Cytokinesis: GAP Gap

Abstract

The central spindle regulates cleavage furrow formation and cytokinesis and is composed of antiparallel microtubules bundled by microtubule-associated proteins and kinesin motors. One key protein in the central spindle is a Rho family GTPase-activating protein, CYK-4/MgcRacGAP, whose target is the subject of two new studies that arrive at divergent models.

An actomyosin-based contractile ring is responsible for generating the force that drives cell division in animal cells. The position of the contractile ring is cooperatively determined during anaphase by astral microtubules that emanate radially from the spindle poles and the central spindle [1], a set of antiparallel bundled microtubules that lies between the two spindle poles. Assembly of the contractile ring is regulated by the small GTPase RhoA. Like most GTPases, RhoA is active when bound to GTP and inactive when bound to GDP. The balance between these two states is regulated by guanine nucleotide exchange factors (GEFs) that activate RhoA and GTPase-activating proteins (GAPs) that stimulate GTP hydrolysis. One critical Rho GEF in cytokinesis, ECT2, is recruited to the central spindle by a putative Rho family GAP, CYK-4/MgcRacGAP (hereafter called CYK-4) [2]. The isolated GAP domain of CYK-4 promotes GTP hydrolysis by Rho family members in vitro; however, it is more active against Rac1 and Cdc42 than against RhoA, despite the fact that RhoA plays a central role in cytokinesis and Rac1 and Cdc42 do not, creating an apparent paradox.

Two new papers [3,4] now investigate the requirement for the GAP activity of CYK-4 in cytokinesis and explore the identity of the target GTPase of CYK-4. While one study concluded that the GAP domain acts on RhoA, the other concluded that it acts on Rac1. Can these divergent observations be reconciled?

CYK-4/MgcRacGAP: A Signaling Nexus

Although the CYK-4 protein has a reasonably simple domain structure, containing a coiled-coil domain, a C1 domain and a RhoGAP domain, it has numerous interaction partners. In addition to acting upon Rho family GTPase(s), the protein forms a stable complex with the kinesin-6 family member ZEN-4/MKLP1 to form the centralspindlin complex, which bundles microtubules at the central spindle. In human and Drosophila cells, and probably in other eukaryotes, CYK-4 also binds to the Rho GEF ECT2 [2,5,6]. In human cells this interaction is critical for RhoA activation and is subject to multiple levels of phosphoregulation by kinases such as Cdk1–Cyclin B and Polo-like kinase 1 [7]. Additional evidence points to an interaction between CYK-4 and the actin-, myosin-, and RhoA-binding protein anillin [8]. While these interactions have been mapped to isolated domains and have been reconstituted in bimolecular reactions, there may be interplay between the binding proteins such that mutations in one domain may affect interactions mediated by another part of the protein.

Phenotypic Fingerprinting

In one of the new studies, Miller and Bement [3] used morpholino oligonucleotides to deplete endogenous CYK-4 from Xenopus blastomeres and reintroduced CYK-4 variants — either wild-type CYK-4, or a mutant lacking the GAP domain, or one in which the conserved catalytic arginine [9], the so-called arginine finger, within the GAP domain of CYK-4 is substituted by alanine. In the other study, Canman et al. [4] isolated temperature-sensitive mutations in Caenorhabditis elegans cyk-4 that conditionally inactivated CYK-4 as a consequence of amino-acid substitutions in regions reasonably close to the catalytic center. Although substitution of the catalytic arginine with alanine has previously been shown to diminish — but perhaps not fully inactivate [10] — the GAP activity of CYK-4, the effects of the new temperature-sensitive mutations have not been documented to date.

In the Xenopus study, replacement of endogenous CYK-4 with a GAP-deficient variant resulted in a broader and more intense accumulation of F-actin at the furrow (Figure 1) [3]. In this study, the activity of Rho family GTPases was monitored by following the cortical localization of GFP fused to protein domains that bind the active form of RhoA, Rac, and Cdc42. Expression of the GAP-deficient variant resulted in hyper-accumulation of active RhoA at the furrow. Notably, the zone of active RhoA widened significantly. No changes in the cortical accumulation of active Rac or Cdc42 were detected. The phenotypes caused by GAP-deficient CYK-4 were quite similar to those observed when constitutively active RhoA was injected. These data point to RhoA being a critical target of the CYK-4 GAP domain, implying that the GAP activity ensures a tightly focused zone of active RhoA. Interestingly, the authors observed oscillations of the contractile ring in embryos expressing a CYK-4 variant that lacked the entire GAP domain. These oscillations are reminiscent of those observed in cultured human cells depleted of the cytokinetic scaffold protein anillin, which binds to both CYK-4 and Rho [8,11,12]. The authors conclude that the CYK-4 GAP domain continuously inactivates RhoA during the process of ingression and that the GAP domain itself may play a mechanical role by immobilizing the ring and preventing its migration in the plane of the membrane.

Figure 1.

A model for the distribution of active forms of Rho family GTPases during cytokinesis in wild-type cells and cells expressing CYK-4 variants.

(A) Formation of an ingressing cleavage furrow is driven by an accumulation of RhoA and its effectors in the furrow region adjacent to the central spindle. (B) In Xenopus embryos, mutation of the catalytic arginine (CYK-4 GAP R→A) results in a broader and more intense zone of active RhoA. Deletion of the GAP domain (CYK-4 ΔGAP) also induces an increase in RhoA activation, but the furrow additionally shows lateral instability. (C) In C. elegans GAP mutants (CYK-4-GAP*), furrow ingression is attenuated and can be rescued by depletion of Rac1. This may be due to accumulation of ectopic Rac activity at the furrow or reduction of Rac activity throughout the cortex, allowing a weaker furrow to ingress more productively.

In the C. elegans study, the GAP domain mutants led to a reduced rate and extent of furrow ingression and reduced accumulation of myosin at the tip of the ingressing furrow [4]. Interestingly, furrow ingression in these mutant embryos resembled that seen in embryos depleted of the CYK-4-binding protein ZEN-4 and in embryos in which the CYK-4–ZEN-4 interaction was disrupted, with the exception that the mutations in the GAP domain did not affect central spindle assembly. Importantly, the cytokinesis defects in the GAP domain mutants could be substantially rescued by depletion of the Rac1 ortholog CED-10 or depletion of Arp2/3 complex members or simultaneous depletion of two Arp2/3 activators (Figure 1). The authors conclude that the absence of CYK-4 GAP activity leads to ectopic activation of Rac1 and Arp2/3-dependent actin polymerization.

An Alternative Interpretation

The GAP domain of CYK-4 is conserved in a wide range of eukaryotic genomes, suggesting that it is functionally relevant and that the domain is active — to varying extents — as a GAP for Rho family members. One must also consider that studies in human and fly cells, and now Xenopus blastomeres, suggest that the GEF activity of ECT2 is stimulated by an interaction with CYK-4 [2,3,6,13]. In these cell types, depletion of CYK-4 leads to a complete block in accumulation of RhoA effectors and a block in furrow ingression that is not significantly different from depletion of ECT2 [6].

It is not yet known whether RhoA activation in C. elegans results from an interaction between CYK-4 and ECT-2. However, this is likely, at least in part, because depletion of CYK-4 diminishes the rate and extent of ingression [4]. Although the newly isolated CYK-4 mutants contain mutations in the GAP domain and retain their ability to bind to ZEN-4/MKLP1, in the absence of biochemical data on their GAP activity and the effect of the mutations on the CYK-4–ECT2 interaction, it is difficult to conclude with certainty that only GAP activity is affected. Indeed, the cytokinesis phenotypes caused by these mutant proteins suggest a failure to fully activate RhoA. These mutants result in a furrow ingression phenotype that is similar to that previously reported for a CYK-4 mutant that disrupts the localization of the centralspindlin complex [14]. One defect that could be shared by these two mutants is a reduction in the accumulation of RhoA at the leading edge of the furrow, which in one case results from a conformational change that disrupts ECT2 activation and in the other from delocalization of centralspindlin. This view is supported by the finding that partial depletion of RhoA or ECT2 exacerbates the phenotype caused by the CYK-4 mutants [4].

The conclusion that CYK-4 targets Rac1 is derived from the striking finding that the phenotype of the CYK-4 mutants can be significantly rescued by depletion of Rac1 or downstream factors, including Arp2/3 complex. This appealing model is consistent with an earlier study reporting that a reduced dosage of Rac genes can suppress a developmental phenotype in Drosophila caused by depletion of the CYK-4 ortholog [15]. If the phenotype results from inappropriate Arp2/3 activation, increased F-actin should be observed in CYK-4 mutant embryos. Because CYK-4 concentrates just in front of the leading edge of the furrow, Arp2/3-nucleated F-actin would be most likely to accumulate at this site. Furthermore, Rac1 depletion would be expected to restore myosin accumulation to the furrow. These parameters were not examined by Canman et al. [4], but furrow ingression in the CYK-4 mutant embryos depleted of Rac1 appears significantly slower than in wild-type embryos, suggesting that myosin accumulation is not restored.

An alternative explanation also warrants consideration, however. Rac1 depletion may rescue the new CYK-4 alleles via a bypass mechanism. Embryos deficient in Arp2/3 exhibit cortical instability [16], suggesting that Arp2/3-mediated actin nucleation occurs throughout the embryo. Furrow ingression is driven by forces that build up at the furrow region and overcome the forces in the cortex that resist deformation. Therefore, depletion of CED-10/Rac1 or Arp2/3 might allow the weaker furrow observed in the CYK-4 mutants (and in embryos in which CYK-4 is not properly localized) to ingress more deeply by making the cortex more compliant (Figure 1). An analogous suppression has been observed during Dictyostelium cytokinesis [17].

Filling the GAPs

These studies should be considered in the light of earlier work that approached the same question. Using homologous recombination in chicken B lymphocytes, endogenous CYK-4 was replaced with a regulated version of CYK-4 [18]. CYK-4 variants lacking the GAP domain or with a mutated arginine finger were reintroduced and scored for rescue of cytokinetic defects resulting from repression of the wild-type copy. Surprisingly, the allele with a mutated arginine finger complemented the defect to the same extent as a wild-type transgene; in contrast, the mutant lacking the GAP domain failed to do so. The kinetics of cytokinetic progression, the distribution of the active forms of Rho family GTPases, and the distribution of Rho family effectors were not examined, however, making it difficult to evaluate how cytokinesis proceeds in these cells.

This question was also addressed in an earlier study in Drosophila embryos [13]. Deletion of three consecutive amino acids of the CYK-4 ortholog — including one of the residues mutated in the newly isolated C. elegans CYK-4 mutants — resulted in a protein that did not localize to microtubules, indicating that mutations in the GAP domain can cause unexpected secondary effects. A mutant carrying a deletion of three residues including the arginine finger was found to localize normally but failed to trigger furrow formation. Whether this defect resulted from hypo- or hyper-activation of particular GTPases was not examined.

In summary, CYK-4 GAP activity is either not required (B lymphocytes), or required to attenuate RhoA activity (Xenopus), or required to build a properly ingressing contractile ring (C. elegans and Drosophila). Furthermore, differences are seen between point mutation of the arginine finger and deletion of the GAP domain, notably defective anchoring of the contractile ring, suggesting that the GAP domain has functions beyond the regulation of GTPase activity.

It is extremely challenging to explain all of these data in a single coherent model. Several considerations suggest that a single answer may be elusive. First, the functional requirement for a GAP domain in a GTPase signaling pathway is necessarily related to the activity of the GEF — if the GTPase is only weakly activated, the GAP activity would be less critical. Second, the requirement for CYK-4 GAP activity could be obscured by the presence of other RhoGAPs. C. elegans embryos contain two highly related GAPs, RGA-3 and RGA-4, which play a significant role in limiting the amount of active RhoA [19,20]. Orthologs of RGA-3/4 are not readily apparent in other systems, suggesting that overall GAP activities may be somewhat variable. Third, not all GAP ‘inactivating’ mutations are equivalent: some may incompletely inactivate the GAP domain and may not destabilize the GAP–GTPase complex [10], whereas others may have additional indirect effects. Fourth, furrow formation is regulated by two parallel pathways [1], the efficiency of which appears to vary in different systems with possible differential requirements for CYK-4 RhoGAP activity. Collectively, these studies indicate that the CYK-4 GAP domain does more than simply attenuate Rho GTPase activity and they leave some gaps to fill.