occurred in the continuous presence of Cln2 (G1 cyclin) or Clb5 (S phase cyclin) . Thus, G1 cyclins and/or S phase cyclins promote centrosome duplication without blocking it. Importantly, in normal cycles the G1 cyclins do not provoke multiple rounds of uncoupled centrosome division.
How might such divisions be suppressed? Interestingly, as in the above experiments in Xenopus and yeast, treatments that eliminate or suppress mitotic cyclin:Cdk1 seem to uncouple centrosome replication. Centrioles multiplied without mitosis in Drosophila upon temperature inactivation of a Cdk1ts, and centrosomes amplified in sea urchin and frog embryos arrested by inhibition of protein synthesis, which presumably blocks cyclin accumulation [11, 12, 35]. Thus, our findings are in accord with previous observations in suggesting that mitotic cyclins are required to enforce coupling of the centrosome cycle to the mitotic cycle.
Suppression of uncoupled centrosome cycles by mitotic cyclin:Cdk1 could be the result of inhibition of one or more steps of the centrosome cycle. Indeed, stabilized versions of the mitotic cyclins or inhibition of the APC/C blocks mitotic exit and blocks daughter-centriole production, showing that mitotic cyclins have either a direct or indirect inhibitory action on centrosome replication .
Because several steps of the centrosome cycle appear coupled to the cell cycle, the cyclin inputs might be complex. For example, the finding that Cdc25stg promotes daughtercentriole maturation in G2 of cycle 14 suggests that cyclin:Cdk1 activation is required for centrosome maturation . However, this is not easily consistent with the observation that inactivation of Cdk1ts allows centriole multiplication without a deficit in daughter-centriole growth . Another possibility is that Cdc25stg removes an inhibitor of daughter-centriole maturation. Indeed, tyrosine phosphorylated Cdc28 of S. cerevisiae inhibits spindle-pole-body duplication, and Cdc25Mih1 reverses this inhibition . In summary, present evidence is consistent with direct or indirect inhibition of centrosome duplication by mitotic cyclin.
We note that the multiple centrioles produced after inactivation of Drosophila Cdk1ts did not separate and that the separation of yeast spindle-pole bodies requires active Cdc28 [35, 36]. We suggest that there is also a positive contribution of mitotic cyclin:Cdk1 to centrosome multiplication but that this requirement is either absent in the early syncytial cycles or that it is satisfied by a low level of mitotic cyclin that persists following RNAi. If a residue of Cdk1 promotes the uncoupled centrosome cycle, why does it not also promote mitosis? Perhaps the cyclin level is too low, the residual Cdk1 activity is localized to the centrosome, or the nuclear cycle is prevented by an unappreciated checkpoint.
If cyclin:Cdk1 provides both negative and positive contributions to the centrosome cycle, a simple model could explain coupling of the centrosome cycle to mitosis. G1 cyclins promote centrosome duplication but also trigger mitotic cyclin accumulation . If kinase inactive mitotic cyclin:Cdk1 inhibits a step in centrosome maturation, this would ensure centrosomes do not divide until mitotic entry, whereupon Cdk1 activation would allow completion of centrosome duplication. Furthermore, if active cyclin:Cdk1 kinase and metaphase activities suppressed centrosome separation, separation of the duplicated centrosome would await mitotic exit. Additional studies will be required to define this multistep coupling mechanism.
The switch from maternal to zygotic regulation at the MBT involves a wholesale reorganization of many regulatory circuits. Although there has been great interest in the mechanisms that time and coordinate this transition, little is known about either the timer or the mechanism. Experiments in frog and fly have suggested that the MBT occurs when the exponential multiplication of nuclei increases the nuclear to cytoplasmic ratio to a threshold [19, 38]. But what provides the readout of the increasing nuclear density? In flies, the capacity to promote mitotic cyclin destruction correlates with an increase in the nuclear to cytoplasmic ratio, or, as emphasized by some authors, it also correlates with the increase in centrosomes and mitotic apparatuses . This relationship between cyclin degradation and nuclear concentration might explain the gradual prolongation of the blastoderm cycles and onset of the MBT. This interphase prolongation has been suggested to allow time for transcription of components necessary at the MBT . Accordingly, knockdown of cyclin synthesis should dramatically influence MBT timing, perhaps directly if cyclin:Cdk levels provide a regulatory input or indirectly if cell-cycle length or nuclear density provides an input.
Knockdown of mitotic cyclins blocked mitosis at the injected pole, modestly extended the cell-cycle length in more distal regions, and usually left the cycle unaffected at the most distal pole. When the unaffected end of such chimeric embryos completed cycle 13, ingression of the cellularization membranes occurred in concert in regions of the embryo in cycle 14, cycle 13, and cycle 12. Thus, local knockdown of cyclins, prolongation of the cell cycle, and reduction of local nuclear density was not sufficient to forestall cellularization. Although we have yet to characterize other MBT parameters, the apparently normal gastrulation of embryos arrested in cycle 13 suggests concerted transition of the various MBT events. These findings are not easily consistent with earlier ideas because the experiment alters many parameters thought to contribute to triggering the MBT. We note that one parameter that is not changed by cyclin RNAi is the increasing centrosome density, which remains a viable candidate for triggering the MBT.
We thank members of the O’Farrell lab (especially Tony Shermoen and Soo-Jung Lee) for helpful comments, and Bill Sullivan and Anne Royou for sharing unpublished data. This research was supported by National Institutes of Health (NIH) postdoctoral fellowship GM8464973 to M.L.M. and by NIH grants GM37193 and GM60988 to P.H.O’F.
Supplemental Data Supplemental Results, Experimental Procedures, four figures, and 19 movies are available at http://www.current-biology.com/cgi/content/full/18/4/245/DC1/.