While there is compelling evidence that CYK-4 plays a role in RhoA activation, the role of its GAP domain remains contentious. The isolated GAP domain of CYK-4 promotes GTP hydrolysis in vitro by Rho family members but has much stronger activity toward Rho GTPases Rac and Cdc42 when compared with RhoA ^{22, 37, 38} . Importantly, however, Rac1 and Cdc42 are dispensable for cytokinesis in a wide variety of contexts ^{36, 39– 41} . Thus, do these in vitro assays properly reflect the function of full-length CYK-4 during cytokinesis in vivo? Does it function as a conventional GAP that inactivates Rac or RhoA or both, or does it participate, unexpectedly, in activating RhoA via ECT-2, or a combination thereof?

CYK-4 was first implicated in cytokinesis on the basis of a temperature-sensitive lethal mutation— cyk-4(t1689ts)—in C. elegans encoding an S15L substitution ( Figure 1A) in the N-terminus of CYK-4 that disrupts its interaction with ZEN-4 ^{16, 42} . Subsequently, Canman and colleagues isolated two additional mutations— cyk-4(or749ts), yielding CYK-4 ^E448K, and cyk-4(or570ts), yielding CYK-4 ^T546I—and both substitutions map to the GAP domain of CYK-4 ⁴³ . However, these mutations are not in the active site, and the biochemical nature of these alleles has not been characterized. Therefore, it is risky to assume that these mutants are solely defective in CYK-4 GAP activity. Indeed, CYK-4 ^E448K exhibits a defect in membrane localization in the gonad ³⁶ . However, because the mutants assemble normal central spindles, they cannot be entirely unfolded at the restrictive temperature. More recently, a mutant targeting the conserved catalytic arginine in the GAP domain was generated (R459A), as was a variant where two surface residues were mutated (K495E and R499E) to block CYK-4 interaction with Rho family GTPases ³⁶ . None of the cyk-4 mutations mentioned here abolishes furrow formation in otherwise wild-type embryos; furrows ingress partially and regress ( Figure 1).

The debate regarding CYK-4 GAP domain function was initiated by the discovery that depleting the GTPase Rac1 (CED-10 in C. elegans) in CYK-4 ^E448K embryos frequently results in complete ingression of the furrow ⁴³ . This was interpreted to mean that the primary function of the CYK-4 GAP domain during wild-type cytokinesis is to inactivate Rac1/CED-10 ^{43, 44} . The experimental result ( Figure 1B) is uncontested and has been confirmed by using either RNAi ^{36, 43} or loss-of-function alleles ^{44, 45} to inactivate Rac1/CED-10. However, there is significant disagreement regarding interpretation of the results and the underlying mechanism of CYK-4 GAP function.

The presence of a parallel pathway for furrow induction contributes to the difficulty in interpreting results concerning the role of the CYK-4 GAP domain. If a given GTPase is the sole, relevant target of the CYK-4 GAP domain during cytokinesis, then its loss should reverse the defects of CYK-4 GAP mutants in a variety of genetic contexts. However, inactivation of Rac1/CED-10 rescues furrow ingression in GAP domain mutants (E448K, R459A, and T546I) ^{36, 43– 45} only in the presence of NOP-1 ³⁶ . In NOP-1 mutant embryos, which are cytokinesis proficient, mutations in the GAP domain of CYK-4 block furrow formation altogether and the activity of Rac1/CED- 10 has no bearing on this behavior ( Table 1) ³⁶ . If mutations in the GAP domain of CYK-4 indeed cause Rac1/CED 10 hyperactivation, then inactivation of Rac1/CED-10 should correct this defect regardless of the presence or absence of NOP-1.

In addition to NOP-1 inactivation, there are other ways in which the centralspindlin-dependent cytokinetic furrow in C. elegans embryos can be isolated from the alternative pathway (see “CYK-4 protein: a brief introduction”). In each case, the CYK-4 ^E448K mutation prevents centralspindlin-dependent furrow formation, and, again, inactivation of Rac1/CED-10 does not restore furrowing ⁴⁵ . Stated another way, loss of Rac1/CED-10 reverses the furrowing defect in CYK-4 GAP mutants only when the centralspindlin-independent pathway for furrow formation is active (that is, when an alternative pool of locally activated RhoA is available). Strikingly, in their recent publication, Canman and colleagues did not use any of these approaches to test whether Rac1/CED-10 is a direct target of CYK-4 GAP activity ⁴⁴ .

To analyze CYK-4 GAP domain function using cyk-4 mutants in a meaningful way, all aspects of a mutant phenotype must be considered. In addition to being incomplete, furrows in embryos with cyk-4 GAP domain mutations have slower ingression kinetics and weakly accumulate active RhoA and its effectors. Although Rac1/CED-10 depletion in cyk-4 GAP mutant embryos permits the completion of furrow ingression, it is unable to restore the other two phenotypes ^{44, 45} . If Rac1/CED-10 was the primary relevant target of CYK-4, embryos deficient in both Rac1/CED-10 and CYK-4 GAP activity would be predicted to resemble Rac1/CED-10–deficient embryos (which closely resemble wild-type embryos).

Canman and colleagues recently suggested that mutations in the GAP domain of CYK-4 do not impair myosin accumulation ( Table 1, asterisk) ⁴⁴ , despite Canman and colleagues’ earlier work demonstrating a defect in myosin at the furrow tip in cyk-4 mutant embryos (see Figure S3 in 43). A variety of mutations in the CYK-4 GAP domain impair the accumulation of myosin, actin, anillin, and a RhoA biosensor ( Table 1, asterisk) ^{2, 36, 45} . Again, these defects are not corrected by depleting Rac1. This is particularly dramatic in NOP-1–deficient embryos where myosin and RhoA activation during cytokinesis are entirely dependent upon centralspindlin ^{2, 36} . These results cannot be explained if the CYK-4 GAP domain acts on Rac1/CED-10. Rac1/CED-10 activates the Arp-2/3 complex ^{46, 47} , a nucleator of branched actin filaments ⁴⁸ . If CYK-4 GAP mutations prevent Rac1 inactivation, hyperaccumulation of Rac1/CED-10 effectors or an increase in branched actin on the membrane would be expected in these mutants. But, in fact, cyk-4 mutations result in a reduction in actin accumulation ⁴⁵ . Critically, there is no evidence that mutations in the GAP domain of CYK-4 cause ectopic Rac1/CED-10 activation in the C. elegans embryo ( Table 1 and Figure 2A). Collectively, the available data do not support models in which Rac1/CED-10 is the primary relevant target of CYK-4 GAP activity during cytokinetic furrowing.

Why does loss of Rac1/CED-10 allow CYK-4 GAP mutants to complete furrow ingression? Inactivation of Rac1/CED-10 partially rescues the cytokinesis defect resulting from selected mutations in CYK-4 and only in the presence of a redundant furrow induction pathway. The extremely specific situations in which restoration to full ingression is observed strongly support that this behavior is mediated by bypass suppression. That is, instead of correcting the primary defect resulting from the mutation in CYK-4, the absence of Rac1/CED-10 attenuates the requirement for RhoA activation in furrow ingression. Depletion of ARX-2, a subunit of the ARP-2/3 complex, has a similar effect ^{43, 45} . Inactivation of Rac1/CED-10 or ARX-2 also partially remedies the cytokinesis defects caused by weak mutations in ECT-2 ⁴⁵ . While RhoA activates formins to generate linear actin filaments at the furrow, the anterior cortex of the embryo is enriched in ARP-2/3 nucleated branched actin ^{49– 52} . Depleting ARX-2 or Rac1/CED-10 induces cortical instability ^{45, 53} , which may allow the weak, NOP-1–dependent furrow to ingress further because the cortex is now more pliant. Alternatively, or in addition, depolymerization of this pool of branched F-actin may allow increased actin polymerization in the furrow because of a reduction in competition for actin monomer ⁵⁴ . In either case, the suppression is due to enhancement of the centralspindlin-independent pathway.

Canman and colleagues have suggested that if Rac1/CED-10 depletion generally promotes ring constriction, then Rac1/CED-10 depletion should also be able to rescue weak contractile rings resulting from temperature-sensitive mutations of the formin CYK-1 and non-muscle myosin II (NMY-2) at intermediate temperatures ⁴⁴ , neither of which is observed. However, as both furrowing pathways are active in this experiment and both strictly require formin and myosin for furrow formation, it is not entirely surprising that Rac1/CED-10 depletion does not compensate when major structural components of the contractile ring are crippled. Direct visualization of the sites of accumulation of branched actin in cyk-4 mutant embryos and assays for the effects of Rac1/CED-10 and the Arp2/3 complex on cortical stability during cytokinesis will help better understand these results.

The data discussed above point to a compelling, albeit counterintuitive, role for CYK-4 in the activation of RhoA during cytokinesis ( Figure 2B). As stated previously, in many model systems, the CYK-4 N-terminus directly interacts with the RhoGEF ECT-2 and is responsible for the relief of autoinhibition and can contribute to the recruitment of the latter to the division plane. Where analyzed, depletion of CYK-4 results in weak or no furrows. This is also the case in C. elegans CYK-4 GAP mutants ( Figure 1). Importantly, as stated in the “ cyk-4 GAP mutants have multiple defects” section above, these CYK-4 mutations result in reduced accumulation of RhoA effectors at the furrow. Disabling parallel furrow induction pathways (see “CYK-4 protein: a brief introduction”) in CYK-4 GAP mutants abolishes the residual accumulation of active RhoA and its targets.

As the above results suggest, if CYK-4 GAP mutants are truly defective in activating RhoA, then experimentally increasing active RhoA levels should restore full RhoA function and cytokinetic furrowing. Importantly, this should occur even in the absence of a redundant furrowing pathway. This prediction was tested in two ways. As mentioned previously, CYK-4 GAP mutants do not furrow in the absence of NOP-1. Remarkably, depletion of RGA-3/4, a conventional RhoGAP that switches off RhoA ^{55, 56} , results in wild-type cytokinetic furrowing in CYK-4 GAP mutants in the presence and absence of NOP-1 ³⁶ . Increasing RhoA activity using an activated allele of the RhoGEF ECT-2 has the same striking effect ³⁶ . These results together support the hypothesis that CYK-4 GAP domain participates in RhoA activation. The results described herein are compiled in poster format viewable on figshare ( 10.6084/m9.figshare.5325619).

In addition to interacting via their N-termini, an association between the CYK-4 GAP and the ECT-2 GEF domain has also been reported ³⁶ . The function of this interaction remains to be elucidated. The GAP domain may facilitate relief of ECT-2 autoinhibition and participate in a ternary complex with RhoA to fully activate the GTPase. While such a model may explain results from C. elegans embryos, CYK-4 function could be highly context-dependent and may vary with diverse cellular parameters.