Initiation of Cyclin B Destruction at the Centrosome

Following mitotic entry, the next critical transition point in the cell cycle is the metaphase to anaphase transition. This is under the control of the spindle assembly checkpoint, which prevents anaphase onset until all chromosomes have achieved attachment to opposite spindle poles (reviewed in [58]). In an elegant approach that made use of cells containing two spindles, it was shown that one spindle could initiate anaphase despite the presence of mono-orientated chromosomes on the second spindle [59]. This implied for the first time that the molecular components of the spindle assembly checkpoint are physically restricted to the spindle structure itself. The obvious location for components of the checkpoint is the kine-tochore/centromere region where attachment of microtubules and tension generated by bipolar attachment can be monitored. In support of this model, multiple spindle checkpoint proteins including Madl, Mad2, BubRl (Mad3), Bubl, Bub3, Mpsl, Aurora B, Rod and Zw10, have all been localized to the kinetochore [58].

The main target of the spindle assembly checkpoint is the multi-subunit ubiqui-tin ligase known as the anaphase promoting complex or cyclosome (APC/C) [60]. Amongst other substrates, the APC/C polyubiquitylates securin and cyclin B targeting them for proteasome-mediated degradation. Destruction of securin initiates anaphase by releasing separase which in turn cleaves the centromeric cohesin molecules that tether the sister chromatids [61]. Destruction of cyclin B promotes mitotic exit. Recognition of substrates by the APC/C requires an additional adaptor subunit that at the time of the metaphase/anaphase transition is the Cdc20 (Fizzy/ Fzy) protein, and in late mitosis/Gl is the Cdhl (Fizzy-related/Fzr) protein [62]. The current view on how the spindle assembly checkpoint prevents anaphase onset is that checkpoint proteins, notably Mad2, Bub3 and BubRl, form a mitotic checkpoint complex (MCC) with Cdc20 preventing it from interacting with and activating the APC/C [63-66]. Once full bipolar attachment has been achieved, Cdc20 is no longer assembled into checkpoint complexes thereby allowing APC/ C-Cdc20 complexes to form initiating the polyubiquitylation of substrates.

The question we are interested in here is where the APC/C-Cdc20 is first activated: is it primarily at the kinetochore or could it rather occur at the centro-some/spindle pole? Circumstantial evidence has come from the localization of APC/C components, notably Cdc16 and Cdc27, as well as Cdc20 to the centrosome, first, in fixed cells [67, 68] and, more recently, in live cells [70, 71, 126]. In reality, though, the localization of these proteins is both complex and dynamic with the APC/C and Cdc20 being found at a number of locations in mitosis including kine-tochores and the cytoplasm. Furthermore, in Drosophila embryos, Cdc16 and Cdc27

do not show entirely overlapping patterns of localization or complex size raising the possibility that there may be multiple, distinct versions of the APC/C operating at different sites [69].

Although present at many sites throughout the cell, there is good evidence that the APC/C is first activated towards cyclin B at the spindle poles. Live cell imaging of GFP-tagged cyclin B destruction in HeLa cells [72], cellularized Drosophila embryos [126] and yeast cells [73] reveals a loss of protein that starts at the spindle pole before moving in a wave along the rest of the spindle and only then onto the cytoplasm (or nucleus in yeast). Perhaps even more exciting is the demonstration that, in a particular Drosophila mutant (cfo), centrosomes detach from their spindles and whilst this does not prevent destruction of cyclin B at the spindle poles, there is no destruction on spindles and embryos arrest in anaphase [74]. This experiment provides powerful evidence that destruction of cyclin B begins on spindle poles and requires an intact physical connection to the spindle to propagate the wave of destruction. It is also possible that a checkpoint is activated in response to centro-some detachment preventing further destruction of cyclin B. Intriguingly, in syn-cytial (early stage) Drosophila embryos, cyclin B is only destroyed on the spindle, whereas it remains present in the cytoplasm [71]. This spatially restricted pattern of destruction suggests that it cannot be the global activation of the APC/C itself that controls the timing of cyclin B destruction, since the APC/C is not spatially restricted. Instead, it is possible that spatial restriction of Cdc20 could be critical. In fact, Cdc20 does appear to be restricted to the spindle during the syncytial stage of insect cell development leading Raff and colleagues to propose that this is why destruction of cyclin B is limited to the spindle in these embryos [71]. Furthermore, they hypothesize that destruction of cyclin B throughout the rest of the cell might depend upon Cdh1 which is only expressed after cellularization, although contrary to the proposed model, Cdh1 is highly concentrated on centro-somes throughout the cell cycle. It remains to be tested whether such spatial restriction of Cdc20 and Cdh1 can explain the temporal pattern of cyclin B destruction in adult vertebrate cells where Cdc20 binds to the APC/C before Cdh1.

Another possibility is that Cdc20 is released from checkpoint complexes in the vicinity of the spindle pole thereby making this the first place that APC/C-Cdc20 can form (Figure 8.2). Following microtubule attachment at the kinetochore

Figure 8.2 Spatial regulation of cyclin B1 destruction at the metaphase/anaphase transition. ► In early metaphase, cyclin B1 is stable and localized to the spindle pole and spindle fibres. Live cell studies in Drosophila embryos indicate that the APC/C ubiquitin ligase is present at spindle poles at this time but is inactive due to the absence of the essential adaptor subunit Cdc20. Cdc20 is thought to be assembled into inhibitory complexes with spindle checkpoint proteins including Mad2, BubR1 and Bub3 at unattached kinetochores. Following microtubule attachment, these complexes are transported along spindle fibers towards spindle poles by minus end-directed motors such as cytoplasmic dynein. Once at the poles, the checkpoint complexes are somehow disassembled allowing Cdc20 to bind and activate the APC/C. For correct operation of the checkpoint, release of Cdc20 and activation of the APC/C depends upon complete attachment of all kinetochores to the spindle. It is not clear how or where this is controlled, but time-lapse imaging in human cells and Drosophila embryos reveals that destruction of cyclin B1 is first observed at spindle poles before spreading outwards along spindle fibres.

Checkpoint complexes form on unattached kinetochores

Kinetochore-MT attachment

Abscission Checkpoint
Cyclin B1 stable on spindle

Early Metaphase

Laie Metaphase

Full attachment of kinetochores to spindle fibres

complexes displaced

Anaphase many checkpoint proteins including Mad2, BubRl, CENP-E, Rod and Zw10, exhibit unidirectional migration from the kinetochores to the spindle poles along spindle fibers [75-77]. Based on the rate of these movements, it is believed that they are mostly driven by the minus end-directed microtubule motor, cytoplasmic dynein [77, 78]. This may be a mechanism to disseminate the checkpoint complexes throughout the spindle or to turn off the checkpoint following microtubule attachment. The next question though is how the Cdc20-checkpoint protein complexes are disassembled. Cdc20 is phosphorylated in early mitosis by Cdkl and MAPK to promote its association with spindle checkpoint proteins and prevent it from binding the APC/C [79]. If the dephosphorylation of Cdc20 that followed checkpoint inactivation occurred primarily at the spindle pole this would lead to disassembly of the checkpoint complexes and restricted formation of APC/C-Cdc20 at this site. Currently, this is pure speculation but could perhaps be addressed with Cdc20 phosphosite-specific antibodies in a similar approach to that described above for showing that Cdkl-cyclin B1 activation occurs first at the centrosome.

Finally, as well as binding of Cdc20, activation of the APC/C requires phosphorylation of APC/C subunits by Cdkl and Plkl, and dephosphorylation of sites phosphorylated by PKA [80-82]. The localization of Cdkl and Plkl to centrosomes during mitosis has already been discussed, so what about PKA? A fraction of PKA clearly localizes to interphase centrosomes [83, 84] as a result of binding to A-kinase anchoring proteins including AKAP450 (also known as AKAP350 or CG-NAP) and pericentrin that are concentrated at the centrosome [85]. Displacement of specific pools of PKA from mitotic spindle poles may involve a shift in binding preference from centrosomal to non-centrosomal AKAPs [86]. Equally important is the localization of the phosphatase, possibly PPl, which removes the phosphates added by PKA. So, although we argued above that APC/C proteins are not spatially restricted, they could still be locally activated by changes in their phosphorylation state.

Clearly, there are many experiments that still need to be done to prove whether or not the APC/C is activated first at spindle poles and, if so, to determine the mechanism for this and whether it relates to Cdc20 localization, activation or possibly phosphorylation of the APC/C. The above discussion has mostly focused on the destruction of cyclin B. Yet the destruction of other substrates may depend upon activation of the APC/C at other sites. Securin is localized throughout the cell as well as on the spindle in mitosis and, temporally, destruction of securin is coincident with that of cyclin B, at least within the constraints of current time-lapse imaging [87]. The localization pattern therefore does not preclude the possibility that securin destruction is initiated at spindle poles, but equally there is no strong evidence to say that it is. Despite this current gap in our understanding, there is growing acceptance that mitotic protein destruction is spatially regulated and that the spindle poles have an important role to play at least in initiating the destruction of cyclin B.

8.6 The Contribution of Centrosomes to Cytokinesis | 153 The Contribution of Centrosomes to Cytokinesis

Following the separation of chromosomes, cytokinesis, or division of cytoplasm, must occur to ensure an equal distribution of genetic material to the two daughter cells (reviewed in [88]). The spatial cues for cytokinesis are coordinated with chromosome segregation as the orientation of cell division is determined by the position of the mitotic spindle [89]. The first visible sign of cytokinesis is the formation of an acto-myosin based contractile ring, which forms perpendicular to the central spindle in late anaphase. As this begins to constrict, the plasma membrane inva-ginates and the cleavage furrow appears, a process that requires synthesis of new plasma membrane. As the furrow further constricts, the microtubule bundles of the central spindle become confined to the ill-defined structure known as the midbody that connects the dividing cells. The final step of mitosis is abscission, when the last remnants of cytoplasmic connections are broken to produce two identical daughter cells, signaling the end of cell division. The molecular processes of cytokinesis and abscission are complex and still far from understood. Intrigu-ingly, though, there is now a wealth of evidence that implicates the centrosome in a number of distinct events that ultimately lead to cytokinesis (summarized in Figure 8.3A).

Using different technologies, three groups recently asked whether cells lacking centrosomes can complete cytokinesis. Firstly, Khodjakov and Rieder used highly focused lasers to selectively obliterate centrosomes (see Chapter 10). Surprisingly, this did not prevent formation of a bipolar spindle [90], but it did interfere with spindle orientation presumably due to loss of astral microtubules [91]. The consequence of having spindles that lacked cortical attachment was incomplete chromosome separation and the formation of thin chromatin bridges connecting the daughter nuclei, a feature known to inhibit cytokinesis [92]. Indeed, 30-50% of cells with laser-ablated centrosomes failed to complete cytokinesis. Secondly, Hinchcliffe and Sluder used needle microsurgery to remove centrosomes together with a portion of cytoplasm from BSC-1 cells. Again, a significant fraction of the acentrosomal karyoplasts were delayed in mitosis and failed to complete cytokinesis [51]. Thirdly, Piel and Bornens showed that an acentrosomal Drosophila cell line, 1182-4, frequently exhibited incomplete cytokinesis leading to the accumulation oftwo or more connected interphase cells [93]. Taken together, these independent experimental approaches clearly indicate that centrosomes are essential for a robust separation of chromosomes, which in turn is needed for subsequent progression through cytokinesis. However, alone they do not necessarily reveal an intrinsic role for the centrosome in the biochemical pathways leading to cytokinesis.

In the context of this chapter, we are particularly interested in whether the cen-trosome acts as a signaling platform to direct events leading to cytokinesis, beyond simply determining the extent of chromosome separation or the plane of cell division. A more direct role for the centrosome in coordinating the timing of cell abscission is suggested by the behavior of individual centrioles during late

Centrosome

Figure 8.3 Proposed functions for the centrosome in cytokinesis. (A) Centrosomes have been implicated in a number of different processes that ultimately lead to, and in some cases are required for, mitotic exit and cytokinesis. On a temporal basis, these can be divided into mitotic spindle and contractile ring positioning, cleavage furrow and midbody formation, cell separation and abscission. However, we emphasize that there likely to be significant overlap in the biochemical pathways required for each of these endpoints. Examples of proteins that localize to mitotic centrosomes and are implicated in these pathways are indicated in dark blue. (B) One of the most intriguing questions relating to the role of the centrosome in cytokinesis is why the mother centriole migrates towards the midbody prior to cell abscission. HeLa cells are shown following methanol fixation and staining with antibodies against a-tubulin (green) and y-tubulin (red). DNA is stained with Hoechst 33258 (blue). Scale bar, 10 mm (see Color Plates page XXVI).

8.6 The Contribution of Centrosomes to Cytokinesis | 155

mitosis. Careful imaging of fixed and live cells revealed unexpectedly that just prior to cytokinesis there is a dramatic splitting of the two centrioles in each spindle pole [93, 94]. The mother centriole moves into the midbody, whilst the daughter cen-triole remains stationary in the center of its respective cytoplasm (see Figure 8.3B). As the mother centriole moves back into the cell center, cytokinesis is completed. The regulation of the movement of the mother centriole is not fully understood but the protein kinase p160ROCK (Rho-associated kinase) may be required since a small molecule inhibitor of this kinase can trigger premature migration of the mother centriole to the midbody and early mitotic exit [95]. These movements also depend on remodeling of the post-anaphase microtubule network as addition of nocodazole causes the mother centriole to remain at the midbody, inhibiting abscission, while removal of nocodazole causes immediate abscission [93] Thus, in certain cell types, mother centriole movements correlate very closely with the timing of cell abscission, although whether they are a necessary prerequisite still remains to be proven.

What could be the purpose of these dramatic centriolar migrations? One possibility is that as the mother centriole migrates away from the midbody, the interphase cytoskeleton is established allowing the generation of opposite forces which propel migrating cells apart [96]. However, during tissue morphogenesis in the animal, cell division and cytokinesis do not usually require cell migration. An alternative hypothesis is that the mother centriole actively transports signaling factors necessary for abscission into close proximity with the midbody. In fact a number of important regulatory proteins localize to the centrosome during early mitosis and then move to the midbody during late mitosis. These include protein kinases, such as polo, and motor proteins, such as the kinesins Eg5 or MKlps [12, 97-99]. MKlpl can bind MgcRacGAP (CYK-4 in C. elegans), a Rho family GTPase-activating protein, to make the centralspindlin complex. This complex is thought to mediate the microtubule bundling that occurs in the central region of dividing cells [100]. It has been proposed that, in Drosophila, the centralspindlin complex of Pav-KLP (the homolog of MKlpl) and RacGAP50C interacts with Pebble, a Rho1-GEF (guanine nucleotide exchange factor), and together this trimeric complex somehow positions the contractile ring and coordinates cytoskeletal remodelling during cytokinesis [101].

The MKlps also physically associate with and are phosphorylated by polo kinases and these processes may be promoted by centrosomal recruitment. Mutations in pavarotti, the Drosophila gene encoding Pav-KLP, or depletion of MKlp2 in mammalian cells leads not only to mislocalization of polo, but also to failure of cytokinesis [97, 102]. Likewise, disruption of polo activity either through genetic mutation or the use of siRNA oligonucleotides results in a failure to complete cytokinesis [103, 104]. The mechanism by which polo regulates cytokinesis is not fully understood although evidence suggests that polo-dependent phosphorylation of NudC (nuclear distribution gene C) is required [105]. Polo kinases also have an important role to play in activating the mitotic exit network in yeast (see below). Thus, a major function of the kinesin motors may be to transport polo, and other regulators, from the centrosome to the midbody. However, as this transport can occur along microtubules, it still does not explain why the mother centriole itself should need to visit the site of abscission.

Apart from the centrosome, other cellular organelles including the Golgi complex and endoplasmic reticulum contribute to the regulation of late mitotic events. In particular, it seems likely that endosomal pathways are required to deliver membrane to the site of cleavage furrow formation. However, even in this process, the centrosome may have a role to play. Arfophilin-2, an ADP ribosylation factor binding protein, is implicated in cytokinesis due to sequence homology with Drosophila nuclear fallout, a centrosomal protein implicated in cellularization and cytokinesis [106]. Arfophilin-2 binds Rab11, another protein implicated in regulating traffic through the recycling endosome compartment. Importantly, RNA-mediated interference of Rab11 in C. elegans leads to specific regression of the cleavage furrow at the final stage of abscission [107]. Both arfophilin-2 and Rab11 have been localized to the perinuclear region in the vicinity of the centrosome implying that cen-trosomes may contribute to cytokinesis through integrating distinct signals in the endosomal recycling pathway [108].

The completion of cytokinesis in fungal cells is absolutely dependent upon a checkpoint mechanism called the mitotic exit network (MEN) in budding yeast and septum initiation network (SIN) in fission yeast (reviewed in [109, 110]). These checkpoints operate though GTPase-regulated protein kinase cascades, many components of which are associated with the SPBs. Hence, it is entirely plausible that the SPB is acting as a solid phase platform to promote these signaling events in much the same way as described earlier for control of the G2/M transition. Importantly, the mitotic exit checkpoint in budding yeast is dependent on the cellular position of the SPB ensuring that mitotic exit and cytokinesis only occur after migration of the nucleus into the bud. The GTPase Tem1p binds to the spindle pole that migrates into the bud via the Bfa1p-Bub2p GAP complex and is kept inactive until late anaphase. Both Tem1p and Bub1p are associated with the daughter SPB. At this point Lte1p, a putative GEF, is released from the cortex of the bud and activates Tem1p, which binds cdc15, Dbf2 and Mob1 triggering the release of the phosphatase Cdc14p from the nucleolus. This in turn dephosphorylates Cdk substrates, promoting Cdk inactivation and allowing mitotic exit. Defects in cyto-plasmic microtubule interactions with the cell cortex and misalignment of the spindle delays Tem1p activation and mitotic exit, thus coordinating cell cycle progression with spindle positioning [111] (see also Chapter 4).

In animal cells, misaligned spindles also delay mitotic progression, raising the possibility that there is conservation of these processes between SPBs and centro-somes, and possibly a conserved spindle positioning checkpoint [112]. Evidence for this comes from the existence of mammalian homologs of some of the MEN components such as Cdc14p, Bub2p and Mob1p [113, 114]. Human Cdc14A phospha-tase localizes to the centrosome and its overexpression causes chromosome segregation defects and cytokinesis failure [115, 116]. Centriolin, a novel protein that localizes to the mother centriole as well as the midbody, shares a limited region of homology with the budding yeast MEN component Nud1p, which anchors the MEN complex to the SPB through direct interactions with Bub2p [117]. Deple-

8.7 A Role for Centrosomes in G1/S Progression? | 157

tion of centriolin by siRNAs causes cytokinesis failure and, ultimately, G1 arrest with chains of cells remaining interconnected by long intercellular bridges. These data support the idea that mammalian cells may possess a regulatory pathway similar to the MEN/SIN that coordinates the final stages of cell division. It is intriguing to speculate that the dependency of fungal cytokinesis on SPB positioning in some way reflects the way that abscission in mammalian cells may be dependent on the repositioning of the mother centriole. Hence, the mother centriole could anchor a regulatory pathway that controls the final stages of mitosis and promotes cytokinesis.

Clearly, the centrosome is intimately involved in late mitotic events. The challenge now is to understand how the centrosome contributes to cytokinesis at the molecular level. We discussed earlier how cyclin B1 destruction is initiated at the centrosome, and it has long been known that failure to degrade cyclin B1 prevents midbody formation and ultimately cytokinesis [118]. However, it is highly unlikely that the sole purpose of the centrosome with respect to cytokinesis is to degrade cyclin B1. The impressive migration of the mother centriole alone suggests a much more direct role for this organelle perhaps in transporting proteins to their site of action at the midbody. These proteins may form complexes with other components of the same signaling pathway whilst still at the centrosome or else after they arrive at the midbody. Either way, this would provide a mechanism whereby active complexes only exist when the inactive constituents come together at a specific localization within the cell, thus regulating the spatial and temporal aspects of the signaling cascade and checkpoints involved in transit through cytokinesis.

0 0

Post a comment