The Rif GTPase is a recent addition to small Rho GTPase family; it shares low homology with other members in the family and evolutionarily parallels with the development of vertebrates. Rif has the conserved Rho GTPase domain structures and cycles between a GDP-bound inactive form and a GTP-bound active form. In its active form, Rif signals through multiple downstream effectors. In the present review, our aim is to summarize the current information about the Rif effectors and how Rif remodels actin cytoskeleton in many aspects.
Rho GTPases were first identified as a family of genes related to Ras-like protein superfamily in the 1980s, and thus were named Rho for Ras homologue . The Rho GTPase family is highly evolutionarily conserved throughout all eukaryotes. It contains 20 members in humans, which can be divided into eight different subfamilies: Cdc42 (cell division cycle 42), Rac, Rho, Rnd, Rif, RhoBTB, RhoH and Wrch . Most Rho small GTPases act as molecular switches, and cycle between a GDP-bound inactive form and a GTP-bound active form, to regulate actin cytoskeleton, cell migration, cell motility, cell polarity, gene expression, vesicle trafficking, cell cycle regulation and nervous system development . Within the Rho families, there are three classical members that have been characterized in detail: RhoA, Rac and Cdc42. Each of them controls a set piece of the actin cytoskeletal architecture. RhoA causes the formation of thick contractile bundles of actin called stress fibres which allow the cell to apply contractile forces across its body; Rac1 activation causes flat sheet-like protrusions from the cell periphery called lamellipodia, which allows the motile cell to move forward; Cdc42 triggers the formation of spike-like projections called filopodia, which help cells sense their external environment [4,5].
The functions of the other members of the human Rho GTPase family are less well understood, and solving the signalling pathways controlled by these proteins has profound importance in cytoskeletal regulation. The Rif GTPase was identified from partial cDNA sequences in the human EST (expressed sequence tag) database using the tblastn algorithm, in a search for novel Rho-family GTPases . The full-length protein encompasses 211 amino acids and contains the same conserved domain structures found in other members of the Rho GTPase family; the GTPase domain with switch I and switch II region, as well as the α3′ helix insert region . The C-terminal prenylation residue leucine is also conserved in Rif, indicating that Rif can be geranylgeranylated and located to molecular membranes. Interestingly, the first 19 amino acids of Rif have no homologue compared with other Rho GTPase family members, indicating that it might mediate some unique protein–protein interaction .
Amino acid sequence alignment of Rif compared with the other members of human Rho GTPase family reveals that Rif is on the same branch as human RhoD, but is quite distantly related to the other family members . The Rho GTPase family is highly conserved in evolution, all of the family members have emerged after the division of the eukaryotes, and most of the human Rho GTPases date back to the division of the chordate phylum . The classical members of the human Rho GTPase family are relatively ancient signalling proteins, which are present in fungi, nematode worms and fruitflies. In contrast, the non-classical human Rho GTPase Rif is found in species evolutionarily higher than Ascidians (including Ascidians themselves), which is the model organism of the invertebrate chordates .
Rif is widely expressed in many human tissues, with the highest level in colon, stomach, spleen and lymphocytes . Actin dynamics are required for the absorptive function of the intestinal epithelium, for nervous cell polarity and capillary growth, and for lymphocyte migration . This tissue distribution pattern suggests that the physiological role of Rif might be partly related with the function of intestinal epithelial cells, hepatocytes and lymphocytes, which all contain microvilli and microvilli-like structures. In the present review, we summarize the current available information regarding how Rif remodels cell cytoskeleton and the signalling pathways behind it.
Rif and filopodia formation
Filopodia are highly dynamic actin-rich cell-surface protrusions that help cells sense their external environment. Cdc42 is the well-known factor to trigger filopodia formation. Cdc42-induced filopodia extend mainly from the periphery of the cell, and Cdc42 also stimulates vinculin, paxillin-containing focal complex formation at the plasma membrane and in particular along and at the tips of filopodia . The actin nucleation at the barbed end of filopodia induced by Cdc42 is mediated by co-operation between WASP (Wiskott–Aldrich syndrome protein)/N-WASP (neuronal WASP) and Arp2/3 (actin-related protein 2/3). Binding Cdc42 to the CRIB (Cdc42- and Rac-interactive binding domain) domain of WASP/N-WASP releases the autoinhibition between the CRIB domain and the C-terminal WCA [WH2 (WASP homology 2) domain, connector region and an acidic peptide] domain, and frees the WCA domain to recruit Arp2/3 and initiate actin nucleation and filopodia elongation . Structure analysis of Cdc42 in complex with the GBD (GTPase-binding domain) of WASP indicated that the GBD makes contact with the Switch I, β2 and α5 regions of Cdc42, but only the Switch I and α5 regions are responsible for the binding specificity .
Overexpressing wild-type Rif or RifQL mutant in HeLa cells all lead to the formation of spike-like filopodia, which are highly dynamic and actin-rich structures . In comparison with Cdc42, Rif-induced filopodia extend from both the cell periphery and the apical cell surface. Rif-induced filopodia do not show co-localization with vinculin-rich focal adhesion complex . The amino acid alignment analysis of Rif with Cdc42 indicated that the residues within the Switch I and α5 regions of Cdc42 that are indispensible for WASP binding are not conserved in Rif. The immunoprecipitation result also shows that Rif cannot interact with WASP in vivo. All of these results imply that Rif does not require the same downstream effector as Cdc42 to induce filopodia formation. Not surprisingly, co-transfection of the dominant-negative Cdc42 and active Rif does not impair Rif-induced filopodia formation, and, similarly, Cdc42-induced filopodia are not inhibited by co-expression with RifTN . Thus Rif does not induce filopodia formation through the Cdc42/WASP/Arp2/3 signalling pathway, Rif and Cdc42 represent two distinct routes in regulating filopodia formation.
As another kind of actin nucleator, DRFs (Diaphanous-related formins) also have the ability to nucleate actin filaments. Co-expression of active RifQL and full-length mDia2 in NIH 3T3 cells leads to the formation of Rif-phenotyped filopodia, and mDia2 is relocalized from cytosol to the filopodia tip . One can imagine that it is at the filopodia tip where mDia2 binds to the dynamic nascent filament and continues the actin filament elongation. By carrying out biochemical analysis, Rif was shown to have a GTP-dependent interaction with the GBD of mDia2 in the yeast two-hybrid assay. Mutation of the key residue His160 in the CRIB-like motif within the GBD does not disrupt the interaction between Rif and mDia2 GBD, showing that Cdc42 and Rif bind different residues within GBD of mDia2 . mDia2 has a strong antoinhibition between its GBD and the C-terminal DAD (Diaphanous autoregulatory domain); a co-immunoprecipitation assay of Rif with the DAD-domain-deleted mDia1 showed that Rif interacts with mDia1ΔDAD in a nucleotide-independent manner. The co-expression of RifTN mutant and mDia2 cannot induce filopodia formation, indicating that, although RifTN has an equal ability to bind mDia2 as RifQL, the full activation of mDia2 needs GTP–Rif . Taken together, mDia2 is required for Rif induction of filopodia, and Cdc42 and Rif produce distinct filopodial projections both in morphology and signalling pathway.
Recent work using time-lapse imaging of live neuronal cells has shown that Rif also interacts directly with mDia1 to induce filopodia formation . This signalling pathway is independent of Cdc42 effectors N-WASP, IRSp53 (insulin receptor substrate protein 53 kDa) and the IRSp53-binding partner Mena (mammalian enabled) . This is consistent with our observation that co-transfection of mDia1 with activated RifQL to HeLa cells leads to a redistribution of mDia1 to Rif-induced filopodia tip and/or along the filopodial length . Knocking down mDia1 using RNA interference in the RifQL expression cells causes the Rif-induced filopodia to form bleb-like structures at the apical surface .
In summary, the non-classical Rho GTPase Rif produces the filopodia structure through controlling actin nucleation effectors mDia1 and mDia2, and these signalling pathways are distinct with the canonical Cdc42/WASP/Arp2/3 filopodia formation pathway, increasing the complexity and specificity of actin cytoskeleton regulation.
Rif and stress fibre formation
Non-muscle cells contain bundles of contractile actin filaments called stress fibres. Stress fibres are composed of bundles of actin filaments and myosin II, and allow non-muscle cells to apply contractile forces. Early work identified RhoA as a crucial regulator of actin stress fibre formation . Downstream of RhoA, there are two key elements: mDia1 and ROCK (Rho-associated protein kinase). These two proteins work co-operatively to induce the formation of thick and organized stress fibres. mDia1 is a formin molecule that catalyses actin nucleation and polymerization; ROCK phosphorylates both myosin and multiple myosin regulatory proteins, such as MYPT (regulatory subunit of myosin light chain phosphatase) and MLC (myosin light chain). Activated RhoA binds to the GBD of mDia1. Binding to the GBD releases an autoinhibitory interaction in mDia1 and stimulates nucleation of new F-actin (filamentous actin) filaments. Activated RhoA can bind ROCK at the C-terminus directly and activate it. The activated ROCK phosphorylates myosin and MYPT, increasing myosin activity and inhibiting the myosin phosphatase. These actions, as a consequence, stimulate cross-linking of actin by myosin and enhance actomyosin contractility .
Expression of wild-type Rif or RifQL mutant also triggers the formation of contractile actin stress fibres. Rif-induced stress fibre formation is completely independent of RhoA, because when dominant-negative RhoA was expressed in HeLa cells, the ability of Rif to induce stress fibres was not affected. Silencing the endogenous Rif expression by RNA interference led to a significant decrease in actin stress fibres under basal conditions. HeLa cells with RifQL and myosin II expression showed a loose sarcomeric array staining pattern of myosin II along the stress fibre induced by activated RifQL .
Rif binds the GBD of mDia1. Silencing mDia1 affected the Rif-induced stress fibre formation. mDia1 has been reported to be a component of focal adhesions, which are the sites of stress fibre initiation and attachment . This has led to the proposal that mDia1 seeds actin stress fibres through nucleation of actin filaments at focal adhesion sites. Despite the fact that RifQL alters the localization of mDia1, from cytoplasm to the tips of Rif-induced filopodia, there was no concentration of Rif or mDia1 at the stress-fibre-attachment site when transfected either together or separately . Rif-induced stress fibres are also dependent on ROCK activity. Both ROCK siRNAs (small interfering RNAs) and ROCK inhibitor Y-27632 inhibit the stress fibre formation induced by RifQL. RhoA activates ROCK by direct interaction with a binding site in the C-terminal region of the protein [19,20]. Although interaction between Rif and ROCK has been shown using yeast two-hybrid analysis, we were unable to demonstrate an interaction between ROCK1 and RifQL by immunoprecipitation, even using a ROCK1 mutant that lacks the C-terminal autoinhibitory domain. Overexpression of RifQL did not increase the phosphorylation of either MLC or MYPT, which are both ROCK targets. Sometimes protein–protein interactions are transient and/or of low affinity and cannot be verified by immunoprecipitation. Rif might have only a transient or weak interaction with ROCK1 and does not increase ROCK activity in cells; however, Rif-induced stress fibre formation depends on the basal level of ROCK activity .
As a non-classical member of Rho GTPase family, Rif demonstrates the possibility of making new interfaces with the canonical stress fibre formation signalling pathway. There are two differences between the Rif-induced stress fibres and RhoA-induced stress fibres. First, Rif-induced stress fibres are restricted to cell lines of epithelial origin, and Rif did not induce stress fibres in fibroblasts. Secondly, the effect of Rif expression on focal adhesion formation was dependent on epithelial cell type, whereas RhoA stimulates stress fibre formation in parallel with an increase the size and maturity of focal adhesions .
Rif and lymphocyte morphology
Circulating lymphocytes maintain a spherical shape, and their surfaces are decorated with actin-based protrusions called microvilli . Lymphocyte microvilli contain parallel bundles of filaments, structurally resembling filopodia. Lymphocytes must undergo dramatic cytoskeletal rearrangements when performing their role in the immune response, particularly during the transition between the blood circulation and the underlying sites of infection . The Rho GTPase family of actin-regulating proteins appears to play a pivotal role in leucocyte morphology regulation, although numerous other protein families and signalling pathways are also involved [23,24].
We found that Rho family member Rif displays a restricted expression pattern and is highly expressed in T- and B-cells, natural killer cells and dendritic cells. This is significant, since Rho family members that show a similar expression pattern play important role in leucocyte signalling [25,26].
Previous work has suggested that Rif has a role in lymphocyte immune functions. Some studies demonstrated that Rif is up-regulated in a subset of transformed FLs (follicular lymphomas), and a significant percentage of FLs transformed to an aggressive DLBCL (diffuse large B-cell lymphoma) . Other research has revealed that Rif is deleted in early-stage mycosis fungoides, which is the most frequent form of cutaneous T-cell lymphoma . The effect of the up-regulation or down-regulation of Rif in these cancer cells is completely unknown. On the basis of all of the above observations, we presume that Rif is related to lymphocyte microvilli actin remodelling, but further work should be carried out with more patients' cell samples and the lymphocytes from Rif-knockout mice.
Rif and neuronal morphogenesis
Neurons are specialized cell types which communicate with other cells through synapses. Neuronal morphology is also determined by the cytoskeletal remodelling controlled by small Rho GTPases. Previous evidence implicates that Rho GTPases play vital roles in neuronal morphogenesis regulation, including migration from the birthplace, extension of axon and dendrites, establishment of neuronal polarity, neurite branching and synapse formation [29,30]. It is widely accepted that the three classical members of the Rho GTPase family, RhoA, Rac and Cdc42, are important players in neuronal development. For example, during the migration process, the growth cone that is at the front of the axon is composed of filopodia and are interconnected by lamellipodia, so Rac and Cdc42 promote neurite outgrowth, whereas RhoA stimulates retraction at the rear. The balance between these two opposite activities mediating by different Rho GTPases is crucial for the morphology and function of the neurons .
Other members of the Rho GTPase family have been shown to be involved in the dendrite development. Rnd1 promotes spine maturation , and Rnd2 triggers dendrite branching . Intriguingly, recent work has shown that Cdc42 and Rif work together in the formation of dendritic spines, which develop from dendritic filopodia . In detail, the dendritic spines are the post-generated actin-rich protrusions from dendritic shafts, which lie along the dendrites where most excitatory synapses reside. The structures of dendrites and dendritic spines shown by electron microscopy indicate that dendritic spine head is composed of branched filaments, whereas the dendritic filaments and spines necks are made of unbranched filaments . A working model has been proposed: the first stage, spine development, starts with the nucleation and elongation of dendritic filaments that are mediated by Rif and its effector mDia2; in the second stage, the spine head begins to form through the formation of branched actin filaments mediated by Cdc42 and its effector Arp2/3; in the meantime, ADF (actin depolymerization factor)/cofilin1 prevents formation of abnormal protrusion from spine heads; finally, the spine matures and maintains its overall morphology .
Primary nervous cells from Rif-knockout mice are helpful for our understanding of the physiological role of Rif in nervous system development.
As a non-classical member of human Rho GTPase family, the emergence of Rif parallels the development of vertebrates, increasing the complexity of the immune system, the nervous system and lots of other processes requiring complex regulation of cell shape and cell migration. Some of the Rif-binding partners are downstream effectors for a wide range of other Rho family members, while the others are unique for Rif. These suggest that there are two ways in which Rif might add functionality to cytoskeletal regulation, either through making novel interactions with the existing pathways or by controlling new signalling pathways. In the future, it will be important to identify more Rif-binding partners using proteomic methods, and in the meantime, we can take advantage of Rif-knockout mice to obtain more in vivo information about the physiological role of Rif and the signalling network between different Rho GTPase members.
Signalling 2011: a Biochemical Society Centenary Celebration: A Biochemical Society Focused Meeting held at the University of Edinburgh, U.K., 8–10 June 2011. Organized and Edited by Nicholas Brindle (Leicester, U.K.), Simon Cook (The Babraham Institute, U.K.), Jeff McIlhinney (Oxford, U.K.), Simon Morley (University of Sussex, U.K.), Sandip Patel (University College London, U.K.), Susan Pyne (University of Strathclyde, U.K.), Colin Taylor (Cambridge, U.K.), Alan Wallace (AstraZeneca, U.K.) and Stephen Yarwood (Glasgow, U.K.).
actin-related protein 2/3
cell division cycle 42
Cdc42- and Rac-interactive binding domain
Diaphanous autoregulatory domain
FERM, RhoGEF (guanine-nucleotide-exchange factor) (ArhGEF) and pleckstrin domain protein 1
insulin receptor substrate protein 53 kDa
myosin light chain
regulatory subunit of myosin light chain phosphatase
Rho-associated protein kinase
Wiskott–Aldrich syndrome protein
WH2 (WASP homology 2) domain, connector region and an acidic peptide
This work was supported by the Welcome Trust.