Rho GTPases in embryonic development

Abstract

In the last decade, several mouse models for RhoA, Rac1, and Cdc42 have emerged and have contributed a great deal to understanding the precise functions of Rho GTPases at early stages of development. This review summarizes our current knowledge of various mouse models of tissue-specific ablation of Cdc42, Rac1, and RhoA with emphasis on early embryogenesis, epithelial and skin morphogenesis, tubulogenesis, development of the central nervous system, and limb development.

Keywords:

• Cdc42
• Rac1
• RhoA
• embryonic development
• mouse models
• cell polarity
• cell migration
• cell fate
• cytoskeleton
• tubulogenesis

Introduction

The Rho family of small GTPases represents a subgroup within the large superfamily of Ras-related small GTPases. Rho GTPases consist of 20 members acting as molecular switches that oscillate between an active GTP-bound state able to activate several downstream effectors and an inactive GDP-bound state.¹ This GDP/GTP cycle is achieved by guanine nucleotide exchange factors (GEFs) that stimulate the exchange of GDP for GTP and by GTPase-activating proteins (GAPs) that enhance the intrinsic GTPase activity, resulting in GTP hydrolysis and protein inactivation.² Over 70 RhoGAPs and 80 RhoGEFs have been identified and most of them show tissue-specific expression.^1,3 Discussing the in vivo mouse models for these GAPs and GEFs would render this review too lengthy, while being too selective would lead to ignore important studies. However, an interesting link can be drawn between tissue-specific Rho GTPase regulators and the conditional deletion of Rho GTPases in specific tissues. In this review, we will detail the phenotypic similarities and differences between various mouse models of RhoA, Rac1, and Cdc42, the best-characterized members of the Rho GTPase family. Previous studies have highlighted the importance of RhoA, Rac1, and Cdc42 in cytoskeletal rearrangements.^1,4 Activation of Cdc42 leads to the formation of filopodia,^5,6 activation of Rac1 induces lamellipodia⁶ and membrane ruffles,⁷ while RhoA activation leads to contractile actin:myosin filaments⁶ and stress fiber formation.⁸ Among other functions regulated by these small GTPases are microtubule dynamics, signal transduction, gene expression, and enzymatic regulation.^1,9 These molecular functions lead to a variety of cellular behaviors, such as cell migration, polarity, survival, morphology, proliferation, differentiation, and regulation of the various phases of the cell cycle. These cellular functions were in part discovered using dominant-negative mutant proteins expressed in various cell lines. Just as moving from studying biochemical reactions in a test tube to cellular functions in a petri dish has helped tremendously in gaining valuable information, in vivo models are leading us to a better understanding of Rho functions within a physiological context. This, in turn, brings us closer to appreciate the many functions of Rho GTPases in normal physiology and human diseases. Since both Rac1¹⁰ and Cdc42-null mice¹¹ are embryonic lethal during the early steps of development, many different conditional knockout mice were generated to investigate their tissue-specific functions in vivo.^12,13 In this review we will discuss various mouse models with tissue-specific ablation of Cdc42, Rac1, and RhoA with emphasis on early embryogenesis, epithelial morphogenesis, skin morphogenesis, tubulogenesis, development of the central nervous system (CNS), and limb development (summarized in Table 1).

Table 1. Mouse models of RhoA, Rac1, and Cdc42 in early embryogenesis

Mouse Line	Cell Type and/or Tissue	Phenotype	References
Rac1
Rac1 null	germline	early embryonic lethal, growth arrest by E7.5, cell morphology defects, massive apoptosis	10, 16
Transthyretin-Cre	visceral endoderm	AVE migration defects, cytoskeletal defects	18
Sox2-Cre	epiblast	no defect in AVE migration, mesodermal cell migration defects, enlarged primitive streak, cardia bifida	17, 18
Grhl3-Cre	surface ectoderm	exencephaly and/or spina bifida	21
Le-Cre	lens placode	abnormal shape of lens pit and/or increased epithelial curvature, reduction in cell elongation	36
Foxg1-Cre	early forebrain	late embryonic lethal, microcephaly, smaller olfactory bulbs, increased cell cycle exit, no anterior commissure, impaired corpus collosal and hippocampal commissure projections, impaired thalamocortical projections, cortical laminar disorganization, disorganized radial glial fibers, Cobblestone lisencephaly, increased apoptosis	54–56
Emx1-Cre	dorsal telencephalon	absence of anterior commissures and corpus callosum	61
Nkx2.1-Cre	MGE progenitors	lethal at 3 week postnatal, reduced MGE-derived cortical interneurons, reduced cell cycle exit, reduced pRb/cyclinD levels	62
Lhx6-Cre	postmitotic MGE interneurons	lethal early postnatal, normal development of neocortex	62
Dlx5/6-Cre	MGE - subventricular zone progenitors	normal neuronal migration	55
Nestin-Cre	CNS progenitors	lethal at 3 wk postnatal, P10 pups had resting tremors and clenching of hindlimbs, enlarged ventricles, smaller cerebellum with defects in cerebellar granule neuron migration, increased apoptosis	66
Nphs1-Cre	podocytes	no obvious phenotype	38
Col2a1-Cre	cartilage - chondrocytes	reduced viability, dwarfism, kyphosis, disorganized growth plate	67
Msx2-Cre	limb bud ectoderm	severe truncation of limbs, disruption of canonical Wnt signaling	68
Prx1-Cre	limb bud mesenchyme	skeletal deformities, soft tissue syndactyly, absence of interdigital programmed cell death	69
Rac1-Synapsin-Cre;   Rac3 null	postmitotic neurons	lethal at 2 week postnatal, epileptic, reduced cortical inhibitory interneurons	64, 65
Cdc42
Cdc42 null	germline	early embryonic lethal, no primary ectoderm, increased apoptosis, cell adhesion defects	11, 22
Keratin 5-Cre	keratinocytes	hair loss at 4 week postnatal, growth retardation, reduced keratinocyte proliferation and cell fate change	31
Pdx1-Cre	pancreas	absence of tubular structures, apical cell polarity defects	35
Nphs2-Cre	podocytes	lethal at postnatal day 12 due to renal failure, severe proteinuria, reduced nephrin and podocin expression, abnormal junctional complexes	38
Six2-Cre	cap mesenchyme	empty bladders, reduced nephrogenesis, reduced glomerular numbers, reduced nuclear Yap	39
Villin-Cre	intestine epithelium	stomach swelling and constipation, intestinal hyperplasia, intestinal epithelium polarity defects, MVID, Paneth cell differentiation defects, increased intestinal crypt depth, impaired Rab8 activation	41, 42
Lgr5-CreER	intestinal stem cells	cell polarity defects, impaired Paneth cell differentiation	41
TetO7-Cre-SFTPC-rtTA	respiratory epithelial cells	lethal at birth due to respiratory failure, defects in cell-cell contacts, severe tubule defects, branching morphogenesis defects, reduced cell proliferation, cell polarity defects, ectopic basement membrane protein secretion.	37
Foxg1-Cre	early forebrain	thicker neuroepithelium, holoprosencephaly, cell polarity defects, defects in adherens junctions	47
Emx1-Cre	dorsal telencephalon	increased cortical thickness, cell polarity and adherens junctions defects, increased intermediate progenitors and early neurogenesis	45
Nestin-Cre	CNS progenitors	lethal at birth, loss of axonal tracks, enlarged growth cones with diffuse F-actin and absence of filopodia, increased cofilin phosphorylation	46
Olig2-Cre	VZ of MGE	rabbit-like hoping gait, defects in cortical interneuron migration, disorganized neuroepithelium, disruption of adherens junctions, impaired motor neuron survival	50
Dlx5/6	MGE - neural progenitors within the SVZ	no obvious phenotype	50
Prx1-Cre	limb bud mesenchyme	skeletal deformities, soft tissue syndactyly, absence of interdigital programmed cell death	70
RhoA
Keratin 5-Cre	keratinocytes	reduced p-cofilin and p-MLC, normal skin development, keratinocyte migration defects	32
Le-Cre	lens placode	abnormal shape of lens pit and/or reduced epithelial curvature, reduced apical constriction, longer cells	36
Nphs2-Cre	podocytes	no obvious phenotype	38
Brn4-Cre	early spinal cord - neuroepithelial cells	late embryonic lethal, disorganized VZ with loss of apical Par3, decreased cell proliferation and increased in cell cycle exit at early neurogenic stages, rosette-like structures	52
Wnt1-Cre	early midbrain, dorsal spinal cord	disruption in adherens junctions, exencephaly-like protrusions, hyperproliferation, corticospinal midline crossing defects, rabbit-like hoping gait	49, 53
Foxg1-Cre	early forebrain	enlarged forebrain, exencephaly-like protrusions, cell polarity defects, rosette-like structures	49
Emx1-Cre	dorsal telencephalon	heterotopic cortex, type II Coblestone lisencephally, defects in radial glial fibers, actin and microtubule destabilization in radial glial cells, rosette-like structures	51
Olig2-Cre	VZ of MGE	rabbit-like hoping gait, defects in cortical interneuron migration, disorganized neuroepithelium, disruption of adherens junction, impaired motor neuron survival	50, 53
Dlx5/6-Cre	MGE - neural progenitors within the subventricular zone	normal cortical interneuron migration	50

AVE, anterior visceral endoderm; CNS, central nervous system; VZ, ventricular zone; SVZ, subventricular zone; MGE, medial ganglionic eminence; MVID, microvillus inclusion disease.

Early Embryogenesis

As mentioned above, Rho GTPases are implicated in a wide range of cellular processes essential during early development. Thus, it was not surprising that germ line deletion of Rac1 and Cdc42 led to early embryonic lethality. At early blastocyst stage (3.5 days after fertilization), before implantation, the mouse embryo consists of two primary tissue types: the trophectoderm that mediates implantation and surrounds the inner cell mass (ICM).^14,15 The ICM then divides into two regions: the primitive endoderm that gives rise to the extra-embryonic structures, and the primitive ectoderm or epiblast, which will develop into the embryo.^14,15 Already at embryonic day 6.5 (E6.5), Rac1-null embryos are smaller, and the epiblast explant lacks lamellipodia actin structures,¹⁰ consistent with the well-documented role of Rac1 in cytoskeletal dynamics^1,9 and indicating that Rac1 plays an important role before the onset of gastrulation (E6). However, it is only 12 years later, after the first publication of the Rac1-null mouse, that we began to understand in more details the functional role of Rac1 during early embryogenesis.^16-18 Using embryonic stem cells that differentiate into embryoid bodies when cultured in suspension as small aggregates, He and colleagues found that Rac1 was translocated from the cytoplasm to the plasma membrane with a higher concentration at basement membrane contacts.¹⁶ Epiblast cells in contact with the basement membrane form the epiblast epithelium, while cells not in contact undergo apoptosis to form the equivalent of the amniotic cavity. In Rac1-null embryoid bodies, massive apoptosis of cells in contact with the basement membrane indicated defective pro-survival signals by Rac1-deficient cells¹⁶ and eventually the epiblast layer disappeared completely.¹⁶ Some evidence in this study also suggested that the GEF DOCK180 and the adaptor protein Crk were important in activating Rac1 to promote survival via the PI3K-Akt pathway.¹⁶

During the early steps of embryonic development, the body axis (anterior-posterior) is specified by collective cell migration of an extra-embryonic organizer before the onset of gastrulation.¹⁴ Two studies from Migeotte and colleagues characterized the role of Rac1 at the time of gastrulation and during the anterior-posterior body axis formation.^17,18 At E5.5, cells from the distal visceral endoderm (DVE) migrate proximally and become anterior visceral endoderm (AVE) cells.^14,15 Nodal and Wnt pathways play a role in specification and migration of DVE cells, respectively, but the factors regulating the direction of migration are not fully understood. AVE cells establish the putative anterior end and create the anterior-posterior axis of E6.0 mouse embryos.¹⁴ In Rac1-null embryos, the anterior-posterior body axis was impaired due to defects in migration of AVE cells.¹⁸ To assess specifically the role of Rac1 in this process and to eliminate the possibility of a migration defect due to a non-cell autonomous effect from the epiblast, conditional deletion of Rac1 in the visceral endoderm was achieved using a transthyretin-Cre recombinase,¹⁸ leaving a functional Rac1 gene in the epiblast and other tissues. Indeed, the Rac1-null AVE cells lost contact with the basement membrane, had reduced cell shape variability and a more diffuse E-cadherin/F-actin expression pattern. AVE cells lacking Rac1 displayed shorter protrusions, and had strong cortical actin. Thus, Rac1 is required in vivo for collective migration of AVE cells and specification of the anterior-posterior axis due to its crucial function in regulating the actin cytoskeleton.¹⁸ Notably, deletion of Rac1 specifically in the epiblast had no effect on AVE cell migration and normal anterior-posterior body axis formation occurred, indicating that Rac1 acts in a cell-autonomous fashion to regulate AVE cell shape and movement.¹⁸ In the second study by Migeotte and colleagues, epiblast-deleted Rac1 embryos showed disorganized somites, developed cardia bifida, and had an enlarged primitive streak.¹⁷ During gastrulation, following basement membrane degradation, epiblast cells undergo epithelial-to-mesenchymal transition (EMT).^14,15 Defects in this process were observed in epiblast-deleted Rac1 embryos as shown by large laminin aggregates attached to the newly formed mesoderm of the enlarged primitive streak.¹⁷ Similar to the failure of AVE cells to migrate, mesodermal cells failed to migrate away from the primitive streak within the gastrulating embryo.¹⁷ Rac1-deficient mesodermal cells displayed abnormal round morphology with only short protrusions that failed to adhere to the extracellular matrix, ultimately causing cell death.¹⁷ Additionally, Rac1-deficient cells from either E6.5 or E7.5 embryos failed to undergo EMT and exit the explant as they lacked both lamellipodia and filopodia actin structures.¹⁷ Similar to the diffuse E-cadherin expression pattern in Rac1-deficient AVE cells, mesodermal cells that lacked Rac1 had diffuse, large focal adhesions marked by vinculin.¹⁷ Thus, Rac1 has a crucial role in both cell-cell and cell-matrix adhesion during collective cell migration in early embryonic development.^17,18 A decrease in cell proliferation accompanied by an increase in apoptosis was also observed in epiblast-deleted Rac1 embryos.¹⁷ In accordance with the role of Rac1 upstream of the PI3K-Akt pathway, crossing these Rac1-deficient mice with mice heterozygous for the negative regulator of the PI3K-Akt pathway (PTEN +/−) rescued cell death but not the migration and morphological defects.¹⁷ Thus, the pro-survival function mediated by the PI3K-Akt pathway is separable from Rac1's function in adhesion and/or cytoskeletal dynamics involved in collective cell migration. Altogether, these studies highlight the fundamental role of Rac1 during the early steps of embryogenesis.

Neural tube closure (neurulation) is one of the most complex developmental events during embryonic development requiring several morphological changes that need to be tightly coordinated.¹⁹ The incidence of neural tube closure defects in the human population (second most common birth defect)¹⁹ as well as the extensive list of transgenic mice displaying exencephaly and/or a spina bifida phenotype²⁰ is a testimony to its complexity. Among the many cellular processes involved, the importance of cell shape change during neural plate formation, neural plate bending, ECM breakdown, actin rich protrusions at the leading edge, and cell-cell adhesion at the dorsal midline are indispensable.¹⁹ When Rac1 was deleted just before the onset of neurulation in the surface ectoderm of the neural ridge, Rac1-deficient mice developed an exencephaly phenotype.²¹ Future studies will be required to characterize at the molecular level the precise role of Rac1 during neurulation.

Just like Rac1-null mice, Cdc42 germ-line deletion in the mouse is embryonic lethal before E5.5.¹¹ Cdc42-null embryos are smaller, disorganized, with diffuse F-actin, similar to Rac1-null mice. In fact, the Cdc42-null phenotype is more striking than Rac1-deletion as these embryos lack the early primitive ectoderm. Consistent with its role in actin polymerization and filopodia formation, Cdc42-deficient embryonic stem cells could not initiate actin polymerization in response to PIP2.¹¹ Interestingly, in a separate study by Wu and colleagues, Cdc42-null embryoid bodies exhibited reduced Rac1 activity, did not develop an ectoderm, and displayed a severe loss of adherens junctions.²² Specifically, Cdc42-deficient epiblast cells showed a diffuse α-catenin, β-catenin, E-cadherin expression pattern in contrast to localized protein expression at contact sites in control cells.²² Similarly, the expression of the tight junction marker ZO-1 was reduced and was more diffuse in Cdc42-null endodermal cells.²² Additionally, one of the key proteins involved in epithelial cell polarization, the atypical PKC, showed reduced expression with a lack of polarized localization toward the lumen.²² Similar to Rac1-null embryoid bodies, the basement membrane failed to prevent apoptosis in attached cells in Cdc42-null embryoid bodies.²² Taken together, dysregulation of the major components of cell-cell junctions during early development strongly argues for an essential role of Cdc42 in the regulation of epithelial polarity in early tissue morphogenesis. Unlike Rac1 and Cdc42, the generation of RhoA-null mice during these early steps of embryogenesis has not been well documented, although briefly mentioned to be embryonic lethal.²³ It remains to be seen if RhoA-deficient embryoid bodies will be characterized in a similar manner as for Rac1 and Cdc42 in the future.

Epithelial morphogenesis

Every tissue is organized with a layer of polarized epithelial cells serving as a barrier between the tissue and its surrounding environment. This layer of polarized cells that makes up the epithelium serves as both a physical and chemical barrier.²⁴ An epithelium is typically exposed to 3 different environments: (1) a lumen on its apical side, (2) neighboring cells on its lateral sides, and (3) a basement membrane on its basal side. At the molecular level, these differences in membrane composition begin with the asymmetrical distribution of polarity complexes. The Par complex is one of the three major polarity protein complexes involved in establishing asymmetry along with Crumbs and Scribble.^25,26 The Par complex targeted to the apical plasma membrane contains Cdc42-Par6-aPKC while Par3-aPKC localizes at the tight junctions where it separates the lumen-facing apical side from the basolateral regions.^25,26 In addition, PTEN-mediated segregation of phosphoinositides has been shown to initiate polarization of epithelial cells by recruiting Cdc42 to apical domains.²⁷ Thus, Cdc42 is crucial in establishing the putative apical side and overall cell polarity. Furthermore, Rac1 has been shown to play an important role downstream of the Par complex formation.²⁸ Since most tissues and organs are composed of a polarized epithelium lining a lumen, we will discuss how loss of Rho GTPases, in particular Cdc42, affects the development of different tissues and organs. This will not only help us appreciate the different functions of Rho GTPases but also understand how different organs develop.

Skin morphogenesis

The specialized organ that serves as the first barrier of the body from the outside environment is the skin. It consists of the epidermis composed of keratinocytes and the underlying dermis. Basement membrane-attached keratinocytes (basal keratinocytes) serve as stem cells while remaining keratinocytes detach from the basement membrane, stop proliferation, and differentiate into the stratum corneum.²⁹ Keratinocyte stem cells can give rise to both sebaceous glands and hair follicles.²⁹ During embryonic development, Wnt signaling has been shown to be crucial for this specific differentiation cascade.³⁰ Given the link between Cdc42 and the Wnt signaling pathway,⁹ Wu and colleagues have investigated the impact of Cdc42 deletion specifically in keratinocytes.³¹ Although these mice were born without any obvious defects, 2 week old pups displayed impaired hair formation and growth retardation. Additionally, these mice started to lose their hair after four weeks and had severe delay in hair follicle morphogenesis. At the cellular level, Cdc42-deficient progenitor cells displayed a reduction in cell proliferation at postnatal day 9 and changed their fate from hair follicle keratinocytes to epidermal keratinocytes. This was in part due to a loss of nuclear β-catenin that led to a loss of LEF-1 transcription factor and reduction of both ZO-1 and occludin expression in hair follicle keratinocytes of 2 wk old mice. At the molecular level, Cdc42-deficient mice had a reduction in both phosphorylated aPKC and GSK3β which was accompanied with an increase in phosphorylated Axin targeting β-catenin to the degradation machinery.³¹ Thus, during epidermal development, Cdc42 is necessary to regulate β-catenin turnover. Taken together, Cdc42 plays a crucial role during skin development by regulating both cell differentiation and epithelial cell polarity. Unlike Cdc42, RhoA is mainly dispensable for skin development but was found to be required for cell contraction and directed migration of keratinocytes in vitro.³²

Tubulogenesis: Pancreatic, lung, kidney, and intestine development

The dynamic change in cell polarity is a fundamental feature of organogenesis. Many organs consist of tubules transporting vital fluids, serving as both transport and diffusion barriers to selectively enable the absorption of nutrients.²⁴ Thus, understanding the genesis of these tubules is essential in developmental biology. Many organs develop from simple tubules that branch and ramify into complex treelike structures,^33,34 favoring an increase in the surface area and exchange of nutrients. A novel role for Cdc42-mediated tubulogenesis during pancreatic development has been established by removing Cdc42 at the onset of pancreatic development.³⁵ Localization of the apical membrane protein mucin was impaired while E-cadherin and basal laminin were properly localized in Cdc42-deficient mice. The unpolarized epithelium subsequently became fragmented and formed large cellular aggregates lacking tubular structures. Interestingly, the Rho kinase (ROCK) inhibitor Y27632 was able to fully rescue tubule formation in Cdc42-deficient explants, suggesting that Cdc42 and RhoA may play opposing roles during pancreatic tubulogenesis. These results demonstrate the importance of Cdc42 during tubule formation by maintaining apical cell polarity. It also showed that Cdc42 controls cell specification non cell-autonomously by favoring the proper microenvironment necessary for regulation of cell fate choices. Altogether, these findings demonstrate the intricate link between apical-basal polarity, morphogenesis, and cell fate during tubulogenesis. In line with Cdc42 and RhoA playing opposing roles during pancreatic development, mutual antagonism between Rac1 and RhoA activities has also been shown to control cell shape and curvature of the invaginating epithelium in the developing mouse lens.³⁶

Interestingly, epithelial-specific deletion of Cdc42 has recently highlighted the essential role of Cdc42 for branching morphogenesis during lung development.³⁷ Cdc42-deficient pups died at birth due to respiratory failure. At E14.5, the developing lung displayed enlarged, dilated, and disorganized tubules and/or peripheral buds together with disruption in branching morphogenesis accompanied by a decrease in cell proliferation. Similar to the Cdc42-null embryos,¹¹ the expression of apical membrane proteins including ZO-1 and Par3 was severely decreased; cortical F-actin was disrupted while junctional E-cadherin was ectopically localized at the apical surface of the epithelium. In addition, the basolateral marker discs large homolog 1 (DLG1) was absent or abnormally distributed in the epithelium. Thus, depletion of Cdc42 led to a loss of tight junctions, adherens junctions, and overall cell-cell adhesion in the developing lungs, resulting in large intercellular spaces and formation of large cysts with detached epithelia. In addition, Cdc42 was found to be essential for the proper positioning of proliferative epithelial cells and consequently for the formation and maintenance of the respiratory tract during lung morphogenesis.

Kidney development has been extensively studied over the past years and has served as a model to study epithelial-mesenchymal interactions, epithelial cell polarization, and branching morphogenesis.³³ Cdc42 has recently been shown to be important for normal podocyte physiology.³⁸ In mice lacking Cdc42 in podocytes, the foot processes of developing podocytes displayed abnormal junctional complexes. Both podocin and nephrin, two podocyte markers, as well as cell polarity proteins were downregulated within the glomeruli. Podocyte-specific deletion of Cdc42 led to elevated albumin in the urine of newborn pups, severe proteinuria at postnatal day 12 and ultimately led to death due to renal failure at 2 weeks, very similar to human congenital nephropathy. Surprisingly, and in contrast to Cdc42, removal of either Rac1 or RhoA in podocytes had no obvious effect on kidney development.³⁸ Additionally, specific deletion of Cdc42 in the cap mesenchyme led to severe defects in kidney development, with hypoplastic kidneys and empty bladders, indicating a dysfunction in the nephron.³⁹ Furthermore, an absence of convolution in glomeruli and renal epithelia indicated a severe defect in nephron morphogenesis. There was also a reduction in glomerular number and proximal tubules in Cdc42-deficient kidneys. Interestingly, the loss of Cdc42 in kidney phenocopied the loss of the protein Yap, an important transcriptional co-activator shown to be essential for nephrogenesis, suggesting that Yap and Cdc42 act together to regulate a genetic program necessary for kidney development. Taken together, it appears that Cdc42 plays a crucial role during kidney development, in particular during podocyte polarization and renal tubulogenesis.

In contrast to the complex morphological structure of the kidney, the intestinal tube is relatively simple. It consists of a sheet of epithelial cells folded to form the crypts (invasions) and villi (protrusions) surrounding the intestinal lumen. Four differentiated cell types reside within the epithelium: the goblet cells, the enteroendocrine cells, the Paneth cells, and the enterocytes.⁴⁰ Additionally, the crypt regions contain stem cells and transit amplifying progenitor.⁴⁰ When Cdc42 was specifically depleted in the mouse intestinal epithelium,^41,42 the mice developed large intracellular vacuolar structures that contain microvillus inclusion bodies in enterocyte cells, a phenotype strikingly similar to the human microvillus inclusion disease (MVID), which is one of the most severe congenital intestinal disorders leading to death within the first year. In this context, Cdc42 ablation resulted in a reduction in the stem cell population that was accompanied with a reduced number of Paneth cells. Furthermore, proliferating stem and progenitor cells increased in mutant mice, resulting in increased intestinal crypt depth. Elevated stem and progenitor cell migration along the villi also caused defects in the apical junction orientation, which impaired intestinal epithelium polarity and may be responsible for the defective intestinal permeability. Interestingly, Sakamori and colleagues have also implicated a functional interaction between Cdc42 and Rab8 as critical for intestine stem cell division, survival, and differentiation. Cdc42-deficient mice have impaired Rab8 activation and inhibited Rab8 vesicle trafficking to the midbody.⁴¹ Moreover, Rab8 was necessary for Cdc42 activation in the intestinal epithelium. Altogether, these studies using Cdc42-deficient mouse models in the intestine confirmed the critical role of Cdc42 in the regulation of cell proliferation, polarity, migration, and differentiation of intestinal cells.

Overall, the importance of the small GTPase Cdc42 as a master regulator of cell polarity is clearly highlighted by these different Cdc42-deficient mouse models in specific organs in which the dynamic control of cell polarity is essential for the proper development of epithelial cell morphogenesis and tubulogenesis.

CNS development

Communication between the billions of neurons in the CNS is mediated by the synapse, a specialized point of contact between the axon of one neuron and the receiving dendrite of another neuron, where neurotransmitters are released. The precise connection between neurons is tightly regulated during development of the CNS. After generation of the three germ layers during gastrulation, the ectoderm gives rise to the neural plate that folds and fuses at the dorsal midline in a process called neurulation.^14,19 At this stage the CNS consists of a neural tube containing neuroepithelial cells that give rise to all neurons and glial cells of the mature nervous system.⁴³ As described above, Rac1 has been shown to be essential for neural tube closure.²¹ After neuroepithelial cells have divided several rounds, they generate the stem cells of the nervous system that will remain into adulthood, the radial glial cells. These cells proliferate by both symmetrical and asymmetrical divisions within the ventricular zone (VZ) to produce postmitotic neurons. These radial glial cells will eventually divide asymmetrically to produce transit amplifying progenitors that will populate a secondary proliferative region, the subventricular zone (SVZ). In addition, radial glial cells that extend across the developing neural tube serve as a scaffold for the newly born cortical projection neurons migrating from the VZ to the cortical plate. In contrast, newly born interneurons within the ganglionic eminences migrate tangentially to the cortex using extracellular cues to guide them.⁴⁴ To reach their final destination, axons travel long distances and navigate through a very complex environment by interpreting many different extracellular signals. The growing axon achieves this important task using a specialized dynamic actin-rich tip: the neuronal growth cone. Once the axon has reached its target, synaptogenesis takes place. From neural tube closure to neuronal migration, neurite outgrowth, axon guidance, and synaptogenesis, Rac1, Cdc42, and recently RhoA have all been shown to play important roles during these defined processes of CNS development.

Cdc42 is expressed at the apical side of VZ progenitors and in postmitotic neurons but is absent in most intermediate progenitors dividing at basal positions.⁴⁵ Interestingly, when Cdc42 was specifically depleted before the onset of neurogenesis in the developing cortex, it led to an increase in basal mitosis with an apical loss of the Par complex and adherens junctions, which resulted in increasing defects in apically directed interkinetic neuronal migration. Loss of the aPKC/Par3 complex and apical cell-cell contacts in Cdc42-depleted cells led to a complete disruption of adherens junctions by mid-neurogenesis. This defect in cell-cell adherence combined with an increased basal mitosis ultimately led to a higher rate of neuron generation and an increase in cortical wall thickness during early corticogenesis.^45,46 Therefore, Cdc42 is required to activate the Par complex and to maintain adherens junctions for regulating mitosis and VZ progenitor fate. In a separate study, conditional deletion of Cdc42 in the developing forebrain displayed a similar phenotype.⁴⁷ Indeed, the apical localization of Par6, aPKC, E-cadherin, β-catenin, and Numb proteins were all abolished as early as E10.5 in the neuroepithelium of Cdc42-deficient embryos, which also exhibited a thicker cortical wall and an increase in basal mitosis. Furthermore, there was a lack of invagination of the roof plate that normally separates the two lateral ventricles, leading to a phenotype characteristic of holoprosencephaly in a sonic hedgehog-independent pathway.⁴⁷ Although Rac1 also interacts with the Par-aPKC protein complex,⁴⁸ mice lacking Rac1 expression in the developing forebrain did not present the holoprosencephaly phenotype,⁴⁷ demonstrating a distinct role for Cdc42 in the morphogenesis of the forebrain. Similar to Cdc42, the removal of RhoA in the forebrain and midbrain neural progenitors led to disorganized apical domains in the VZ with rosette-like structures, disruption of adherens junctions, and dysplasia of the brain.⁴⁹ Notably, RhoA-deficient neural progenitors showed increased cell proliferation, reduced cell cycle exit, and higher expression of hedgehog target genes. In each RhoA-deficient mouse model, there was the appearance of exencephaly-like protrusions in the midbrain and forebrain at later stages of development.⁴⁹ Furthermore, loss of RhoA and Cdc42 in the VZ of the medial ganglionic eminence demonstrated defects in the migration of cortical interneurons.⁵⁰ However, deletion of both genes in the SVZ did not lead to major defects in cortical interneuron migration,⁵⁰ suggesting that other Rho GTPases may compensate for the loss of RhoA and Cdc42 in the ganglionic eminences of the SVZ. Additionally, in a recent study from Cappello et al. (2012), conditional removal of RhoA in the developing cerebral cortex led to the formation of a double cortex, highlighting the importance of RhoA in the regulation of radial glial-mediated neuronal migration during development of the CNS.⁵¹ Moreover, the tissue-specific deletion of RhoA in the neuroepithelium of the developing spinal cord has also highlighted the critical role of RhoA in the maintenance of apical adherens junction integrity, organization of the neuroepithelium in the developing spinal cord and left-right locomotor behavior.^52,53

Three independent groups have investigated the effect of a forebrain-specific deletion of Rac1 beginning at E9.5,^54-56 after neural tube closure. This conditional Rac1 deletion led to late embryonic lethality. Overall, Rac1-deficient embryos had smaller forebrain, olfactory bulb, cortex, and striatum with enlarged ventricles.^54-56 There was a lack of separation between the SVZ/VZ⁵⁴ and a decrease in proliferation of neural progenitor cells that correlated with an increase in cell cycle exit and apoptosis during early corticogenesis, which resulted in microcephaly.^54,56 In the developing cortex, the increase in cell cycle exit was more dramatic within the intermediate progenitor cells occupying the SVZ.⁵⁴ These studies strongly demonstrated the importance of Rac1 in the regulation of proliferation and differentiation of progenitor cells during cortical neurogenesis, in particular within the intermediate progenitor population. Since Rac1 is an important signaling mediator of many classic axon guidance cues, including the netrin/DCC signaling pathway,^57-60 it was not so surprising to observe a complete absence of the anterior commissure in the brain of Rac1-deficient embryos.^55,61 Furthermore, the corpus callosal and the hippocampal commissural axons failed to cross the midline in Rac1-deficient embryos while there was also projection and defasciculation defects of the corticothalamic and thalamocortical axons. However, no defect in the axonal outgrowth of telencephalic neurons was observed in Rac1-deficient embryos but a reduction in cortical interneurons within the cortex indicated a defect in tangential migration.^55,62 Intriguingly, cultured Rac1-deficient cortical neurons displayed an increase in the number of neurites compared with control neurons.⁵⁵ These in vivo and in vitro data are certainly in marked contrast to many previous in vitro studies reporting a role for Rac1 in neurite outgrowth using dominant-negative mutant expression.^58,63 The recent study by Vaghi et al. (2012) investigating Rac1 and the neural specific Rac3 in the development of cortical and hippocampal GABAergic neurons suggests that both Rac1 and Rac3 contribute synergistically to the regulation of cortical and hippocampal neuronal migration and differentiation, which may explain the epileptic phenotype of the Rac1/Rac3 double mutant mice.^64,65 Lastly, several defects were also observed in the development of the cerebellum lacking Rac1.⁶⁶

Taken together, these studies clearly establish fundamental roles of RhoA, Rac1, and Cdc42 during early development of the CNS.

Limb development

Conditional deletions of both Rac1 and Cdc42 in mouse cartilage and limb bud mesenchyme have recently demonstrated the essential role of Rac1 and Cdc42 during chondrogenesis and limb bud morphogenesis. Indeed, chondrocyte-specific deletion of Rac1 led to multiple developmental defects.⁶⁷ Pups displayed severe kyphosis at birth and died shortly after birth due to respiratory distress. They were smaller due to growth retardation, with a smaller skeleton, shorter tibias, femurs, and skull, and a deformed rib cage.⁶⁷ At embryonic day 16.5, when ossification is normally underway, Rac1-deficient mice only displayed hypertrophic cartilage. Thus, in Rac1-deficient cartilage, primary and secondary ossification was delayed.⁶⁷ Furthermore, these mice had disorganized growth plates with chondrocytes of abnormal shape and size. Moreover, Rac1-deficient chondrocytes showed defects in adhesion and spreading on collagen and fibronectin with a disorganized actin cytoskeleton, reduced cell proliferation, increased apoptosis, and deregulation of key gene regulator of cartilage development such as Indian hedgehog (Ihh).⁶⁷ On the other hand, mice with a Rac1 deletion in the embryonic limb bud ectoderm displayed disrupted canonical Wnt signaling leading to severe truncations of the limb, which phenocopied β-catenin deletion.⁶⁸ Interestingly, conditional deletion of Rac1 in the limb bud mesenchyme also led to skeletal deformities in the autopod and soft tissue syndactyly due to a complete absence of interdigital programmed cell death (PCD).⁶⁹ Similarly, deletion of Cdc42 in the limb bud mesenchyme led to shorter limbs and several skeletal abnormalities.⁷⁰ Specifically, there was a bifurcated sternal bar, malformation or absence of the xiphoid process, abnormal calcification of the craniums, and the secondary palate failed to fuse. Notably, the cleft palate defect observed in Cdc42-deficient mice resulted in a suckling disorder in newborn pups leading to death by malnourishment.⁷⁰ In addition, the growth plate displayed a wider hypertrophic cartilage and the reduction in Col10 and Mmp13 expression suggested an inhibition of chondrocyte differentiation.⁷⁰ Furthermore, syndactyly was caused by fusion of metacarpals and defective interdigital PCD, similar to the Rac1-deficient mouse. Overall, these mouse models strongly demonstrate a crucial role for Rac1 and Cdc42 in chondrogenesis and interdigital PCD during limb development. Of great interest, recent findings have also implicated Rac1/Cdc42 regulation in the Adams-Oliver syndrome, a rare human developmental disorder characterized by the congenital absence of skin (aplasia cutis congenita) in combination with limb abnormalities,^71,72 phenotypes that are very similar to those of the Rac1/Cdc42 mouse models during limb development.

Concluding Remarks

From the early stages of embryonic development, Rho GTPases are implicated in the regulation of cell proliferation, survival, migration, polarization and differentiation. Consequently, dysregulation of these cellular functions results in severe developmental defects. The recent in vivo mouse models of RhoA, Rac1, and Cdc42 not only lead to a better insight into Rho family functions but also provide invaluable animal models for human diseases. Clearly, we are still only beginning to appreciate the complexity and many functions of the Rho family of small GTPases in mammals. In vivo studies of the large number of Rho regulators in specific cell types and tissues can only add to the complexity of Rho signaling pathways but will also contribute to increasing our knowledge of Rho functions in a more physiological manner and in specific cellular contexts.

Disclosure of Potential Conflicts of Interest

No potential conflict of interest was disclosed.

Acknowledgments

We acknowledge the limited scope of this review and, as such, apologize that we were unable to reference more works related to this field. We are grateful to Jonathan DeGeer for critically reading this manuscript.

Funding

This work was supported by grants from CIHR MOP-14701. N.L.V. is a William Dawson Chair and FRSQ Chercheur National. P.M.D. is supported by an FRSQ doctoral's Training Award.