Wnt and Rho GTPase signaling in osteoarthritis development and intervention: implications for diagnosis and therapy

Wnt and Rho GTPase signaling play critical roles in governing numerous aspects of cell physiology, and have been shown to be involved in endochondral ossification and osteoarthritis (OA) development. In this review, current studies of canonical Wnt signaling in OA development, together with the differential roles of Rho GTPases in chondrocyte maturation and OA pathology are critically summarized. Based on the current scientific literature together with our preliminary results, the strategy of targeting Wnt and Rho GTPase for OA prognosis and therapy is suggested, which is instructive for clinical treatment of the disease.


Introduction
Th e disability burden and prevalence of osteoarthritis (OA) in both developed and developing countries is increasing due to an aging population. OA is a degenerative joint disease that is characterized by cartilage degradation and osteophyte formation. It involves multiple components of the joint, including the synovial joint lining, peri-articular bone and adjacent supporting connective tissue elements [1]. Current OA treatment modalities mainly function as intermittent symptom relief without long-term improvement in disease prognosis due to our current limited understanding of OA pathophysiology. Better understanding of the underlying mechanisms of OA initiation and progression might therefore facilitate identifi cation of appropriate therapeutic targets for OA treatment [2].
Th e mechanism of OA is currently not well defi ned, as multiple factors can in more than one way lead to articular cartilage destruction and loss of joint function. Recently, increasing numbers of studies have implicated chondrocyte terminal diff erentiation (hypertrophy-like changes) in the pathogenesis of OA. Th is is similar to the chondrocyte diff erentiation process during endochondral ossifi cation (EO). Th e close resemblance between terminal diff erentiation in OA cartilage and EO suggests that new OA therapeutic targets can potentially be identifi ed from EO biology. Normal articular chondrocytes located at the ends of long bones do not develop into the hypertrophic state, thus avoiding terminal diff erentiation. However, OA chondrocytes lose their stable phenotype and undergo hypertrophy, which is characterized by cell enlargement as well as expression of various chondrocyte maturation and osteogenesis markers such as COLX [3], matrix metalloproteinase (MMP)13 (also known as collagenase 3) [3][4][5], a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS)-5 [6][7][8], osteopontin, osteocalcin, Indian Hedgehog [9], Runx2 [10], vascular endothelial growth factor (VEGF) [11], and trans glutaminase-2 (TG-2) [12].
Th e developmental biology of EO is of key importance in understanding the process of OA, and there is much scientifi c evidence indicating that signaling pathways modulating joint formation and homeostasis are of central importance in the pathogenesis of OA. Th e Wnt signaling pathway is well established to be a key regulator in EO [13,14], a process through which bone and articular cartilage are formed. At the same time, most studies support the notion that activation of Wnt/β-catenin signaling is associated with articular chondrocyte matrix catabolism and stable phenotype loss [15]. Recent years have also seen a number of studies indicating that Rho GTPases play central roles in both chondrocyte diff erentiation and articular chondrocyte physiology, which will be discussed below. complex of casein kinase, axin, the adenomatous polyposis coli tumor suppressor protein (APC) and glycogen synthase kinase 3β (GSK3β) [16]. However, when Wnt ligands bind to cell membrane receptors, signaling through the frizzled receptors inhibits this degradation process, thereby increasing the levels of free cytoplasmic β-catenin. Accumulation of cytoplasmic β-catenin results in its translocation to the nucleus, where it binds to transcription factors such as lymphoid enhancing factor (LEF)/T cell factor (TCF) to generate a transcriptionally active complex that targets genes such as those encoding MYC, cyclin D1, MMP3 and CD44 [17]. In addition, there are some natural extracellular inhibitory factors that regulate canonical wnt signaling, including members of the secreted frizzled receptor protein (sFRP) family, Dickkopf (Dkk) proteins [18], Wnt inhibitory factor [19], cerberus [20] and sclerostin [21] (Figure 1).
Th e Rho family of GTPases includes 20 members, which are 'Ras-like' proteins. Amongst these, Cdc42, Rac1, and RhoA have been intensively studied. Guanine nucleotide exchange factors (GEFs), GTPase-activating proteins (GAPs) and guanine nucleotide dissociation inhibitors (GDIs) are all regulators of the switch between the active and inactive forms of Rho-GTP. Rho GTPases have also been referred to as 'molecular switches' for transducing signals from the chondrocyte extracellular matrix to aff ect cytoskeletal actin dynamics and cellular morphology, which in turn regulate cell proliferation, apoptosis and gene expression [22].
Meanwhile, a new study indicates that Rho GTPases play a role in nuclear transportation of cytoplasmic βcatenin. Constitutive activation of Rac1 in colon cancer cells signifi cantly enhances TOPFlash promoter activity and nuclear β-catenin accumulation. Th is eff ect is inhibited by dominant-negative Rac1 [23]. Mutation of RacGap50C, a negative regulator of Rac1, in Drosophila embryos stimulated canonical Wnt signaling [24]. Similarly, Rac1-specifi c activator Tiam1 was demonstrated to transcriptionally activate β-catenin/TCF complexes in response to Wnt3a [25]. In another study, Wu and colleagues [26] reported that Rac1 acted cooperatively with JNK2 activation during β-catenin phosphorylation and nuclear localization. Th is was further supported by phenotype similarity between Rac1 and β-catenin ablation in mouse limb bud ectoderm. Although neither stabilization nor nuclear localization of β-catenin requires RhoA activation, Wnt3a induction of osteogenic diff erentiation of stem cells requires both RhoA and βcatenin activation [27] (Figure 1).
By contrast, much less attention has been paid to noncanonical Wnt signaling, which is characterized as being β-catenin/TCF independent. One example of noncanonical Wnt signaling is the planar cell polarity pathway, which promotes cell organization in a particular orientation [28,29], through the action of Rho GTPases on assembly of the actin cytoskeleton [30,31].

Roles of Wnt and Rho GTPases in regulating chondrocyte hypertrophy and maturation
Canonical Wnt signaling is known to induce chondrocyte hypertrophy and fi nal maturation. During skeletal develop ment and growth, chondrocyte hypertrophy, calcifi cation, and expression of MMPs, ADAMTS and VEGF in limb buds or growth plates require activation of canonical Wnt signaling [32,33]. Forced expression of the constitutively active form of LEF in chick chondrocytes stimulates ectopic EO [34]. Additionally, mis-expression of Frzd-1, a Wnt antagonist, led to delayed chondrocyte maturation, metalloprotease expression and marrow/ bone formation [35], thus suggesting a positive role of Wnt signaling in promoting chondrocyte maturation. Th ese data confi rmed the pivotal role of Wnt-β-catenin in chondrocyte maturation and hypertrophy during EO.
Recent studies also suggest that GTPases play a signifi cant role in both chondrocyte development and maturation. Rac1 and Cdc42 are co-expressed in both articular and growth plate chondrocytes, and they function to accelerate the rate of chondrocyte diff erentiation by increasing COLX promoter activity [36]. Kerr and colleagues [37] found that levels of active Rho GTPases increased with chick chondrocyte maturation. Th e activated Rac1 expression induced chondrocyte enlargement and MMP13 upregulation, suggesting a positive role of Rac1 in chondrocyte maturation. Additionally, Rac1 and Cdc42 are required for chondrocyte condensation mediated by N-cadherin and act as positive regulators of chondrogenesis [38]. Th e regulatory eff ect on chondrocyte diff erentiation was verifi ed by gene mutation studies in mice. In vivo, Rac1-defi cient growth plates displayed delayed ossifi cation, reduced chondrocyte proliferation and increased apoptosis [39], partly due to reduced mitogenic activity through Rac1-inducible nitric oxide synthase-nitric oxide signaling in EO [40]. Similar results were observed in limb bud development. One study reported that the specifi c deletion of Rac1 (Msx-2 cre) caused severe truncations of limb buds due to impaired nuclear transport of β-catenin [26]. Studies by Kamijo and colleagues reported that both Rac1 [41] and Cdc42 [42] are essential for interdigital programmed cell death through regulation of Bmp, Msx1, and Msx2 gene expression.
A study by Beier and colleagues [43] demonstrated an antagonistic eff ect of RhoA/ROCK signaling on chon drocyte diff erentiation, in contrast to Rac1/Cdc42 signaling [44]. Over-expression of RhoA in ATDC5 cells resulted in delayed hypertrophic diff erentiation with reduced COLX and MMP13 expression. However, pharmacological inhibition of RhoA/ROCK by Y27632 increases Sox9, COLII and aggrecan mRNA levels during chondrogenesis in monolayer culture systems. Th e observed eff ects of RhoA/ROCK signaling appeared to be antagonistic in a three-dimensional micromass culture system [45]. Similarly, the study by Lassar and colleagues also reported that RhoA/ROCK signaling regulated Sox9 transcriptional activity through actin polymerization mediated by protein kinase A phosphorylation of Sox9 [46]. By contrast, studies of D'Lima and colleagues [47] demonstrated that ROCK, a downstream eff ector of RhoA, directly phosphorylates Sox9, which in turn regulates chondrogenesis. Th is suggests that RhoA functions through signaling pathways other than ROCK in modulating chondrogenesis [48]. Recently, Sox9 has been demonstrated to correlate with Mef2c in modulating chondrocyte terminal diff erentiation [49], suggesting that Rho GTPases may function upstream of Sox9 during chondrocyte diff erentiation.
In summary, Rac1/Cdc42 and RhoA/ROCK signaling pathways are all expressed during chondrogenesis and have adverse eff ects on chondrocyte terminal diff er entiation (hypertrophy-like change). Th e Rac1/Cdc42 signaling pathway accelerates chondrocyte hypertrophy while the RhoA/ROCK signaling pathway delays chondrocyte maturation through regulation of Sox9, as illustrated in Figure 2, but the underlying mechanisms are still poorly understood.

Canonical Wnt signaling and pathological changes in osteoarthritis
Wnt-β-catenin signaling is activated in both human and mice OA cartilage. In fact, many animal model studies utilizing a genetic approach have strengthened this view. Mechanical injury, a major cause of OA, leads to downregulation of Wnt antagonist FRZB and up-regulation of ligand Wnt16 and target genes encoding β-catenin, Axin-2, C-JUN and LEF-1 [50]. Furthermore, trans criptome analysis demonstrated that expression of Wnt1inducible signaling protein 1 (WISP-1) is increased twofold in cartilage lesions compared to healthy intact cartilage [51]. Th ese fi ndings indicate that Wnt signaling may function as an OA initiation factor upon mechanical injury. Corr and colleagues [52,53] fi rst reported that Arg200Trp and Arg324Gly Frzb variants, encoding sFRP3, an extracellular inhibitor of Wnt-β-catenin signaling, contributed to genetic susceptibility of women to hip OA. However, the same conclusions were not reached by another two groups that investigated other populations [ 54,55]. Although Min and colleagues [56] thought that these two variants are also associated with other generalized OA at multiple sites, there is still no direct evidence implicating Frzb variants in knee OA. Frzb knockout mice display increased cartilage damage and thicker cortical bone formation [57]. Given the close relationship between bone shape and OA development, Baker-Lepain and colleagues [58] believed that SNPs in Frzb are associated with the shape of proximal femur and further contribute to hip OA development. However, some pertinent questions remain: do these two variants increase wnt ligand binding with the Frizzled protein to activate Wnt-β-catenin signaling; and does mis-function of Frzb in chondrocytes directly lead to OA or Frzb modulation of bone shape, disrupting mechanical loading on cartilage and consequently leading to OA? Th e inhibition of Dickkopf-1 (Dkk1), a negative regulator of Wnt-β-catenin signaling, has been reported to be able to reverse the bone-destructive characteristics of rheumatoid arthritis to the bone-forming characteristics of OA [59]. Another study on the mouse OA model also demonstrated that control of Dkk1 expression prevents joint cartilage deterioration in osteoarthritic knees through attenuating the apoptosis regulator Bax, MMP3 and RANKL (receptor activator of nuclear factor kappa-B ligand) [60]. Additionally, Blom and colleagues [61] showed that stimulation of Wnt-induced signaling protein 1 (WISP1) in chondrocytes resulted in IL1dependent induction of MMPs and aggrecanase, suggesting induction of chondrocyte maturation. LRP5 is located in chromosome 11q12-13, which is thought to be an OA susceptibility locus [62]. Lrp5-/-mice displayed increased cartilage degradation, probably due to low bone mass density [63]. Th ese studies thus provide indirect evidence for Wnt-β-catenin participation in OA progression. Zhu and colleagues [64] provided direct evidence for the fi rst time that β-catenin is implicated in the development of OA. Th e conditional activation of β-catenin in articular chondrocytes of adult mice caused OA-like cartilage degradation and osteophyte formation, and this was associated with accelerated chondrocyte maturation and MMP13 expression. Later, the authors reported a somewhat contradictory fi nding that selective suppression of β-catenin signaling in articular chondrocytes also causes OA-like cartilage degradation in Col2a1-ICAT (inhibitor of β-catenin and TCF) transgenic mice [65]. Th is led Kawaguchi [66] to hypothesize that β-catenin induces chondrocyte maturation similarly to Runx2, whereas it suppresses chondrocyte apoptosis similarly to osteoprotegerin (Table 1).
Although most current studies in the scientifi c literature demonstrate the involvement of canonical Wnt-βcatenin signaling in OA development, the role of this signaling pathway in OA pathophysiology is actually dependent on patient characteristics. For instance, two SNPs in FRZB were initially thought to be associated with an increased risk of primary hip OA among female patients [52,53]. However, confl icting data were reported by diff erent studies [54,55]. Th e relationship between FRZB SNPs and human OA development may be dependent on the characteristics of the patient population, that is, sex and age-related diff erences. Excessive or insuffi cient β-catenin signaling in mice chondrocytes has been shown to increase susceptibility to OA phenotype [64,65], thus suggesting that balanced β-catenin levels are essential for maintaining homeostasis of articular chondrocytes. Factors impairing this balance could lead to pathological changes in chondrocytes by promoting either terminal diff erentiation or apoptosis.
Moreover, because OA is a systemic joint disease aff ecting overall joint tissues, including cartilage, subchondral bone and synovium, imbalance of β-catenin signaling in tissues other than cartilage could also initiate or promote OA development. For example, because canonical Wnt signaling has direct roles in osteogenesis, excessive Wnt signaling can also lead to increased bone formation, which might be associated with osteophyte formation. Two Wnt antagonists, sFRP1, which binds to RANKL [67], and DKK1, which promotes osteoprotegerin secretion [58], can alter the balance between osteoclast and osteoblast development. Additionally, upregulated DKK1 levels in synovial fi broblasts contribute to synovial hypervascularity in OA [68], which would imply that modulating DKK1 expression in synovial fi broblasts may be a potential therapeutic strategy for OA-induced synovitis and joint degradation.

Rho GTPases and pathological changes in osteoarthritis chondrocytes
With increasing recognition of the role of Rho GTPase activities in chondrocyte hypertrophy-like changes, their eff ects on OA have attracted much attention and have been investigated using both human genetic studies and animal models. Epidemiological studies from diff erent groups reported a relationship between SNPs in RhoB and OA susceptibility in some populations [69,70]. Meanwhile, rodent OA models treated with the Rho kinase inhibitor AS1892802 displayed alleviation of cartilage damage [71]. RhoB is downregulated in OA articular chondrocytes and is thought to be responsible for signifi cant DNA damage observed in the pre-apoptotic phenotype of OA chondrocytes [72]. RhoA-ROCK signaling is thought to be involved in early phase response to abnormal mechanical stimuli, which is accepted as a contributory factor to OA initiation and progression [73]. In addition, RhoA-ROCK signaling has also been demonstrated to interact with other patho logical factors associated with OA such as transforming growth factor-epidermal growth factor receptor signaling factors [74], IL1a, insulin-like growth factor-1 (IGF-1) [75] and leptin [76], suggesting a global role of RhoA-ROCK in OA progression. With regards to the Rac1/Cdc42 signaling pathway in OA progression, Cdc42-GTP content decreases [77] while Rac1-GTP increases with chondrocyte aging. Th is provides new insights into agerelated OA development. Additionally, Rac1 regulates CTGF/CCN2 gene expression [78], which is upregulated in OA, and has been shown to be benefi cial for articular cartilage regeneration in a mono-iodoacetate (MIA)induced OA model and articular cartilage defect model [79]. A recent study by Long and colleagues [80] showed that Rac1 is involved in Fnf-induced MMP13 production, thus suggesting a metabolic role of Rac1 activation in cartilage ( Figure 3).
Th e role of Rho GTPases in OA progression may not only be limited to cartilage, but may also involve synovium and osteochondral bone. Rac and its regulators -GEFs and GAPs -have been proven to play vital roles in STAT signaling transduction [81][82][83][84][85], which is essential for the infl ammatory response, thus suggesting the important role of Rac GTPases in OA joint infl ammation [86]. Our preliminary results also showed that intraarticular administration of the Rac1 inhibitor NSC2376 effi caciously decreases mRNA transcript levels of pro-infl ammatory factors in joint tissue (unpublished data). Moreover, Rho GTPases also have important roles in mature osteoclasts by regulating the formation of actin rings and resorption lacunae [87] and are required for osteoclast diff erentiation [88]. Th e defi nitive role of Rho GTPase expression in osteochondral bone that contributes to OA progression needs to be further studied. Our preliminary study investigating human OA cartilage shows that Rac1 is activated in OA chondrocytes and the level of Rac1-GTP is greatly upregulated by IL1b in a chondrocyte monolayer culture system (unpublished data), suggesting the important role of Rac1 in proinfl ammatory factor-induced OA progression. Furthermore, primary chondrocytes from OA calcifi ed cartilage (one phenotype of OA) is signifi cantly inhibited by the Rac1 specifi c inhibitor NSC23766, as demonstrated by Alizarin Red staining (unpublished data). Constitutive over-expression of Rac1 resulted in up-regulation of COLX, Runx2 and ADAMTS-5 and intra-articular injection of NSC23766 delayed mice OA development (unpublished data). Due to the high level of expression of Rac1 in human and mouse articular chondrocytes (Figure 4), further studies are focusing on the role of Rac1 in OA development in vivo, and its underlying mechanism. Additionally, the defi ned role of Rho GTPase in OA progression should be further investigated with animal models utilizing both genetic and pharmacological tools.

Table1. Overview of the roles of various elements of the Wnt signaling pathway in osteoarthritis development, as demonstrated by human genetic studies or animal models
As mentioned earlier, Wnt/β-catenin signaling activation leads to elevated articular chondrocyte catabolism, hypertrophy-like changes and cartilage degradation, which are all key features of OA [66]. Rho GTPases have recently been discovered to function as key mediators of β-catenin nuclear translocation and the available data demonstrated signifi cant roles of GTPases in chondrocyte hypertrophy, maturation and OA development [69][70][71][72][73][74][75][76][77][78][79][80]. Interaction between canonical Wnt signal ing and GTPases independent of actin cytoskeletal changes in OA development has not yet been addressed. Th e preliminary results from our study indicate that Rho GTPase modulation of OA may partially function through control of β-catenin nuclear translocation in canonical Wnt signaling.

Wnt signaling and Rho GTPases as targets for OA treatment
Current treatment modalities of OA, including pharmaco logical and surgical procedures, are mainly focused on promoting partial regeneration and relieving pain. For example, acetaminophen, non-steroidal anti-infl am matory drugs (NSAIDS) and cyclooxygense 2 (COX-2) [89] are all utilized to relieve arthritic pain and can achieve good short-term results. Surgical treatment, including lavage, abrasion arthroplasty and microfracture, has long been considered as a palliative therapy for pain, possibly due to removal of infl ammatory factors and bone marrow mesenchymal stem cell-mediated fi brous cartilage regeneration on the subchondral bone [90]. Concerns about later re-emergence of pain and durability of the newly formed fi brous cartilage by micro-fracture makes it imperative to develop more eff ective OA treatment modalities.
Recently, tissue engineering for cartilage regeneration has achieved much progress. Autologous chondrocyte implantation has often been used to treat simple cartilage defects [91,92]. However, chondrocytes in the newly formed cartilage by these procedures are likely to undergo calcifi cation and hypertrophy-like changes, thereby aff ecting cartilage function [93]. Th erefore, to improve therapeutic effi cacy and maintain the functional status of regenerated cartilage, OA treatment should be focused on removing the causes or risk factors of OA. Small molecules targeted to OA-specifi c molecular pathophysio logy may be a good strategy. Available evidence suggests a critical role of Wnt signaling in EO as well as OA development. Excessive levels of some Wnt ligands and β-catenin have been observed in degenerating cartilage. However, this seems to be a paradox because several Wnt signaling antagonists, including DKK1, FRP1, FRP2, and FRP4, are strongly expresssed in OA synovium and cartilage. Th is may possibly be explained by the conjecture that both gain or loss of function of Wnt/β-catenin signaling would disrupt cartilage homeostasis and lead to pathological changes associated with OA. Aberrant expression of Wnt ligands and Wnt antagonists in synovium may function as an early signal to initiate OA, which in turn can be utilized as an easily accessible OA prognostic marker.
Both genetic and experimental studies have highlighted the great potential of locally modulating the Wnt signaling pathway to alter OA prognosis. Rho GTPases have been recently discovered to modulate β-catenin nuclear translocation and control β-catenin/TCF transcription activity. An altered level of Rho GTPases in articular chondrocytes might therefore be recognized as a new marker for OA development. Hence, Rho GTPases may be good targeting candidates to develop small molecule drugs for OA therapy. In fact, many ROCK inhibitors have recently emerged and have been reported in the patent literature. Some of these are utilized for infl ammatory disorders such as multiple sclerosis and asthma. In particular, fasudil hydrochloride, a potent ROCK inhibitor, has been clinically used to treat cerebral vasospasm [94] and pulmonary hypertension [95].
Although blocking the activity of some members of the Rho GTPases family is able to prevent chondrocytes from undergoing hypertrophy and ossifi cation, there are several pertinent problems to be solved before this strategy can be utilized as a means of OA therapy. Th eoretically, Rho GTPases interact with the Sox9 and Runx2 pathways in maintaining a fi ne balance between chondrogenesis and chondrocyte terminal diff erentiation.
Th e underlying mechanism needs further investigation to identify more specifi c intervening signal molecules implicated in chondrocyte hypertrophy-like changes. Alternatively, Rho GTPase eff ectors could be more promising drug targets, because each of these eff ectors mediates specifi c roles of Rho GTPases. To date, modulating Rho GTPases to prevent chondrocytes from undergoing hypertrophy-like change has been evaluated in several animal studies and have demonstrated signifi cant effi cacy in OA therapy [71]. However, many scientifi c questions about the application of Rho GTPases for OA treatment still remain to be answered.
Last but not least, since Wnt and Rho GTPases have important signaling roles in numerous cell types, systemic administration of modulators of these pathways could be dangerous. Localized drug delivery may be a solution. Some biomaterials, such as chitosan and alginate microspheres, may serve as delivery vehicles for controlled drug release in designated tissues. Because Wnt and Rho GTPase signaling pathways modulate both early chondrogenesis (which should be promoted for cartilage repair) and hypertrophic diff erentiation (which should be suppressed), there should ideally be programmed drug administration for initial activation of these signaling pathways to promote chondrogenesis, followed by inhibition at a later time point to prevent chondrocyte terminal diff erentiation. Unpublished results from our lab showed that mesenchymal stem cells seeded on biomaterials incorporated with cytokines promoted cartilage repair. Th ereafter, intra-articular injection of Rho GTPase inhibitors at a later time point could block terminal diff erentiation of the newly formed chondrocytes.

Conclusion
OA articular chondrocytes undergo hypertrophy-like changes, which is a similar process to EO. Wnt/β-catenin and Rho GTPases, mainly RhoA, Rac1 and Cdc42, are well recognized as crucial regulators or mediators of chondrocyte development and chondrocyte hypertrophy during EO. It is now well established that Wnt/β-catenin and Rho GTPases have similar roles in OA progression and local modulation of the Wnt signaling pathway delays OA development. Preliminary studies have illustrated that Rac1 inhibition suppressed OA articular chondrocytes from undergoing hypertrophy-like changes both in vivo and in vitro. Moreover, Rac1 inhibitors may also be promising drugs for preventing chondrocyte ossifi cation in cartilage tissue engineering. Other members of the Rho GTPase family may also possess similar potential as molecular targets for OA therapy. It was only in the last few years that the roles of Rho GTPases in modulating chondrocyte development and OA were intensively studied. Th eir regulatory eff ects on chondrocyte hypertrophy-like change warrants the use of Rho GTPase activators or inhibitors for OA prevention and cartilage tissue engineering. However, several concerns need to be addressed before Rho GTPase modulation is utilized as a means of OA therapy: the dosage and timing of intervention should be carefully investigated; appropriate controlled release systems may potentiate sustained function of Rho GTPases in OA joints; and drugs targeting specifi c eff ectors of Rho GTPases should be further developed to avoid side eff ects.

Competing interests
The authors declare that they have no competing interests.