|Home | About | Journals | Submit | Contact Us | Français|
RhoA GTPase is a key intracellular regulator of actomyosin dynamics and other cell functions, including adhesion, proliferation, survival, and gene expression. Most of our knowledge of RhoA signaling function is from studies in immortalized cell lines utilizing inhibitors or dominant mutant overexpression, both of which are limited in terms of specificity, dosage, and clonal variation. Recent mouse gene targeting studies of rhoA and its regulators/effectors have revealed cell type-specific signaling mechanisms in the context of mammalian physiology. The new knowledge may present therapeutic opportunities for the rational targeting of RhoA signaling-mediated pathophysiologies.
RhoA is a founding member of the Rho GTPase family and serves as an intracellular molecular switch, cycling between a GTP-bound active form and a GDP-bound inactive form. RhoA activity and intracellular localization are regulated primarily by guanine nucleotide exchange factors (GEFs),2 GTPase-activating proteins (GAPs), and guanine nucleotide-dissociation inhibitors (GDIs). GEFs catalyze the release of GDP from RhoA and thereby promote GTP binding to RhoA. GAPs stimulate the intrinsic GTPase activity of RhoA, causing the hydrolysis of GTP and RhoA inactivation. GDIs inhibit the dissociation of GDP from RhoA and also extract RhoA from the membrane where RhoA executes its biological functions. Through direct interaction with downstream effectors, activated RhoA transduces signals and regulates multiple cellular processes (1). RhoA is critical for several fundamental cell functions, including migration, adhesion, survival, cell division, gene expression, and vesicle trafficking (Fig. 1) (1).
In this minireview, we describe the recent progress in studying RhoA signaling in mammalian physiology using mouse genetic models, especially conditional rhoA knock-out mice and RhoA regulator or effector gene-targeted mice. Similar to what we have learned by mouse gene targeting studies of other Rho GTPase family members such as Cdc42 and Rac (2–4), the mouse model studies of RhoA signaling confirm some conclusions of earlier in vitro studies, but more importantly, it sheds light on novel cell type-specific signaling roles of RhoA under diverse physiologic conditions.
The cell function and mode of regulation of RhoA have been intensively studied over the past 2 decades to help establish a biochemical signaling module of RhoA (Fig. 1) (1). The majority of the cell functional studies were carried out in tissue culture and, in many cases, depended on immortalized cell lines, which may behave differently from the primary cells that they originated from. In addition, tissue culture conditions cannot fully recapitulate the physiologic environment. Importantly, the cell signaling network can be altered in immortalized cells and by in vitro conditions.
In addition, the tools used in cell culture systems to study RhoA signaling also have limitations. A commonly used approach to manipulate RhoA activity is through dominant mutant overexpression. The use of the mutants may elicit off-target effects, as they are often overexpressed over endogenous RhoA levels and can tie up multiple regulatory GEFs or GAPs to affect other related Rho GTPase signaling activities or to affect the dynamics of effector interactions at precise intracellular locations. Another tool for studies of RhoA signaling has been the Clostridium botulinum C3 exoenzyme toxin, which ADP-ribosylates the conserved Asn-41 residue of Rho GTPase subfamily members (RhoA, RhoB, and RhoC), thereby inactivating their effector-binding domain (5, 6). Although it shows reduced ability in inactivating other Rho GTPases (e.g. Rac1 and Cdc42), C3 toxin is not useful for dissecting the RhoA function from that of RhoB or RhoC (5, 6), which might be involved in distinct signaling functions (7), and may affect non-Rho signaling such as JNK and p38 (8) with a toxic effect (9). Thus, caution needs to be applied in interpreting cell biological results obtained by these approaches.
Considering the limitations of conventional methods in studying RhoA signaling function, genetic manipulation of RhoA is superior in understanding the bona fide role of RhoA in individual tissue/cell types. Recently, several laboratories have generated conditional RhoA knock-out mouse models. These RhoA loss-of-function mouse models help confirm some previously observed RhoA cell functions. Significantly, they have also revealed some unexpected roles of RhoA in specific cell types under varying physiologic conditions (Table 1).
Significant knowledge about RhoA cell function derives from earlier studies in fibroblast cell lines. Dominant RhoA mutant overexpression or C3 toxin treatment showed that RhoA is essential for the organization of the actomyosin network and for focal adhesion (1). Subsequent rhoB or rhoC mouse knock-out studies found no such phenotypes (10, 11); therefore, it has been postulated that RhoA is the major regulator of actomyosin and focal adhesion dynamics among Rho GTPases. Surprisingly, recent studies of rhoA null MEF cells found no actomyosin or focal adhesion defects (12). Consistently, deletion of rhoA did not alter the activity of downstream signaling targets, the actomyosin machinery myosin light chain (MLC) and cofilin, or MAL/SRF-mediated transcription, a pathway known to depend on F-actin dynamics (1). A significant increase in RhoC expression and activity was observed in rhoA null MEF cells. Furthermore, deletion of rhoC in combination with knockdown of RhoB in rhoA null MEF cells fully disrupted the formation of stress fibers and focal adhesion, indicating a redundancy of RhoA, RhoB, and RhoC in the regulation of cytoskeletal organization and cell-extracellular matrix interaction (Fig. 2A).
RhoA was found to be required for cell cycle progress in G1/S phase transition and cytokinesis (13, 14) and for controlling p21Cip1 and p27Kip1 cell cycle regulator expression and stability (15, 16). In rhoA null MEF cells, cell cycle G1/S phase progression and p21Cip1 and p27Kip1 expression appeared normal, but cytokinesis was defective, despite the unaltered MLC activity (12). Therefore, RhoA seems to control cytokinesis by utilizing a signaling pathway parallel to MLC activity. In addition, despite a high degree of conservation in sequence and subcellular localization among RhoA, RhoB, and RhoC (7), a RhoA-unique pathway must regulate cytokinesis (Fig. 2A). It will be interesting to pinpoint such a mechanism.
Mutant overexpression and inhibitor studies in cultured keratinocyte cell lines have shown that RhoA signaling is essential for regulating skin cell proliferation and differentiation (17) and cell-cell contacts (18). Different from that in MEF cells (12), keratin-5 Cre-driven rhoA deletion in keratinocytes in vivo was sufficient to reduce phosphorylation of MLC and cofilin (19). Primary keratinocyte culture showed additional defects upon rhoA deletion: impaired junction maturation, a mild cytokinesis defect, and decreased directional migration. However, rhoA deletion did not affect development of the epidermis or hair follicles or the formation and maintenance of cell-cell contacts, including adherens junctions and tight junctions, in vivo. Furthermore, no cytokinesis defects were detected in rhoA null keratinocytes in vivo (19), and functionally, the rhoA-deficient epidermis displayed no defects in wound healing. The discordance between in vivo and in vitro knock-out studies may be attributable to compensation by the related RhoB because rhoA deficiency increased RhoB expression in keratinocytes. These data, along with the observed differences between MEFs and keratinocytes, indicate that RhoA signaling and function are cell context-dependent, thereby emphasizing the influence of environment on cell signaling mechanisms.
The role of RhoA in regulating cardiac physiology and pathophysiology remains largely unclear. On one hand, RhoA activity is increased by insults such as H2O2, glucose deprivation, and ischemia/reperfusion (I/R) (20). Modest expression of constitutively active RhoA activates PKD, which appears to be beneficial to protect against injuries (i.e. I/R) (21). On the other hand, overexpression of RhoA in cardiomyocytes induces apoptosis by up-regulating Bax (22). Furthermore, prolonged expression of active RhoA during development results in hypertrophy (21), and pharmaceutical inhibition of its effector Rho kinase (ROCK) suppresses hypertrophy induced by pressure overload (23). Knock-out studies implicate the cardio-supportive function of RhoA. Hearts of cardiac-specific rhoA knock-out mice, driven by β-myosin heavy chain-Cre, had significantly increased lactate dehydrogenase (LDH) release and infarct size after I/R challenge (21). Similarly, PKD phosphorylation was diminished. Thus, RhoA protects the heart against I/R damage through PKD and is important to prevent LDH release and heart infarction (Fig. 2B).
Apical constriction of epithelial cells, which is commonly observed during development processes, including gastrulation, neural tube folding, and lens placode invagination (24), is driven by polarized contraction of the actomyosin cytoskeleton at the apical side of epithelial cells (25). RhoA and its downstream signaling components such as ROCK and MLC were shown to be apically enriched during the apical constriction of epithelial tissues such as the lens placode, and inhibition of the RhoA pathway via inhibitors reduced apical constriction (26). Consistently, Le-Cre-driven deletion of rhoA in the lens epithelium impaired apical constriction (27). In the apical half of the cell, rhoA deletion led to decreased MLC activity and decreased the contraction force necessary for constriction. In the basal half, distribution of Rac1 and its downstream effector Arp2/3 was expanded, leading to protrusion of actin filaments and cell elongation (27). Thus, RhoA regulates lens epithelium morphology by activating MLC and actomyosin contractility in the apical half and restricting Arp2/3 activity and polymerization of protrusive actin in the basal half (Fig. 2C).
Another aspect of eye development regulated by RhoA is transient closure of eyelids. Normally, mice are born with closed eyelids, and this is maintained up to postnatal day 12 (28). Deletion of ROCK-I or ROCK-II or mutation of MAP3K1 (a downstream target of RhoA-ROCK) resulted in an eyes-open-at-birth (EOB) phenotype (29, 30). Interestingly, deletion of rhoA in the eyelid epithelium driven by Le-Cre did not impair eyelid development. However, in the Map3k1 hemizygotic background, conditional deletion of rhoA delayed eyelid closure, possibly through regulation of MAP3K1 expression (31).
Neural development is a complex process involving the proliferation, differentiation, and migration of neurons and their supporting glia (32–34), and RhoA may be involved in all of these functions (1). To form proper neural circuits, neurons must also form axons, which extend over sometimes long and complicated paths to connect with specific target cells (35). Actin cytoskeletal dynamics plays a key role in axon growth and guidance, and in vitro studies have shown that RhoA activity increases actomyosin contractility and induces collapse of the growth cone and repulsion of the axons (36).
The function of RhoA in regulating adherens junctions in the central nervous system has been studied using several cell type-specific Cre transgenic models, e.g. driven by Wnt1-Cre in the mesencephalon, Foxg1-Cre and Brn4-Cre in the forebrain, and Olig2-Cre in the spinal cord. Deletion of rhoA results in dysplasia and disrupts the distribution of adherens junction components such as N-cadherin, αE-catenin, β-catenin, and F-actin (37–39). In rhoA knock-outs, rosette-like structures, seen previously in N-cadherin null mutants (40), were observed in the brain and spinal cord, and cells were found to invade the lumen of the neural tube (38), suggesting a disturbance of adherens junctions and a disruption of cell-cell contact in the neuroepithelium.
Although RhoA appears to have a similar role in regulating cell-cell contact in the brain and spinal cord, it has differential functions in regulating the proliferation of these two central nervous system tissues. Brain-specific deletion of rhoA (driven by Wnt1-Cre or Foxg1-Cre) caused expansion of neural progenitor cells as a result of increased proliferation and reduced cell cycle exit (37); this is similar to αE-catenin knock-out brains (41). In sharp contrast, spinal cord-specific deletion of rhoA (driven by Brn4-Cre) reduced neuron proliferation and caused precocious cell cycle exit, perhaps as a result of premature differentiation of neural progenitor cells (38). Interestingly, phosphohistone 3-labeled mitotic cells dispersed throughout the neuroepithelium in rhoA-deficient brain and spinal cord, instead of the tight luminal localization in the wild-type neural tube (37, 38), suggesting an alteration of the neural progenitor niche.
The function of RhoA in regulating neural circuit organization has also been intensively investigated. Deletion of rhoA in ventral or dorsal spinal cord motor neuron progenitors by Olig2-Cre or Wnt1-Cre resulted in a rabbit-like hopping gait, a typical locomotor circuitry disruption phenotype. Hopping phenotypes were also documented in mutant mice for axon outgrowth repellant ephrin B3 or its receptor EphA4 (42–44). Ephrin B3 is highly expressed at the midline of the developing spinal cord, and it prevents the axon expressing its EphA4 receptor from crossing the midline (43, 45). The axons do in fact cross the midline in ephrin B3 or EphA4 mutant mice, leading to a disruption of locomotor circuitry (42, 44). Axons of rhoA-deficient corticospinal neurons and interneurons display a similar midline-crossing phenotype, as well as a loss of ephrin B3 expression at the spinal cord midline, causing a deficiency in ephrin B3-EphA4 signaling (Fig. 2D) (38). In contrast to corticospinal neurons or interneurons, the central and peripheral projections of rhoA null (driven by Wnt1-Cre) dorsal root ganglion (DRG) neurons appear normal (46). Consistently, Sema3-mediated axon repulsion, a process previously described as RhoA-dependent behavior (47), was maintained in rhoA null DRG neurons in vitro. This could be due to a compensation by RhoC, the expression of which was significantly increased in rhoA null DRG neurons.
During neocortical development, neurons are sorted into six horizontal layers, and neurons generated in the proliferative zones migrate to their destinations in the cortex (33). Deletion of rhoA in the cerebral cortex (driven by Emx1-Cre) resulted in two migrational phenotypes by two different mechanisms. In the first, rhoA null neurons migrated faster than wild-type cells, with some mutant cells migrating to the pial surface, and resembled the cobblestone lissencephaly phenotype. In the second phenotype, the heterotopic cortex was generated underneath the normatopic cortex, as in subcortical band heterotopia (SBH) (48). This SBH phenotype appears to be due to cytoskeletal defects within radial glia rather than neurons themselves. rhoA null neurons transplanted into the wild-type cortex migrated normally, but not the other way around. The radial glial scaffold was severely disrupted after loss of rhoA cell somata were scattered rather than tightly aligned, and processes were disorganized and no longer connected the upper and lower halves of the cerebral cortex. Furthermore, severe disruptions in both actin and microtubule cytoskeletons were observed in radial glial cells, and this might be the underlying mechanism for the SBH phenotype.
The hematolymphoid system is derived from primitive hematopoietic stem and progenitor cells (HSPCs). Inhibition of RhoA activity by overexpressing dominant-negative mutant RhoA enhanced HSPC proliferation and engraftment (49). In contrast, increasing RhoA activity by knocking out of one of the major RhoA GAPs in the primitive HSPCs (p190B) also enhanced HSPC engraftment (50). These observations made it difficult to interpret the role of RhoA in regulating HSPCs in the suppression-of-function and gain-of-function models. In contrast, loss of rhoA in hematopoietic lineages driven by Mx1-Cre resulted in complete disruption of hematopoiesis and pancytopenia. Interestingly, only blood progenitors, rather than hematopoietic stem cells (HSCs), were severely affected by the loss of rhoA. rhoA null HSCs retained the ability to compete with wild-type counterparts, whereas the progenitors could not. Consistently, rhoA−/− hematopoietic progenitor cells (HPC), rather than HSCs, displayed a significantly increased TNFα-mediated necrosis and cytokinesis failure (Fig. 2E) (51). Within the blood progenitor populations, rhoA knock-out reduced MLC activity, which is thought to be indispensable for cytokinesis (14); however, it remains to be determined whether this contributes to the cytokinesis failure phenotype.
G-protein-coupled receptor agonists (e.g. thrombin, ADP, and thromboxane) are important in regulating the behavior of the megakaryocyte-platelet lineage (52). G13, a major RhoA activator (53), is critical for the shape change of platelets, and the G13-RhoA-ROCK axis is involved in platelet activation (54). Deletion of rhoA by PF4-Cre led to accumulation of megakaryocytes and reduction of platelets, indicating a blockage of the final step of thrombopoiesis (55). rhoA-deleted platelets displayed reduced responses to agonists of G13 and, to a lesser extent, Gq signaling. However, no defects in actin assembly were observed in rhoA-depleted platelets, which resembled the case of rhoA-deleted keratinocytes or MEFs. However, in contrast to keratinocytes or MEFs, in which a compensatory increase in RhoB or RhoC expression was observed after deletion of rhoA (19), RhoA is the major Rho subfamily protein expressed in platelets (56), and it is unlikely that RhoB or RhoC can compensate for the deletion of rhoA. rhoA-deleted platelets also displayed reduced integrin-mediated clot retraction but normal spreading on fibronectin, an interesting result, given that both functions are downstream of integrin outside-in signaling. It is not surprising that the PF4-Cre-driven rhoA-deleted mouse model showed prolonged tail bleeding and decreased vulnerability to ischemic brain infarction. Thus, RhoA may be a valid target for preventing and/or alleviating thrombus formation.
B cells are also derived from HSCs through a common lymphoid progenitor cell, progenitor B (proB) cell, precursor B (preB) cell, and immature B cell hierarchy. In the spleen, immature B cells give rise to transitional B cells and finally mature non-circulating marginal zone and circulating follicular B cells (57). Deletion of rhoA in the entire hematopoietic lineages (driven by Mx1-Cre) increased common lymphoid progenitor cells but significantly reduced proB, preB, and immature B cells in bone marrow (58), resembling the blockage of megakaryocyte-platelet differentiation (55). In contrast, deletion of rhoA from the B cell lineage by CD19-Cre did not alter the numbers of proB, preB, or immature B cells in bone marrow, possibly due to the poor deletion efficiency of CD19-Cre in bone marrow (58, 59). In the spleen, rhoA deficiency resulted in significant reductions in T cells and marginal zone and follicular B cells not because of impaired proliferation but instead reduced survival. Indeed, deletion of rhoA decreased transcription of the receptor for the B cell survival factor BAFF and thus blunted the response to BAFF-induced inhibition of apoptosis (58).
Complementing the rhoA gene targeting studies, recent work specifically deleting selected RhoA regulator or effector genes in mice have also contributed to our understanding of RhoA signaling function in vivo (Table 2). RhoA activity is regulated primarily by three classes of regulators, i.e. GEFs, GAPs, and GDIs. Although expression of some RhoA regulators is ubiquitous (e.g. DLC-1 (60) and p190 (61)), many are tissue and cell type-specific (e.g. Rgnef (62), mgcRacGAP (63), and p190B (64)). RhoA activity in a given cell may result from a combined effect of the various regulators expressed in the cell. In addition, whereas some GEFs and GAPs may signal specifically through RhoA, others may regulate multiple Rho GTPase family members (53, 65, 66). Thus, depending on the expression pattern of various regulators and potential cross-talk with other Rho GTPases, deletion of a given RhoA regulator may or may not result in a similar phenotype as the rhoA knock-out. For example, knocking out the GEF ect2 in MEF cells led to a cytokinesis failure (67) similar to that of rhoA gene deletion (12), but no detectable cytokinesis defects were seen in other GEF knock-out MEFs. Interestingly, deletion of the GAP mgcRacGAP also caused cytokinesis failure (63, 68), indicating that RhoA activity is tightly controlled by the mgcRacGAP and Ect2 GAP/GEF pair during cytokinesis. In another example, deletion of the GAP p190 blocked callosal axons from crossing the midline (69), which is consistent with the aberrant midline crossing of axons of corticospinal neurons and interneurons in rhoA-deleted mice (39), suggesting that the p190-regulated RhoA activity controls this axon behavior. In contrary, some knock-out models of RhoA regulators also display distinct phenotypes from rhoA knock-out mice. For example, in contrast to rhoA knock-out in MEF cells (12), deletion of the GEF rgnef blocked focal adhesion formation (62). As mentioned above, deletion of the GAP p190B a major negative regulator of RhoA, led to increased long-term HSC activity and reduced progenitor activities (50), whereas rhoA knock-out resulted in a differentiation block from HSCs to blood progenitors. It is possible that in a complex process such as hematopoiesis, RhoA activity is tightly regulated by the interplay of multiple regulators. Furthermore, in addition to RhoA, the regulators may also control the activities of other Rho GTPases or signaling components in a defined cell lineage. Gene targeting studies of specific RhoA regulators can be complementary to rhoA knock-out studies for delineating the cell type-specific signaling module of RhoA.
Activated RhoA executes its function by recruiting downstream effectors (Fig. 1). Like RhoA regulation by multiple regulators, engagement of the effectors by RhoA may be complicated by the expression pattern of effectors and possibly a shared involvement by related Rho GTPases. Not surprisingly, effector knock-out and rhoA knock-out mouse models have been reported to show both consistent and discordant phenotypes. For example, neuroepithelial integrity was impaired in both rhoA and mDia knock-outs (37, 38, 70), suggesting that the RhoA-mDia axis might be critical in regulating adherens junctions in the developing central nervous system. In contrast, although the rhoA-deleted central nervous system showed no obvious cytokinesis defects, knocking out citron kinase in mice caused a severe cytokinesis blockage (71). Similarly, deletion of rhoA in the hematopoietic lineage caused pancytopenia and blockage of differentiation at the HPC stage, whereas mDia deletion resulted in a reduced T cell lineage and myeloproliferative phenotype (72, 73). Deficiency in Rho kinases, a major class of effectors of RhoA, can also cause different phenotypes compared with rhoA knock-out in mice. For example, in contrast to the rhoA knock-out eyelid epithelium, in which an EOB phenotype was evident only in the Map3k1 hemizygotic background (31), deletion of either ROCK-I or ROCK-II induced EOB (29, 74), suggesting that other signaling input to ROCKs, in addition to RhoA, is important for regulating eyelid closure. In blood monocytes, ROCK-I binds directly to the tumor suppressor PTEN to regulate the PTEN-mediated phosphatidylinositol 3,4,5-triphosphate, Akt, and cyclin D1 pathway and cell migration (75). Therefore, in addition to the possibility of combined activities of multiple RhoA effectors in a certain tissue, interactions among different effectors, which can also be subject to RhoA-independent signal regulation, may play a compensatory or combinatory role downstream from RhoA. It remains a challenge to sort out which effector pathway(s), e.g. RhoA-ROCK, RhoA-mDia, and/or RhoA-citron, is engaged by RhoA to mediate a signaling function. The gene targeting studies also suggest that the potential therapeutic effects of effector inhibition, such as by ROCK inhibitors, do not necessarily phenocopy RhoA inhibition in applications.
In addition to previous cell biological studies, genetic studies in lower organisms such as the yeast Saccharomyces cerevisiae and the fruit fly Drosophila melanogaster have implicated RhoA orthologs in key processes of organismal viability and morphogenesis (76–78). The recently developed genetic mouse models of RhoA, coupled with RhoA regulator/effector mouse gene targeting efforts, help establish RhoA as an essential signal transducer required for the survival and several fundamental behaviors of mammalian cells. Importantly, the mouse model studies indicate that RhoA can play an essential non-redundant role in one signaling function in a given cell type while being dispensable for a similar function or involved in a different function in another cell type. The cell-type specificity may reflect a pathway-specific nature of the RhoA signaling module and its cross-talk with other related Rho GTPases, similar to the case with Cdc42 and Rac (2–4).
Despite potential concerns associated with the mouse gene targeting approach due to compensatory signaling, the mouse knock-out studies offer an improved method in deciphering the physiologic function of RhoA in mammal tissues. Future studies will better define the interaction between RhoA and its specific regulators and effectors in vivo in formulating a more comprehensive RhoA signaling module in each unique cell type and tissue environment. In addition, the mouse genetic studies may reveal information about the possible involvement of RhoA signaling in human diseases, and RhoA and its regulator/effector knock-out mice will be useful for evaluating the contribution of RhoA signaling in pathophysiology. To this end, abnormal RhoA signaling has been associated with human cancer, neuronal disorders, and asthma (20, 79, 80), and the rational design and targeting of the RhoA signaling pathway (81–84) may bear therapeutic value.
*This work was supported in part by National Institutes of Health Grants R01 CA150547, R01 HL085362, and P30 DK090971.
2The abbreviations used are: