Seeing is believing: tools to study the role of Rho GTPases during cytokinesis

ABSTRACT

Cytokinesis is required to cleave the daughter cells at the end of mitosis and relies on the spatiotemporal control of RhoA GTPase. Cytokinesis failure can lead to changes in cell fate or aneuploidy, which can be detrimental during development and/or can lead to cancer. However, our knowledge of the pathways that regulate RhoA during cytokinesis is limited, and the role of other Rho family GTPases is not clear. This is largely because the study of Rho GTPases presents unique challenges using traditional cell biological and biochemical methods, and they have pleiotropic functions making genetic studies difficult to interpret. The recent generation of optogenetic tools and biosensors that control and detect active Rho has overcome some of these challenges and is helping to elucidate the role of RhoA in cytokinesis. However, improvements are needed to reveal the role of other Rho GTPases in cytokinesis, and to identify the molecular mechanisms that control Rho activity. This review examines some of the outstanding questions in cytokinesis, and explores tools for the imaging and control of Rho GTPases.

KEYWORDS:

• RhoA
• biosensors
• optogenetics
• contractility
• cytokinesis

Rho GTPases in cytokinesis

As the master regulator of cytokinesis, RhoA plays a central role in the regulation of cytokinesis (Figure 1). Upon its activation, RhoA regulates contractile ring assembly at the equatorial cortex [1–4]. Active RhoA is generated by Ect2, a guanine nucleotide exchange factor (GEF), which requires the formation of an anaphase-specific complex with Cyk4/MgcRacGAP to be active (Figures 1; 5–7). Mitotic phase-GTPase activating protein (MP-GAP) uniformly dampens RhoA activity throughout the cortex, which presumably is over-ridden by Ect2 in the equatorial plane (Figures 1; 8). Active RhoA binds to and activates effectors including diaphanous-related formins to promote the nucleation of long, unbranched F-actin, and Rho-dependent kinase (ROCK) to phosphorylate myosin light chain for the assembly of bipolar nonmuscle myosin filaments (Figures 1; 9, 10). This core RhoA module is conserved among metazoans [11]. In human cells, disruption of the central spindle causes an increase in the breadth of active RhoA, while Cyk4 or Ect2 depletion causes early cytokinesis defects due to failed contractile ring formation and insufficient levels of active RhoA in the equatorial plane [6,7,12].

Figure 1. Rho GTPase signalling pathways that regulate cytokinesis. RhoA is required for cytokinesis and is considered the master regulator of contractile ring assembly. Active RhoA is generated in the equatorial plane during anaphase by the GEF Ect2, which exchanges GDP for GTP. Outside of this region, MP-GAP inactivates RhoA via stimulating GTP hydrolysis. Ect2 requires binding to the centralspindlin complex (a tetramer of Cyk4 and MKLP1] for its activity. Binding to RhoA also relieves the autoinhibition of Ect2. At the cortex, MKLP1 is inhibited by 14-3-3, which is relieved by Aurora B kinase phosphorylation. Plk1 phosphorylation of Cyk4 is also required for Ect2-binding. In the equatorial plane, active RhoA forms a contractile ring through its effectors formin and ROCK, which control actin polymerization and myosin activation, respectively. Another RhoA effector, anillin, crosslinks the ring to the overlying membrane to control ring position, and stabilizes RhoA at the membrane. Cyk4 also may downregulate Rac1 activity in the equatorial plane to decrease cortical stiffness. This could occur by preventing the activation of PAK and Arp2/3, which are required for myosin activity and the assembly of branched F-actin, respectively. The role of Cdc42 is not clear, but may regulate cortical properties similar to Rac1.

A major question in the field is how a tight zone of active RhoA is maintained for ring assembly and ingression. Both spindle-dependent and -independent mechanisms control the ring, and since there are many reviews that already describe these mechanisms, we will focus on the central spindle pathway. The central spindle is composed of antiparallel bundles of microtubules that form during anaphase between segregating chromosomes. Cyk4 is part of the centralspindlin complex with kinesin-6 MKLP1 that is required for central spindle assembly [13], and the dogma in the field is that Cyk4-binding to Ect2 spatiotemporally controls the generation of active RhoA in the equatorial plane. However, this model has been refined in more recent years. Ect2 contains several lipid-binding domains in its C-terminus that are required for its activity, and centralspindlin also associates with the membrane via the C1 domain of Cyk4 which is required for its function in regulating Ect2 [12,14–16]. Subsequent studies found that membrane localization and not central spindle recruitment of Ect2 is required for its activity [17]. Cyk4 also requires phosphorylation by Plk1 for Ect2-binding (Figures 1; 18–20). However, the Plk1-regulation of Ect2-centralspindlin could involve additional mechanisms. Plk1 activity negatively regulates microtubule-bundling which would release Ect2-centralspindlin complexes for transition to the overlying membrane where they are regulated by other factors such as Aurora B kinase (Figures 1; 21, 22). Thus, the central spindle could help direct complexes to the equatorial plane, but is not required for their activation per se.

Additional pathways also regulate the core RhoA module – either via feedback within the module, or by other GEFs and GAPs (Figure 1). For example, positive feedback within the Ect2-Cyk4 complex stimulates GEF activity, which was recently proposed to occur in part through the binding of active RhoA to a domain in Ect2 that releases it from autoinhibition [23,24]. Another major regulator of feedback is anillin, a highly conserved scaffold protein which stabilizes active RhoA for downstream signalling (Figure 1). Anillin has binding domains for actin, myosin, RhoA, septins and lipids and is required for ring positioning in many metazoan cell types [25–28]. Anillin binds cooperatively to RhoA and phospholipids, and requires active RhoA for its cortical recruitment [29]. Recent studies showed that anillin could induce the formation of PI_4,5P₂ lipid nanodomains and increase the membrane retention of active RhoA to facilitate its interaction with effectors such as Rho kinase that control actomyosin assembly [25,29]. Although less clear, negative regulators of RhoA also influence ring ingression. A recent study showed that in C. elegans, GCK-1 (germinal centre kinase-1/mammalian germinal centre kinase III subfamily) and its partner CCM-3 (cerebral cavernous malformations-3) are recruited by ANI-1 (anillin) to the equatorial cortex where they feed back to negatively regulate RHO-1 and dampen contractility [30]. Their model is that GCK-1 and CCM-3 downregulates RHO-1 by recruiting RGA-3 (MP-GAP) to the cortex acting as a brake to the contractile ring. Examples of additional regulators of RhoA include other GEFs, such as GEF-H1, MyoGEF and LARG, and GAPs such as p190RhoGAP, although their relative requirements are not clear [31–35].

The role of Rac in cytokinesis is controversial. Loss-of-function studies do not support a significant role for Rac in cytokinesis [24,36]. However, depending on the cell type and context, Rac could be important for furrow positioning, cell adhesion and recovery of cortical polarity for cell spreading. Studies in the one-cell C. elegans embryo have led to conflicting models for the role of Rac in cytokinesis based on the need to specifically downregulate Rac in the equatorial plane vs. a more global regulation of Rac. One model proposes that CYK-4 has GAP activity towards Rac in the equatorial plane, which facilitates ingression by decreasing branched F-actin in the furrow [37]. Rac leads to the activation of Arp2/3, which regulates the formation of branched, F-actin to confer different cortical properties compared to unbranched F-actin. The depletion of Rac/ced-10, or its effector Arp2/3/arx-2, partially suppresses cytokinesis phenotypes caused by an inactivating mutation (E448K) in the GAP domain of cyk-4 [24,37,38]. However, another interpretation of this data is that the Rac-mediated suppression of cyk-4 phenotypes occurs by globally softening the cortex to permit ingression when the ring is weakly formed, as CYK-4 is required for Ect2 to activate RhoA. In embryos with defective nop-1, a C. elegans-specific regulator of contractile ring assembly that functions redundantly with cyk-4, Rac/ced-10 depletion does not suppress the cytokinesis phenotypes caused by the cyk-4 GAP mutant, which would be predicted if Rac1 were the direct target [24,39,40].

In mammalian cells, active Rac is enriched at the poles during mitotic exit, while Rac and its effectors appear to be inactivated at the cell equator [41,42]. Mutating the GAP domain of Cyk4 causes cytokinesis failure and increases the formation of focal adhesion complexes. Further, the overexpression of constitutive active Rac hinders cytokinesis and RhoA recruitment, suggesting that Rac must be inactivated at the cell equator to inhibit cell adhesion pathways, which may be detrimental for cytokinesis [41]. Similar to C. elegans, depleting Rac or its effectors PAK1 and ARHGEF7 can rescue the defects caused by the GAP mutant [41]. However, the same alternative interpretation of data could apply as described for C. elegans where active Rac and its effectors are required to dampen cortical contractility outside the furrow. This model is also supported by studies in Dictyostelium, where furrow ingression occurs more rapidly when cortical tension is reduced upon removal of RacE/dynacortin, suggesting that the balance of branched and linear F-actin controls the rate of ingression [43]. Collectively, studies in a diverse range of cell types support a model where linear F-actin and myosin contractility is enriched in the equatorial plane and generates tension that is balanced by different cortical properties at the poles [38,43,44–47].

The evidence supporting a role for Cdc42 in cytokinesis is even less clear compared to Rac and has been under-studied compared to the other GTPases. Cdc42 depletion does not appear to cause cytokinesis phenotypes in several model systems [36,37,48,49]. However, Cdc42 constitutive activation blocks cytokinesis in Xenopus embryos [50], and in HeLa cells [51]. It also disrupts cellularization in Drosophila embryos, a process that shares some similarities with cytokinesis [48]. Cdc42 may have stronger roles in cells with cortical polarity, where it could influence ring positioning and ingression kinetics [52,53]. Also, there could be some redundancy with Rac since both GTPases can recruit effectors such as Wave or Wasp that control Arp2/3, and PAK, which can regulate myosin contractility and adhesion [54]. New techniques and methodologies that permit better spatiotemporal resolution are needed to revisit the role of Cdc42 in cytokinesis in specialized cell types.

Biosensors to study Rho GTPases

The tools used to visualize active Rho include biochemical techniques, fixed cell immunofluorescence and biosensors. Pulldowns using the purified recombinant Rho-GTP binding domain (RBD; also referred to as GBD for GTPase binding domain) of Rho effectors with high affinity for the GTP-bound state enable the specific precipitation of GTP-bound Rho [55,56], which can be quantified compared to controls by densitometry on western blots [56–58]. Until Yonemura published their ground-breaking paper describing the use of TCA fixation to visualize active RhoA, no other method had been able to demonstrate its localization during cytokinesis [58]. More than 90% of Rho in cells is tightly bound to GDI, making most of this pool inaccessible for extraction [56,57]. Crosslinking fixatives lock this pool into place, while precipitation-based methods like methanol cause their extraction when permeabilized, and neither method revealed differential staining at the equatorial cortex. While it is not clear why TCA fixation works, one hypothesis is that it precipitates RhoA rapidly enough to protect the membrane-bound pool from loss during permeabilization [58,59]. The ability of TCA fixation to recover active RhoA may also be anillin-dependent [59]. Since active RhoA is at the membrane, the interpretation is that this preserved pool reflects active RhoA. Unfortunately, biochemical assays and imaging relying on cell fixation are ill-suited for studying dynamic events due to poor spatial (western blots) or temporal (fixed cell imaging) resolution; these challenges can be remedied by using biosensor probes to report Rho activity in vivo.

Current methodologies to visualize active Rho GTPases in cells rely on probes that bind to or reflect the active conformation, as summarized in Table 1 (due to space limitations, this is not an exhaustive list). Many of these biosensors rely on Förster resonance energy transfer (FRET), the energy transfer between two fluorophores via dipole-dipole coupling [60]. The efficiency of this energy transfer is inversely proportional to the sixth power of the distance, making FRET a sensitive method to detect protein-protein interactions in vivo. The Rho FRET probes typically consist of a Rho GTPase, a Rho-binding domain (RBD) from an effector, and two fluorescent proteins with a spectral emission-absorption overlap amenable for FRET [61]. The probes are introduced exogenously where changes in their localization or energy transfer are interpreted to report for changes in Rho activity. One of the first biosensors for studying Rho GTPases was the FLAIR (fluorescence activation indicator for Rho proteins) biosensor used to image changes in Rac activity in Swiss 3T3 fibroblasts [62]. The fusion of GFP to Rac provides a donor for Alexa-546 dye coupled to the CRIB domain from PAK (p21-activated kinase), such that the binding of GFP to the Alexa dye causes FRET [62]. With this design, the FRET signal reports for the net activation of Rac by its upstream regulators. To circumvent the challenge of introducing a dye to cells, a new generation of probes was developed where the fluorescent proteins were genetically encoded in the FRET sensor. The Raichu probes (Ras and interacting protein chimeric unit) were designed with the RBD and/or Rho sandwiched between donor and acceptor pairs [42,63]. To report for changes in GEF or GAP activity, Raichu-RhoA was designed with RhoA and the RBD from PKN sandwiched between CFP and YFP [42]. Similar probes were developed for Rac and Cdc42. The binding of active Rho to the RBD enhances FRET and reports for an increase in GEF activity. Another probe, Raichu-RBD, contains the RBD from Rhotekin and lacks RhoA. In this case, binding of endogenous active RhoA to the RBD causes a reduction in the FRET signal [42].

Table 1. List of biosensors and tools designed for studying Rho GTPases

Biosensors	Readout	Sensing Elements	Reference
RhoA/B/C Biosensor
GFP-RhoA	Intensity	RhoA	101
YFP-RBD-RhoA-CFP (Raichu-RhoA]	FRET	RBD [PKN), RhoA	42
eGFP-rGBD	Intensity	RBD (Rhotekin] [M. musculus)	68
YFP-RhoA	Intensity	RHO-1 [C. elegans]	6
RBD-CFP-YFP-RhoA (RhoA FLARE/RhoA-1 G]	FRET	RhoA, RBD [Rhotekin)	64
AHPH-GFP	Intensity	RBD [C-anillin]	28
CyPet-RBD, YPet-RhoA (bimolecular RhoA Biosensor]	FRET	RBD [Rhotekin), RhoA	102
AHPH-GFP	Intensity	RBD [C-ANI-1; C. elegans]	40
mTFP1-RBD-Venus-RhoA (RhoA-2 G]	FRET	RBD [Rhotekin), RhoA	103
RBD-mVenus-Cerulean3-RhoA (RhoA DORA]	FRET	RhoA, RBD [PKN1)	65
AniRBD-GFP (Rho1 GTP sensor]	Intensity	RBD [C-Anillin; Drosophila)	76
Cerulean-Venus-RhoA [GDI-RhoA FLARE]	FRET	RhoA	104
mCyRFP1-RhoA, mMaroon1-RBD-mMaroon1 (RhoA-CyRM]	FLIM-FRET	RhoA, RBD [Rhotekin)	105
mRFP-RhoA-RBD-eGFP (Raichu-RhoA-FRET biosensor]	FLIM-FRET	RhoA, RBD [PKN)	106
GFP-LET-502 (RHO1-biosensor]	Intensity	ROCK [LET-502; C. elegans)	30
dimericTomato-2xrGBD	Intensity	RBD [Rhotekin]	72
GFP-RhoB	Intensity	RhoB	101
RBD-mVenus-Cerulean3-RhoB (RhoB FRET Sensor]	FRET	RBD [PKN1), RhoB	66
RBD-mCerulean-mVenus-RhoC (RhoC FLARE]	FRET	RBD [ROCK1), RhoC	107
RBD-mVenus-Cerulean3-RhoC (RhoC FRET Sensor]	FRET	RBD [PKN1), RhoC	66
IT-Rho	Intensity	Rho [X. laevis]	80
Cy3-Rho (GDI-bound]	Intensity	Rho [X. laevis)	80
Cdc42 Biosensor
GFP-Cdc42	Intensity	Cdc42	101
YFP-RBD-Cdc42-CFP (Raichu-Rac]	FRET	RBD [PAK), Cdc42	63
eGFP-wGBD (or mRFP-wGBD]	Intensity	RBD [N-WASP)	68
GFP-GBDwsp-1	FLIM-FRET	RBD [WSP-1; C. elegans]	79
mCerulean-PBD-PBD-cp229Venus-Cdc42	FRET	PBD [PAK], Cdc42	108
mCyRFP1-Cdc42, mMaroon1-RBD-mMaroon1 (Cdc42-CyRM]	FLIM-FRET	Cdc42, RBD [Pak3)	105
Cerulean-Venus-Cdc42 [GDI Cdc42 FLARE]	FRET	Cdc42	104,
ddFP-RBD, ddRFP-Cdc42 (bimolecular R-Cdc42]	Intensity	RBD [Pak1),Cdc42	109
IT-Cdc42	Intensity	Cdc42 [X. laevis]	80
Cy3-Cdc42 (GDI-bound]	Intensity	Cdc42 [X. laevis)	80
Rac1/2/3 Biosensor
Rac-GFP, PBD-Alexa546 (FLAIR]	FRET	Rac, PBD [PAK)	62
GFP-Rac1	Intensity	Rac1	101
YFP-RBD-Rac1-CFP (Raichu-Rac1]	FRET	RBD [PAK), Rac1	63
Ypet-Rac1, PDB-CyPet (Rac1-FLARE]	FRET	Rac1, PBD [PAK)	102
mTFP1-PBD-Venus (Rac1-2 G]	FRET	PBD [PAK), Rac1	110
mCerulean-cpVenus-Rac1 [GDI Rac1 FLARE]	FRET	Rac1	104
miRFP720-PBD-PBD-miRFP670-Rac1 (NIR Rac1 Biosensor]	FRET	PBD [PAK), Rac1	100
Ypet-PBD, Turquoise-Rac1 (Rac1 FLARE DC1g]	FRET	PBD [PAK), Rac1	111
ddFP-PBD, ddRFP-Rac1 (R-Rac1]	Intensity	PBD [PAK), Rac1	109
mCerulean-PBD-PBD-mVenus-Rac1 biosensor	FRET	PBD [PAK], Rac1	112
GFP-Rac2	Intensity	Rac2	101
mCerulean-PBD-PBD-mcp229Ven-Rac2 biosensor	FRET	PBD [PAK], Rac2	113
mCerulean-PBD-PBD-mcp229Ven-Rac3 biosensor	FRET	PBD [PAK], Rac3	114

Abbreviation: AHPH, anillin homology domain (AHD] and pleckstrin homology (PH) domain of Anillin; ddFP, dimerization-dependent fluorescent protein; DORA, Dimerization Optimized Reporter for Activation; FLAIR, fluorescence activation indicator for Rho proteins; FLARE, fluorescence activation indicator of Rho proteins; IT, internally-tagged GFP; mTFP, monomeric Teal fluorescent protein 1; NIR, near-infrared.PAK, p21-activated kinase; PBD, p21 binding domains [PBD) derived from PAK1; PKN, Protein Kinase N; Raichu, Ras and interacting chimeric unit; RBD, RhoA-binding domain; RhoA-1 G, first-generation RhoA biosensor; RhoA-2 G, RhoA biosensor second generation;

Disclosure statement: This is by no means an exhaustive list of the probes designed to detect Rho-family GTPases. For simplicity, the biosensors description includes only the fluorescent proteins and sensing elements.

Despite their successful use in cell motility and wound healing, the early FRET probes were less reliable during cytokinesis. Another biosensor was designed with the fluorophore pair sandwiched between the RBD from Rhotekin and RhoA, with RhoA at the C-terminus to permit post-translational modification (Figure 2(a)) [64]. Using this biosensor, active RhoA was detected at the cleavage furrow during cytokinesis of HeLa cells [31]. Using this design, a similar RhoA biosensor was developed and named Dimerization Optimized Reporter for Activation (DORA) which reports for changes in active RhoA in interphase cells and may also work in dividing cells [65]. Instead of using the RBD from Rhotekin, they sandwiched the fluorophore pair (YFP and CFP derivatives) between the RBD of Protein Kinase N1 [PKN1) and full-length RhoA. RhoB and RhoC biosensors were similarly generated and used to compare the changes in active RhoA, RhoB or RhoC localization in endothelial cells under different activating conditions [66] While the Hordijk group observed RhoA/B/C activation in response to nocodazole treatment (depolymerizes microtubules], which presumably releases active GEF-H1, they did not examine the localization of these GTPases during cytokinesis in non-treated cells [66].

Figure 2. RhoA biosensors. RhoA localizes to the plasma membrane through its lipid-modified tail. RhoA is activated by guanine nucleotide exchange factors (GEFs) and inactivated by GTPase activating proteins (GAPs), while GDP dissociation inhibitors (GDIs) sequester inactive RhoA (RhoA-GDP) in the cytosol. Active RhoA [RhoA-GTP) binds to effectors to cause a change in cytoskeletal dynamics. Different probes have been developed to selectively detect active RhoA. a) A FRET biosensor generated by 64,includes the RhoA-binding domain [RBD] of Rhotekin at the N-terminus, with RhoA at the C-terminus. FRET is generated when RhoA is active and binds to the RBD. b) Another probe designed by 67,contains the RBD from Rhotekin fused to GFP. This probe localizes to active RhoA during cytokinesis in Xenopus oocytes and echinoderm embryos. c] Another probe designed by 6,fused C. elegans RHO-1 with YFP. This probe localized as expected for active RhoA during cytokinesis in cultured human cells. RHO-1 may have a slower rate of GTP hydrolysis compared to human RhoA, causing it to bind to conserved effectors and localize. d) Another probe designed by 28,contains the C-terminus of anillin [anillin homology domain (AHD) with RBD and C2 domains, and a PH domain] fused to GFP. While the RBD binds to active RhoA, the PH domain binds to the membrane and mediates cooperative RhoA-binding [29]. This probe has been used to successfully visualize the localization of active RhoA during cytokinesis in cells from different organisms.

There has been limited success with the use of non-FRET probes to detect active RhoA during mitosis. In Xenopus and echinoderm embryos, a fluorophore fused to the GBD (RBD) of Rhotekin beautifully reports for active RhoA during cytokinesis (Figures 2(b); 67, 68). An equivalent probe reports a signal during mitosis in mast cells, but does not work well in HeLa cells, Drosophila or C. elegans embryos [42,59,69–71]. In 2021, the Rhotekin-based GBD (RBD) biosensor was improved by increasing the number of fluorescent proteins, using dimeric vs. monomeric forms of fluorescent protein, and by increasing the number of GBD’s [72]. Based on membrane localization in response to RhoA activation in HeLa cells, the Goedhart group found that the dimericTomato-2xrGBD sensor was most improved compared to prior versions. The probe shows beautiful localization to the furrow in HeLa cells and could be ideal for use in studies of mechanisms regulating cytokinesis [72]. The caveat to using any of the GBD/RBD-based probes is that they could compete with endogenous complexes for binding to active Rho.

Tagging native proteins, entire effectors or more domains compared to only the RBD could provide context that is required to more accurately report for Rho activity. Interestingly, expressing a fluorophore fused to C. elegans RHO-1 in human cells localizes similar to fixed cells during cytokinesis (see below), even though human RhoA fails to localize (Figure 2(c)) [6]. This probe could work because RHO-1 remains in the GTP-bound state longer vs. human RhoA and could compete with endogenous RhoA for localization and/or effector binding [unpublished observations; 6]. In C. elegans embryos, GFP:RHO-1 localizes cortically and to the contractile ring, albeit weakly [73]. More recently, the C-terminus of anillin, which includes an RBD, C2 and PH domain (the RBD and C2 form the anillin homology domain; AHD), has been used as a reporter in C. elegans, Drosophila, and in cultured human cells (Figure 2(d)) [28,40,74–77]. This likely works well because it contains multiple binding domains that better reflect RhoA’s localization to specific subsets of phospholipids. The RBD and C2 domains bind cooperatively to RhoA and phospholipids [29], and recent studies showed that anillin increases the residence time of active RhoA via promoting the clustering of PI_4,5P₂ lipids, which are favoured by RhoA [25,27,78]. In addition, the RBD of anillin may bind weakly to RhoA compared to other effectors to facilitate the hand-off of RhoA vs. competing with their binding [25,78]. The caveat with using this probe is that it also binds to other cortical proteins such as septins and Ect2, which also could influence its localization [27,75]. Further, depending on the study, it could be ideal to use other effectors to monitor the localization of active RhoA. For example, CRISPR was used to endogenously tag LET-502 in C. elegans (human ROCK) with GFP to monitor changes in active RhoA during cytokinesis [30].

Other non-FRET probes to study RhoA and Cdc42 in vivo have also been described. For Cdc42, a fluorescent gene fused with the Cdc42-binding domain from WASP reports for active Cdc42 in Xenopus oocytes, the one-cell C. elegans embryo and in metaphase mast cells [68,71,79]. A different approach was used to design probes for RhoA and Cdc42 by (i) inserting GFP directly into a solvent-based loop within Rho (IT-Rho) and co-expressing it in Xenopus oocytes with GDI and at 36% above endogenous Rho expression, and, (ii) generating recombinant, prenylated Rho that was ligated to Cy3 at the N-terminus, then bound to GDI and injected into oocytes ~41% or 53% above endogenous levels (for RhoA and Cdc42, respectively) [80]. The signals associated with these two different types of probes overlapped with previously described reporters for active RhoA (mRFP-2xrGBD) and Cdc42 (mRFP-wGBD) during wound healing [68,81]. These new probes localized to the furrow during cytokinesis and could be useful for studying mechanisms where the endogenous machinery is not compromised [80].

Optogenetic tools to manipulate Rho GTPase activity

It is also highly desirable to control the activity of Rho GTPases with high spatiotemporal precision to study their function in specific biological processes. Due to the pleiotropic requirements of GTPases for multiple events, loss or gain-of-function studies have been challenging to interpret as these perturbations are chronic and allow indirect effects to accumulate. The recent development of optogenetic tools to control Rho activity in mammalian cells have permitted functional studies of Rho in cytokinesis (Table 2). In 2016, Wagner and Glotzer used tunable light-controlled interacting protein tags (TULIPs) to photomanipulate a RhoGEF and induce RhoA activity throughout the cell cycle. To accomplish this, they relied on the use of photosensitive protein domains that interact upon exposure to blue light. Their design involved the fusion of a tandem PDZ tag to the DH domain of the GEF LARG (photorecruitable GEF), while GFP-tagged LOVpep localizes to the membrane by fusion to the transmembrane protein Stargazin [82]. Upon illumination with blue light, LOVpep undergoes a conformational change, enabling the PDZ domains to bind reversibly to LOVpep and recruit GEF to the membrane where it can activate RhoA (Figures 3(a); 82, 83). They showed that exogenously activated RhoA was sufficient to initiate furrowing in any region of the cell cortex, and in any cell cycle phase [82]. In the same year, the Petronczki group used the Cry2-CIB1 dimerization system to optogenetically activate RhoA by locally targeting Ect2 to the plasma membrane during cytokinesis [17]. To do this they fused the N-terminal part of CIB1 to GFP and a membrane-targeting motif (CIBN-EGFP-CAAX), while Cry2 was fused to mCherry and Ect2 lacking the C-terminal PH and PBC domains required for membrane localization (Cry2-mCh-ECT2) [17]. Upon illumination with blue light, Cry2 binds to CIB1, bringing Ect2 to the membrane. Using this system, they found that membrane-localized Ect2 is necessary to support furrow ingression, which also requires Cyk4-binding [17]. Another optogenetic tool relies on the oligomerization of Cry2 to activate RhoA [84]. A Cry2-mCherry-RhoA fusion clusters upon illumination with blue light to generate photoactivatable RhoA in mammalian cells, although it is not clear why clustering is sufficient to activate RhoA. Within seconds of exposure to light, the initially diffuse Cry2-mCherry-RhoA fusion translocates to the plasma membrane and is reversible (Figure 3(b)). To date, only one study reports the use of this tool during mid-late cytokinesis [85], and it is not clear if it will be useful during earlier stages of ring assembly and ingression.

Table 2. List of optogenetic tools designed for studying Rho GTPases

Optogenetic tools	Wavelength	Elements	Reference
Optogenetic Clustering and activation of RhoA	488 nm	Cry2-RhoA	84
Cry2-CIB1 dimerization	488 nm	Cry2-mCh-ECT2, CIBN-EGFP-CAAX	17
TULIPs -Light-mediated control of RhoA activity	405 nm	photorecruitable GEF [LARG]	82
LOVTRAP – reversible, light-induced protein dissociation	450–490 nm	LOV2, Zdk-RhoA	86
Optogenetic Clustering and activation of RhoA	488 nm	Cry2PHR-mCH-RhoA	85
Mitotrap	488 nm	Cry2-VHH, Tom20-CIB-stop	85

Abbreviation: TULIPs, Tunable, light-controlled interacting protein tags; LOVTRAP, LOV2 Trap and Release of Protein.

Disclosure statement: This is by no means an exhaustive list of the optogenetic tools designed to detect Rho-family GTPases.

Figure 3. Optogenetic tools to activate RhoA. Two different approaches are shown that have been designed to control RhoA activation. a) Illumination generates conformational changes that permit protein-protein interactions between the PDZ domain and LOVpep. Membrane-anchored LOVpep binds to PDZ fused with the GEF domain from LARG, bringing the GEF to the plasma membrane where it can activate RhoA. Fluorophores are also fused to the different components to visualize their localization (GFP with LOVpep and mCherry with PDZ-GEF). b) Illumination causes Cry2 clustering. Light-induced clustering activates RhoA when fused to Cry2 via an unknown mechanism. Adding a fluorophore, such as mCherry also permits visualization of clustered active RhoA.

Other optogenetic tools have been developed but have had limited use for studies of Rho during cytokinesis. A light-inducible dissociation system such as LOVTRAP could be used to tightly and reversibly control the activity of Rho GTPases by controlling the localization of upstream regulators [86]. One of the proteins from the light-dissociation pair Zdk/LOV2 can be tagged to Rho or one of its regulators, while the other is targeted to a specific location. Their interaction can be reversibly controlled by light, such that Rho or its regulator are sequestered to an organelle. Several analogous systems have been developed using different light-responsive proteins. For example, a Mitotrap assay was used to demonstrate a role for CLIC4 in cytokinesis [85]. The Prekeris group co-transfected HeLa cells expressing endogenous GFP-CLIC4 with Cry2-GFP-VHH (GFP-VHH encodes a GFP nanobody) and Tom20-CIB-stop (Tom20 binds to the outer membrane of mitochondria). When pulsed with blue light, Cry2 binds to CIB bringing GFP–CLIC4 to the mitochondria. Using this approach, they found that the removal of CLIC4 from the plasma membrane caused an increase in cortical blebbing and furrow regression demonstrating its role in cytokinesis [85].

Harnessing Rho GTPase tools for cytokinesis studies

Biosensors are useful tools to study the dynamic changes and spatial distribution of active Rho, but as described previously, have limitations for use during cytokinesis. Improvements in the quantum yield and stability of fluorophores, imaging infrastructure and computational tools could improve the use of existing biosensors for cytokinesis studies. Different fluorescent proteins have unique properties that affect their functionality as a visual tool in various experimental settings. FPbase (https://www.fpbase.org) is a searchable database for fluorophores, where researchers can calculate the Förster distance (R₀) for a donor/acceptor pair using the FRET calculator (https://www.fpbase.org/fret/ [87]). Many researchers use fluorescence lifetime imaging (FLIM) to measure FRET independent of protein concentration, which can be variable when using exogenous probes. FLIM is a quantitative, intensity-independent fluorescence imaging technique that produces image data based on the fluorophore’s fluorescence lifetime. Thus, by measuring the differences in the fluorescence decay rate from a sample, FLIM can distinguish probes with similar fluorescence spectra and is therefore a widely used approach for multiparameter imaging [88,89].

The generation of entirely new approaches to detect conformational changes in Rho with high signal-to-noise ratios may be required. It may also be necessary to control expression or use tools that are less likely to interfere with normal Rho function in cells. For example, endogenously integrating fluorophores, or controlling their expression using inducible promoters. Current biosensors detect active Rho upon binding to the RBD from specific effectors, but it is clear that the affinity for Rho may differ depending on the effector. In addition, Rho may require cooperative binding with other components to reach its optimal affinity for these domains. Even more of a concern is that the probe could outcompete endogenous interactions that are required for normal function. Using this knowledge could guide the design of probes better suited for detecting active Rho during cytokinesis, which may differ for other biological processes.

Using a different approach altogether to sense active Rho could lead to more significant advancements in the development of biosensors. Nanobodies or intrabodies detect specific conformations of their target proteins [90]. When coupled with fluorophores and transfected in cells as an expressing transgene, they can be used to visualize the localization of their target protein in a specific state (e.g. Figure 4(a)). For example, the nanobody RH57 was recently identified by a synthetic phage display library that selectively bound to the active form of RhoA/B/C when expressed as an intracellular antibody [intrabody). This nanobody was functionalized  for use as a bioluminescence resonance energy transfer (BRET] activity sensor by fusing RLuc8 (luciferase) to RhoA, which generates light from luciferin to excite GFP fused to RH57 [90]. Like FRET, BRET can be measured ratiometrically, but offers lower phototoxicity, less bleaching, minimal autofluorescence, and an improved signal-to-noise ratio compared to FRET [91]. Another study identified an intrabody that binds selectively to active RhoB, which was further functionalized by adding the F-box domain of E3 ubiquitin ligase (F-lb) to selectively degrade active RhoB via the ubiquitin-proteasome pathway [92]. These tools, or modifications of these tools could be used to study Rho GTPases during cytokinesis.

Figure 4. Multiplexing tools to study RhoA. a) A nanobody that selectively detects active RhoA can be expressed as a fusion with a fluorophore tag [e.g. GFP). Nanobodies can bind to clefts and cavities that are otherwise inaccessible to conventional antibodies or effector proteins, and thus are less likely to compete for binding with effector proteins. b) The nanobody-based biosensor for active RhoA could be coupled with optogenetics to control RhoA activity. In this example, the GEF domain is fused to mVenus-labelled Zdk1, which binds to the LOV2 domain. Illumination with 457-nm light causes their dissociation to release the GEF for RhoA activation, reported by the change in localization of the RhoA biosensor. Zdk1 or LOV2 also can be fused to other domains that further control subcellular localization. c) FRET probes that detect active different RhoGTPases also can be multiplexed with optogenetics to control RhoGTPase signalling. This example, based on 100, includes a near-infrared FRET biosensor for active Rac, where Rac is fused to the C-terminus. The probe contains a far-red FRET donor-acceptor pair, as well as PBD1 and PBD2 domains that interact in the absence of active Rac. Active Rac binds to the PBD1 domain causing dissociation of the PBD2 domain to permit FRET. The optogenetic tool is similar to B] where a GEF for Rac can be released upon illumination with 457-nm light to activate Rac1. In addition, RhoA biosensors can be combined with this system with tags such as GFP that are compatible with the NIR biosensor and optogenetic tool (not shown).

Split-fluorophores could also offer improvements for studies of Rho GTPases in cytokinesis. In bimolecular fluorescence complementation (BiFC), a fluorescent protein is split into non-fluorescent fragments, and fluorescence is restored when these fragments interact to reconstitute the fluorophore [93]. Proteins of interest can be fused to the fluorophore fragments and their interaction is measured based on the reconstituted fluorescence. Using this approach could be advantageous to visualize active Rho during cytokinesis since such a small pool of Rho is active, and BiFC has a high signal-to-noise ratio. More recently, a tripartite split-GFP system was developed that splits GFP into three fragments [94,95]. Similar to the BiFC approach, the beta strands of GFP are separated into multiple fragments. These include two small fragments, GFP10 and GFP11, which can reconstitute GFP when they come together with a larger fragment, GFP1-9. This method may offer an advantage as the smaller fragments are less likely to cause toxicity [96] and can be used to label endogenous proteins [97]. However, the irreversibility of split-fluorophore complementation prevents its use in capturing dynamic events during cytokinesis. Currently, few reversible BiFC systems have been described, and may suffer from loss of quantum yield or brightness [98,99]. The development of bright, reversible BiFC systems may elevate BiFC as a viable alternative to FRET.

Future studies should consider multiplexing optogenetic tools that control Rho GTPases with a biosensor(s) to coordinate changes in activity with the visualization of a response with high spatiotemporal resolution. For example, a fluorophore labelled-nanobody that specifically detects active RhoA could be coupled with an optogenetic system for controlling RhoA activity (Figure 4(b)). Recently, a multiplexed imaging approach was designed by combining an optogenetic tool for Rac with two biosensors for active Rac and RhoA [100]. The Hodgson group engineered the first near-infrared (NIR) biosensor that detects active Rac1 and is compatible with CFP/YFP biosensors. Using these tools, they optogenetically controlled Rac activity, while imaging the NIR Rac1 and CFP/YFP RhoA biosensors to visualize simultaneous changes in Rac and RhoA activity within the same cell. An example of multiplexing the NIR sensor with optogenetics to control Rac activity is shown in Figure 4(c). Their multiplexed system demonstrated that Rac1-RhoA antagonism occurs during protrusion-retraction cycles in migrating fibroblasts [100]. Such a tool could also be useful for cytokinesis, especially where knowledge of multiple Rho GTPases is lacking. Combining these with an ‘n-partite’ approach could also be useful for studying cytokinesis with minimal cellular perturbation and toxicity.

Concluding remarks

Given their pleiotropic roles, it has been challenging to define the function of Rho GTPases in cytokinesis. While numerous studies have shown that RhoA is required for cytokinesis, the pathways that regulate its activity and localization are still not well-understood. In particular, the crosstalk and feedback that influence the function of RhoA GEFs and GAPs, as well as other GTPases remain to be clarified. In recent years, significant progress has relied on imaging and optogenetic tools to study Rho GTPases in different biological processes. Nevertheless, such efforts in cytokinesis are lacking, and our understanding of the regulation of different Rho GTPases in cytokinesis is still far from complete. Recent technological advancements should offer the improved spatiotemporal resolution needed to capture highly dynamic events, and combined with multiplexing existing or new biosensors (e.g., that incorporate nanobodies) and/or with optogenetic tools will significantly enhance the study of Rho GTPases for their role in cytokinesis.

Acknowledgments

We thank Dr. Michael Glotzer and reviewers for helpful comments and feedback. We acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (NSERC Discovery Grant 04161-2017), and Concordia programs for funding including the Horizon Postdoctoral Fellowship and Concordia University Research Chair program.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Natural Sciences and Engineering Research Council of Canada [RGPIN-2017-04161].