Rac1 plays an essential role in axon growth and guidance and in neuronal survival in the central and peripheral nervous systems

Background Rac1 is a critical regulator of cytoskeletal dynamics in multiple cell types. In the nervous system, it has been implicated in the control of cell proliferation, neuronal migration, and axon development. Results To systematically investigate the role of Rac1 in axon growth and guidance in the developing nervous system, we have examined the phenotypes associated with deleting Rac1 in the embryonic mouse forebrain, in cranial and spinal motor neurons, in cranial sensory and dorsal root ganglion neurons, and in the retina. We observe a widespread requirement for Rac1 in axon growth and guidance and a cell-autonomous defect in axon growth in Rac1−/− motor neurons in culture. Neuronal death, presumably a secondary consequence of the axon growth and/or guidance defects, was observed in multiple locations. Following deletion of Rac1 in the forebrain, thalamocortical axons were misrouted inferiorly, with the majority projecting to the contralateral thalamus and a minority projecting ipsilaterally to the ventral cortex, a pattern of misrouting that is indistinguishable from the pattern previously observed in Frizzled3−/− and Celsr3−/− forebrains. In the limbs, motor-neuron-specific deletion of Rac1 produced a distinctive stalling of axons within the dorsal nerve of the hindlimb but a much milder loss of axons in the ventral hindlimb and forelimb nerves, a pattern that is virtually identical to the one previously observed in Frizzled3−/− limbs. Conclusions The similarities in axon growth and guidance phenotypes caused by Rac1, Frizzled3, and Celsr3 loss-of-function mutations suggest a mechanistic connection between tissue polarity/planar cell polarity signaling and Rac1-dependent cytoskeletal regulation.


Background
Small GTPases act as molecular toggle switches, cycling between active GTP-bound and inactive GDP-bound states to regulate a wide variety of intracellular processes in response to extracellular signals [1]. Rac1, a widely expressed member of the Rho GTPase family, controls actin dynamics and plays an essential role in cell proliferation and migration in multiple contexts [2].
In mice, constitutive deletion of Rac1 leads to early embryonic lethality [3]. However, selective deletion of Rac1 in the anterior neural tube starting at embryonic day (E)8.5 permits survival until birth, and at late gestation, these embryos show an absence of the anterior commissure, a failure of the corpus callosum and hippocampal commissure to cross the midline, and defects in corticothalamic and thalamocortical projections [4]. Loss of Rac1 also leads to a reduction in the proliferation and survival of neural progenitors and a severe defect in tangential migration of inhibitory interneurons [5,6]. Selective deletion of Rac1 in the cerebral cortex blocks midline crossing of axons in the corpus callosum and anterior commissure, but there is little or no effect on the development of the corticospinal, corticothalamic, and thalamocortical tracts [7].
The tissue polarity/planar cell polarity (PCP) signaling system is one of several cell-cell signaling systems that control axon guidance. PCP signaling is conserved from insects to mammals, and it has been most intensively studied in the context of epithelia, where it establishes and maintains polarity within the plane of the epithelium relative to global anatomic structures. PCP signaling appears to converge on the cytoskeleton, and a variety of experiments in the context of convergent extension and inner ear sensory hair cell development suggest the direct or indirect involvement of small GTPases as effectors of PCP signaling [8][9][10][11][12].
Mutations in either of two core PCP genes, Frizzled (Fz)3 and Celsr3, cause profound and virtually identical defects in axon guidance in mice, including defects in corticothalamic, thalamocortical, nigrostriatal, and spinal sensory axon tracts [13][14][15][16][17][18][19]. We recently conducted a comprehensive analysis of axonal development in both the central and peripheral nervous systems in Fz3 −/− mice [17,18], and we were intrigued by the potential similarities between the Fz3 −/− axon guidance defects and the axon guidance defects reported by Chen et al. [4] and Kassai et al. [7] in forebrain-specific Rac1 knockout mice. Here we present a comprehensive analysis of axonal development in conditional mutants of Rac1, including the first analysis of the Rac1 mutant phenotype in sensory and motor neurons. We have used the same methodologies that were used in our earlier studies of Fz3 mutants, and the resulting comparisons of Fz3 and Rac1 loss-of-function phenotypes reveal several striking similarities.

Results
Defects in forebrain axon tracts in Foxg1-Cre;Rac1 CKO/− embryos To compare axon growth and guidance phenotypes produced by loss of Rac1 with our earlier studies of Fz3 lossof-function mutations in the forebrain, we immunostained coronal sections of E18.5 Foxg1-Cre;Rac1 CKO/− and control Rac1 CKO/+ brains for neurofilament and compared them to an analogous set of coronal sections from Foxg1-Cre;Fz3 CKO/− embryos, shown in Figures 1 and 2 of Hua et al. [18]. (Foxg1-Cre is expressed prior to E10.5 in the anterior neural tube, leading to Cre-mediated recombination in all or nearly all CNS cell types in the developing telencephalon). Figure 1a-c' shows that Foxg1-Cre;R-ac1 CKO/− forebrains are smaller than littermate controls and they exhibit axon growth and guidance defects that closely resemble the defects seen in Foxg1-Cre;Fz3 CKO/− forebrains (compare to Figure 2B-B" of Hua et al. [18]). In particular, Fig. 1a'-c' shows that in Foxg1-Cre;Rac1 CKO/− forebrains the lateral ventricles are enlarged, axons are missing from the subcortical zone, the corpus callosum is misrouted away from the midline, axons are missing from the striatum, and the anterior commissure is missing. Among thalamocortical axons, none traverses the striatum to innervate the cortex. Instead, a majority of these axons are misrouted so that they project to the contralateral thalamus, resulting in a dense U-shaped axon tract along the inferior border of the thalamus. The remaining minority of thalamocortical axons are misrouted to the ventral ipsilateral cortex. The misrouting of thalamocortical axons in Foxg1-Cre;Rac1 CKO/− brains is indistinguishable from the phenotype observed in Foxg1-Cre;Fz3 CKO/− brains (compare Fig. 1c' with Figures 1 and 2 in Hua et al. [18]).
Inhibitory interneurons in the cerebral cortex are born in the medial ganglionic eminence and migrate tangentially across the cortex between E12 and E16 [20]. Using Lhx6.1 and GAD67 as markers at E16.5 and E18.5, Chen et al. [4] observed a nearly complete absence of interneurons in the cerebral cortex in Foxg1-Cre;Rac1 CKO/CKO embryos. Using calbindin as a marker for a subset of interneurons, Fig. 1d-g' confirms those results by showing that, at E13.5, migrating interneurons are present in the control cortex but are absent from the Foxg1-Cre;R-ac1 CKO/− cortex.
To test the possibility that combined haplo-insufficiency for Fz3 and Rac1 might reveal a genetic interaction, we compared Rac1 CKO/+ vs. Foxg1-Cre;Rac1 CKO/+ ;Fz3 +/− coronal sections of E18.5 brains immunostained for neurofilament. Heterozygous deletion of either Fz3 or Rac1 has no detectable effect on brain size or the pattern of neurofilament staining ( [13,14]; Fig. 1a-c, h-j). However, heterozygous deletion of both Fz3 and Rac1 in the forebrain (Foxg1-Cre;Rac1 CKO/+ ;Fz3 +/− ) results in an absence of the corpus callosum and a brain that is reduced to a size intermediate between wild type (WT) controls and In the Foxg1-Cre;Rac1 CKO/+ ;Fz3 +/− brain, the axons that would normally comprise the corpus callosum collect adjacent to the midline, a defect that is seen in a variety of axon guidance mutants [21]. This genetic interaction is suggestive of a mechanistic link between Rac1 and Fz3 function, although it is also possible that it represents an additive effect in a system that is sensitive to genetic perturbation.
Survey of peripheral nerve defects in Olig2-Cre;Rac1 CKO/− embryos To broadly survey peripheral nerve phenotypes caused by loss of Rac1, we visualized axons in whole-mount E11.5 Olig2-Cre;Rac1 CKO/− embryos by immunostaining for neurofilament ( Fig. 2a-d'). Olig2-Cre is expressed in many cranial and spinal motor neurons beginning at E9.5 [22][23][24][25]. This analysis revealed a severe thinning of the VIth, XIIth, C1 segmental, and phrenic nerves in Olig2-Cre;Rac1 CKO/− embryos, a set of nerves that partially overlaps the set that are thinned in Fz3 −/− and Olig2-Cre;Fz3 CKO/− embryos (compare Fig. 2a-c' to Figures 1 and 6 in Hua et al. [17]). Whole-mount neurofilament staining of Olig2-Cre;Rac1 CKO/− fore-and hindlimbs at E13.5 shows a severe reduction in the diameter of the dorsal nerve and a more modest reduction in the diameter of the ventral nerve in the hindlimb, with only a modest reduction in the diameter of both dorsal and ventral nerves in the forelimb (Fig. 2e-h'). Among these four nerves, the dorsal nerve in the hindlimb also showed the greatest thinning in Fz3 −/− and Olig2-Cre;Fz3 CKO/− embryos [17].
To distinguish motor vs. sensory axon contributions to the peripheral nerve defects, the Hb9-GFP reporter was used to selectively visualize cranial and spinal motor axons in Olig2-Cre;Rac1 CKO/− and control Rac1 CKO/+ embryos (Fig. 3). This analysis revealed a complete or nearly complete loss of immunostaining of coronal brain sections at E18.5 shows an enlarged lateral ventricle (encircled by a white line), absence of multiple axon tracts, and misrouting of thalamocortical axons in the embryonic Foxg1-Cre;Rac1 CKO/− brain. In the Foxg1-Cre;Rac1 CKO/− brain, corticospinal axons are partially obscured by misrouted thalamocortical axons. Numbered white arrows (control) and arrowheads (mutant) point to corresponding axon tracts: 1, axons in the cortical intermediate zone; 2, corpus callosum; 3, axons in the striatum; 4, anterior commissure; 5, thalamocortical axons; and 6, corticospinal tract. N = 2 for each genotype. In this and all other figures, images are shown from a representative embryo of the indicated genotype. Scale bar, 500 μm. d-g' Calbindin immunostaining of E13.5 sections shows the absence of calbindin + interneurons in the Foxg1-Cre;Rac1 CKO/− neocortex. e, e', g, g' are enlarged views of boxed regions in (d, d', f, f'), respectively. In the WT neocortex, interneurons (green arrows) exhibit elongated bipolar morphology parallel to the plane of the cortex, indicative of tangential migration. In the Foxg1-Cre;Rac1 CKO/− forebrain, calbindin + interneurons are seen in the ventral telencephalon but are not found in the neocortex. The lateral ventricle is already enlarged by this stage. Faint background immunostaining is present in the vasculature. CGE caudal ganglionic eminence, LGE lateral ganglionic eminence, MGE medial ganglionic eminence. N = 3 for each genotype. Scale bars: d', 500 μm; g', 100 μm. h-j' Neurofilament (NF) immunostaining of coronal Rac1 CKO/+ vs. Foxg1-Cre;Rac1 CKO/+ ;Fz3 +/− brain sections at E18.5. The corpus callosum is indicated by a white arrow in (h, i). The corresponding non-crossing fibers are indicated by a white arrowhead in (h', i'). N = 2 for each genotype. Scale bar, 500 μm. k Planes of sections in (a-g') the motor component of the VIth and XIIth nerves (Fig. 3a vs. 3a' and Fig. 3d vs. 3d') and a massive reduction in the number of motor neurons in the VIth, Xth, and XIIth cranial nerve nuclei at E12.5 in Hb9-GFP;Olig2-Cre;Rac1 CKO/− embryos (Fig. 3b vs. 3b' and Fig. 3c vs. 3c'). We did not assess motor axons within the Xth cranial nerve, but we presume that their numbers are correspondingly reduced. Axons within the VIIth nerve were Hb9-GFP negative and were unaffected in Olig2-Cre;Rac1 CKO/− embryos (Fig. 3b vs. 3b'). In the E12.5 Hb9-GFP;Olig2- Quantitative analysis of limb motor axon defects and spinal motor neuron loss in Olig2-Cre;Rac1 CKO/− embryos In earlier work, we quantified a distinctive axon stalling phenotype at the base of the limbs in Fz3 −/− embryos by measuring dorsal and ventral nerve diameters in whole mounts of neurofilament-stained limbs at E12.5 and E13.5 [17]. For the dorsal nerves, measurements were made at locations proximal and distal to the point of axon stalling (see Figure 4 in Hua et al. [17]). We have performed the same analysis on E12.5 and E13.5 Olig2-Cre;Rac1 CKO/− and WT littermates (Fig. 4a, b). At both ages, Olig2-Cre;R-ac1 CKO/− fore-and hindlimb ventral nerve diameters were reduced, on average, to 65-73 % of the diameters of littermate controls. Similarly, the Olig2-Cre;Rac1 CKO/− fore-and hindlimb dorsal nerve diameters proximal to the Fz3 −/− stalling point were reduced, on average, to 59-69 % of the diameters of littermate controls. Interestingly, the Olig2-Cre;Rac1 CKO/− dorsal nerve diameters distal to the Fz3 −/− stalling point were reduced, on average, in the forelimb to 47-53 % of the control diameters and in the hindlimb to 16-21 % of the control diameters. These data indicate that in Olig2-Cre;Rac1 CKO/− embryos there is a severe axon stalling defect in the hindlimb dorsal nerve and a moderate axon growth defect in the three other limb nerves that were quantified in Fig. 4a, b. These patterns of nerve diameter reduction closely match the patterns observed in Fz3 −/− limbs, both qualitatively and quantitatively.
Our earlier analysis of Fz3 −/− embryos showed that axon stalling leads to a secondary loss of spinal and cranial motor neurons. In the late E11.5 lumbar spinal cord, motor neuron loss was largely limited to the lateral division of the lateral motor column (LMC; Figure 5 in Hua et al. [17]). A similar analysis of Olig2-Cre;Rac1 CKO/− lumbar spinal cords showed an increase in the number of apoptotic motor neurons at E11.5, as determined by immunostaining for cleaved caspase3 (Fig. 4c), and a reduction in motor neurons in both the medial and lateral divisions of the lateral motor column (LMC M and LMC L ) relative to littermate controls ( Fig. 4d-f). Cell loss in both divisions rather than a single division implies the presence of developmental defects in the Olig2-Cre;Rac1 CKO/− spinal cord that are broader than the defects in the Fz3 −/− spinal cord.
To assess the autonomy of the motor axon growth defects in Olig2-Cre;Rac1 CKO/− embryos, dissociated cells were cultured from E11.5 spinal cords of Hb9-GFP;O-lig2-Cre;Rac1 CKO/− and littermate control Hb9-GFP;O-lig2-Cre;Rac1 CKO/+ embryos, and among GFP + neurons, axon lengths were quantified after 2 days in culture. In a sample of 51 Hb9-GFP;Olig2-Cre;Rac1 CKO/− and 69 control axons, the axons of Rac1-deficient motor neurons had a mean length approximately one third of the length of the control group (P < 10 −11 ) (Fig. 4g). Thus, loss of Rac1 impairs axon growth in a cell-autonomous manner. In dissociated neuronal cultures, a similar cell-autonomous axon growth defect was observed among Rac1-deficient cerebellar granule cells [26], but no axon growth defect was observed among Rac1-deficient cerebral cortical pyramidal neurons [4]. The results of Chen et al. [4] could reflect the presence of compensating Rho family members in cortical neurons.
Wnt1-Cre is expressed in migrating neural crest cells, including those that form the DRG and cranial sensory ganglia [27]. As Rac1 acts in both cell migration and axon guidance, it would not be surprising to find that defects in one or both of these processes affect sensory neuron development. To assess Rac1 effects at the earliest stages of DRG development, we compared the DRG in coronal Axon tract and peripheral nerve defects in Wnt1-Cre;Rac1 CKO/− embryos. a-g' Neurofilament (NF) immunostaining of whole-mount E11.5 embryos shows that in Wnt1-Cre;Rac1 CKO/− embryos, multiple axon tracts are aberrant including the following: diffuse axonal projections in the hindbrain and the IIIrd and IVth nerves (green arrows in (b) vs. red arrows in (b')), the Vth nerve (green arrows in (c) vs. red arrows in (c')), the Xth (green arrow in (d) vs. red arrow in (d')) and XIIth nerves (green arrowhead in (d) vs. red arrowhead in (d')), multiple nerves in the forelimb (green arrows in (e) vs. red arrows in (e')), and ventral axons from the DRG (green arrows in (f) vs. red arrows in (f')). The sympathetic chain ganglia are reduced in size (green arrows in (g) vs. red arrows in (g')). To show the individual hindbrain axons, the neurofilament signal intensity was increased in the (b) and (b') images. d-e', g, g' are maximum intensity projections of a subset of Z stacks containing the entire nerves or chain ganglia. N = 3 for each genotype. Scale bars: a', 1 mm; b'-g', 100 μm sections of Wnt1-Cre;Rac1 CKO/− and littermate control embryos at E11.5. This comparison revealed a dramatic reduction in the size of the DRG and number of DRG neurons in the brachial and thoracic regions in Wnt1-Cre;Rac1 CKO/− embryos, but there was no difference between mutant and control embryos in the expression of DRG markers (Brn3a, Islet1/2, Sox10), and there was little or no difference between mutant and control embryos in the relative density of DRG neurons with activated cas-pase3 at this stage (Fig. 6a-d).
At present, it is not clear whether the primary defect in Wnt1-Cre;Rac1 CKO/− DRG development is in neural crest migration, DRG axon growth and/or guidance, or some combination of the two, since we have not identified the missing neural crest cells. However, it is apparent that DRG neurons are affected early in their development, since they are already missing from the DRG at E11.5.
Defects in retinal ganglion cell axon guidance and survival in Pax6-alpha-Cre;Rac1 CKO/− and Six3-Cre;Rac1 CKO/− mice The growth of retinal ganglion cell (RGC) axons to their central targets presents a favorable system in which to study long-range axon guidance [28]. Eliminating Rac1 predominantly in the peripheral retina beginning at~E9.5 with the Pax6-alpha-Cre driver [29] does not affect the viability or overall health of the embryo or postnatal mouse and thus permits a phenotypic analysis in  adulthood. Figure 7a shows that, in adult Pax6-alpha-Cre;Rac1 CKO/− mice, the diameter of the optic nerve and the thickness of the inner retina (in the retinal periphery) are both reduced, consistent with the expression of the Pax6-alpha-Cre transgene in the periphery but not the central~50 % of the retina. Within the thinned inner retina, the pattern of lamination and the locations and structures of major cell types-including cholinergic, GABAergic, dopaminergic, calbindin-expressing, and calretinin-expressing amacrine cells, and horizontal cells, blood vessels, and astroctyes-appear essentially normal (Fig. 7a). These data closely resemble the phenotype that results from the selective loss of the majority of RGCs, as observed, for example, in Brn3b −/− , Math5 −/− , and Brn3b −/− ;Brn3c −/− mice [30][31][32][33][34].
Intraocular injection of fluorescent cholera toxin subunit B in adult Pax6-alpha-Cre;Rac1 CKO/− mice shows an essentially normal pattern of innervation of RGC axon targets, including the suprachiasmatic nucleus (SCN), lateral geniculate nucleus (LGN), olivary pretectal nucleus (OPN), pretectal area (PA), superior colliculus (SC), and the medial terminal nucleus (MTN) (Fig. 7b). These data are consistent with the expectation (based on the pattern of Pax6-alpha-Cre expression) that RGCs within the central~50 % of the retina should retain Rac1 function. To selectively visualize the axons of RGCs that had lost Rac1, we introduced a Cre-activated alkaline phosphatase (AP) knock-in allele that is expressed exclusively in RGCs (Brn3b CKOAP ; [35]). In the peripheral retina, Pax6-alpha-Cre expression is predicted to lead to Cre-mediated recombination of both Rac1 CKO and Brn3b CKOAP , simultaneously eliminating Rac1 and activating expression of AP in RGCs. As seen in the left column of Fig. 7c, control brains from adult Pax6-alpha-Cre;Brn3b CKOAP/+ mice show AP histochemical staining in the same set of retinorecipient areas that were labeled by intraocular injection with cholera toxin subunit B (Fig. 7b). In contrast, the right column of Fig. 7c shows that brains from adult Pax6-alpha-Cre;Brn3b CKOAP/+ ;Rac1 CKO/− mice exhibit almost no AP staining in retino-recipient regions. We ascribe the occasional presence of focal zones of AP activity in retino-recipient territories to rare RGCs in which the AP reporter was activated by Cre-mediated recombination but the Rac1 CKO allele was not recombined. These data indicate that loss of Rac1 leads to a severe deficiency in RGC axon growth, guidance, and/or RGC survival. The defect appears to be cell-autonomous since it is not corrected by neighboring Rac1 +/− RGCs that develop normally.
A more severe retinal phenotype is observed when Cremediated deletion of Rac1 is driven by Six3-Cre, a transgene that is expressed in all cells in the neural retina beginning at~E11 [36]. In adult Six3-Cre;Rac1 CKO/− retinas, the optic nerve is reduced to <20 % of the WT diameter and the inner nuclear layer and inner plexiform layer are dramatically thinned (Fig. 8a). A variety of retinal cell types are present but with substantially distorted morphologies and reduced abundances (Fig. 8a). These include cholinergic, GABAergic, dopaminergic, calbindin-expressing, and calretinin-expressing amacrine cells. Horizontal cells, marked by neurofilament and calbindin expression, appear to be almost completely absent. Neurofilament immunostaining shows accumulation of neurofilament at the inner face of the retina, and glial fibrillary acidic protein (GFAP) immunostaining of retina flat mounts shows enhanced staining in a pattern consistent with reactive astrocytes. By DAPI staining, the numbers of cells in the ganglion cell layer and the inner nuclear layer are reduced to <50 % of the WT level, but the outer nuclear layer is largely unaffected. Tracing adult RGC axons by intraocular injection of fluorescent cholera toxin subunit B in adult Six3-Cre;Rac1 CKO/− retinas shows an almost complete absence of RGC axons in retino-recipient regions (Fig. 8b). At P0, the number of RGCs is similar to the number in WT control retinas, as determined by immunostaining for transcription factor Brn3a (data not shown); earlier work demonstrated that RGC cell death secondary to axon pathfinding defects occurs at~P10 [37].  Under this interpretation, the observations in Figs. 7 and 8 imply that loss of Rac1 in a subset of RGCs (Pax6-alpha-Cre;Rac1 CKO/− ) leads to axon growth defects and eventual death in this RGC subset and the secondary loss of a relatively small fraction of amacrine cells, whereas loss of Rac1 in the vast majority of RGCs in Six3-Cre;Rac1 CKO/− retinas leads to axon growth defects and eventual death in nearly all RGCs and the secondary loss of a large fraction of amacrine cells. In Six3-Cre;Rac1 CKO/− retinas, in which Cre is expressed in retinal astrocytes, we also observe defects in postnatal retinal vascularization, which involves endothelial cell growth on an astrocyte scaffold, and this could account for part of the axon development and inner retinal phenotypes. However, these considerations do not apply to the Pax6-alpha-Cre;Rac1 CKO/− retina, as Pax6alpha-Cre is not expressed in retinal astrocytes [38].

Discussion
This study provides a systematic analysis of the consequences of Rac1 deletion in neurons in the telencephalon, cranial and spinal motor columns, sensory ganglia, and retina. In each of these neuronal populations, we observe defects in axon growth and/or guidance with loss of Rac1. Among motor neurons and retinal ganglion cells, these defects are associated with neuronal loss, consistent with the known dependence of neuronal survival on axon targeting. Additionally, we observe cell migration defects among telencephalic interneurons and a large reduction in the number of DRG neurons at E11.5. This early DRG phenotype presumably reflects a defect in the proliferation and/or migration of DRG progenitors from the neural crest. In contrast with earlier studies that used dominant negative or constitutively active mutants of Rac1 or other Rho family GTPases (for example, [39][40][41][42][43][44]), the loss-of-function approach used here does not have the potential to perturb signaling by closely related small GTPases or their effectors, and therefore, the observed phenotypes can be confidently assigned to Rac1.

The breadth of Rac1 action in nervous system development
The cell biological processes that underlie nervous system development can be roughly divided into seven general classes: (1) cell proliferation; (2) cell migration; (3) axon growth, guidance, and branching; (4) dendritic growth and branching; (5) synaptogenesis; (6) remodeling/plasticity; and (7) cell survival or death. Rac1 and related Rho GTPases play essential roles in each of these processes, as briefly summarized in points (1-7) below.
(1) Rac1 in the embryonic forebrain is required for proliferation of neural progenitors [5,6], and in the adult, it is required for learning-evoked neurogenesis in the hippocampus [45,46]. (2) In the telencephalon, Rac1 is required for tangential migration of interneurons from the medial ganglionic eminence ( [4,47]; this work), as are RhoA and Cdc42 [48]. In the rostral migratory stream, Rac1 is required for neuronal migration [49]. (3) Down-regulation of Rac1 activity by the breakpoint cluster region GTPase activating protein decreases dendritic branching in vivo and in vitro [50], and constitutive activation of Rac1 increases dendritic growth and branching in the cortex and hippocampus [51,52]. (4) In Drosophila, Rac1 and the closely related Rho GTPases Rac2 and Mtl control axon growth, guidance, and branching [8,53], and in the mouse, Rac1 is required widely in the central and peripheral nervous systems for axon growth and guidance ( [4,7]; this work). In the hippocampus, Rac1 is additionally involved in axon pruning [54]. (5) In C. elegans, the reorganization of the actin cytoskeleton and the clustering of synaptic vesicles in response to netrin signaling require Rac1 [55], and in mice, EphB signaling via the GEF Tiam requires Rac1 for excitatory spine remodeling [56]. (6) The synaptic functions of Rac1 continue throughout life, as determined by defects in hippocampal plasticity and impaired spatial memory in mice with defects in Rac1 signaling [57,58]. (7) Finally, Rac1 regulates cell survival by mechanisms that may include both direct and indirect effects. In the ventricular zone and subventricular zone, deletion of Rac1 leads to accelerated cell cycle exit and reduced survival of progenitors, an effect that may reflect Rac1 regulation of cell cycle progression and apoptosis [5,6]. Among cranial and spinal motor neurons and RGCs, cell death may be secondary to defects in axon growth and guidance (this work).

Comparison of Rac1 and Fz3/Celsr3 phenotypes
As noted in the "Background" section, the initial impetus for this study was to compare Rac1 and Fz3 phenotypes. With these two phenotypes now defined in detail using the same methodologies, it is apparent that they share a number of features ( [4,7,13,14,17,18], this work). Most prominently, the distinctive misrouting of thalamocortical axons-with the majority projecting to the contralateral thalamus and a minority projecting to the ventral cortex-is virtually identical in Foxg1-Cre;Rac1 CKO/− , Fz3 −/− , and Celsr3 −/− embryos, as well as in Foxg1-Cre;Fz3 CKO/− and Foxg1-Cre;Celsr3 CKO/CKO embryos [15,16], suggesting that there may be a stereotyped default pathway for these axons if they do not traverse the normal route through the internal capsule. Mutations in Rac1, Fz3, and Celsr3 also lead to a loss of the corticospinal tract, the corticothalamic tract, and the anterior commissure. However, the loss of the corpus callosum in Foxg1-Cre;Rac1 CKO/− embryos is not seen in Fz3 −/− or Celsr3 −/− embryos, where the corpus callosum is only mildly affected. Motorneuron-specific loss of Rac1 in Olig2-Cre;Rac1 CKO/− embryos produces a distinctive stalling of axons within the dorsal nerve of the hindlimb but a much milder loss of axons in the ventral hindlimb nerve and in forelimb nerves, a pattern that is virtually identical to the one previously observed in Fz3 −/− limbs. A comparison between the Olig2-Cre;Rac1 CKO/− and Olig2-Cre;Fz3 CKO/− phenotypes shows generally similar effects on cranial motor nerve axons and motor nuclei, but among different nuclei, the severity differs between the two genotypes. For example, a comparison of Fig. 2b and 2b' in this work with Figure 1B and B' of Hua et al. [17] shows a more severe loss of axons in the XIIth cranial nerve and adjacent C1 spinal nerve in the Olig2-Cre;Rac1 CKO/− embryo compared to the Fz3 −/− embryo.
Evidence for an interaction between PCP signaling-or, more generally, noncanonical Wnt signaling-and Rho family GTPases comes largely from studies of epithelial polarity in Drosophila and morphogenetic cell movements in Xenopus and zebrafish (reviewed in [59]). Early work based on dominant negative mutants and ectopic expression demonstrated a role for Rho family GTPases in establishing the cytoskeletal organization that underlies PCP-based polarity [60,61] Later work, including experiments with loss-of-function alleles, implied that the relationship between PCP signaling and Rho GTPases may be indirect, with each representing a distinct pathway of information flow that converges on the cytoskeleton [8][9][10]62]. Similar interpretations may apply to experiments with mammalian epithelia, in the context of directed cell movements and asymmetry in terminally differentiated cells [11,63].
At present, data regarding the mechanism of Fz3/Celsr3 signaling in axon growth and guidance are limited. Aside from the experiments reported here, we are not aware of any experiments that have examined a possible relationship between Rho family GTPases and Fz3/Celsr3 function in axon guidance. Based on pharmacologic experiments in embryonic spinal cord explants, Frizzled-mediated guidance appears to be under the control of atypical protein kinase C zeta and phosphatidylinositol-3-kinase [64]. With respect to extrapolations from PCP mechanisms in epithelia, it is possible that axon guidance mechanisms are significantly different, since epithelial polarity depends on the core PCP proteins Vangl1 and Vangl2, whereas recent work suggests that axon guidance does not [19]. Additionally, the rapid extension and retraction of filopodia that accompany the exploratory movements of growth cones suggest that axon guidance may utilize PCP protein complexes that are more labile than the ones formed between epithelial cells. Interestingly, the work of Onishi et al. [65] has found a correlation between endocytic recycling of Fz3 at filopodial tips, Fz3 phosphorylation, and growth cone guidance. These data, together with the known role of Rho GTPases in regulating actin dynamics in growth cones [66], suggest that one potential mechanistic link between Rho family GTPases and Fz3 could be at the level of local regulation of actin polymerization/depolymerization in filopodia.

Conclusions
The experiments reported here indicate a widespread requirement for Rac1 in axon growth and guidance and a cell-autonomous defect in axon growth in Rac1 −/− motor neurons in culture. Following deletion of Rac1 in the forebrain, thalamocortical axons were misrouted in a pattern that is indistinguishable from the pattern previously observed in Frizzled3 −/− and Celsr3 −/− forebrains. In the limbs, motor-neuron-specific deletion of Rac1 produced a distinctive pattern of axon stalling that is virtually identical to the one previously observed in Frizzled3 −/− limbs. These similarities in axon growth and guidance phenotypes caused by Rac1, Frizzled3, and Celsr3 loss-offunction mutations suggest a mechanistic connection between tissue polarity/planar cell polarity signaling and Rac1-dependent cytoskeletal regulation.

Alkaline phosphatase (AP) histochemistry
Throughout this work, the day of finding a copulation plug was counted as embryonic day (E)0.5, and the day when mice were born was counted as postnatal day (P)0. AP staining was performed on 200-μm-thick brain sections from P36 mice. Mice were anesthetized with ketamine/xylazine and then sacrificed by trans-cardiac perfusion with 4 % paraformaldehyde (PFA) [w/v, dissolved in phosphate buffered saline (PBS), pH 7.4], and brains were dissected out and further fixed in 4 % PFA overnight at 4°C. Brains were embedded in 3 % low melting point agarose (w/v, dissolved in PBS) and sectioned at 200 μm on a vibratome. AP histochemistry was performed essentially as previously described [17,18,71]. Vibratome sections in PBS containing 2 mM MgCl 2 were heated in a water bath at 69°C for 90 min and then equilibrated in AP-staining buffer (0.1 M Tris, 0.1 M NaCl, 50 mM MgCl 2 , pH 9.5) overnight at room temperature. AP histochemistry was carried out in AP-staining buffer containing 0.34 μg/ml 4-nitro blue tetrazolium chloride (NBT) and 0.175 μg/ml 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) Immunostaining was conducted as previously described [17,18]. For retina cryosections, the enucleated eye was immersion fixed in 4 % PFA for 30 min at 4°C, and then, the cornea and lens were carefully removed, and the eye cup was immersion fixed in 4 % PFA at 4°C overnight, washed three times in PBS, equilibrated in 30 % sucrose (w/v, dissolved in PBS), embedded in optimal cutting temperature (OCT) compound (Sakura Finetek; Torrance, CA), frozen, and sectioned at 14 μm on a cryostat. For spinal cord cryosections, embryos were immersion fixed in 4 % PFA at 4°C for 2 h, washed three times in PBS, equilibrated in 30 % sucrose, embedded in OCT compound, frozen, and sectioned at 14 μm on a cryostat. For vibratome sections, embryos or brains were immersion fixed in 4 % PFA at 4°C overnight, washed three times in PBS, embedded in 3 % low melting point agarose, and sectioned at 120 μm on a vibratome. For whole-mount E11.5 embryos, and E12.5 and E13.5 limbs, samples were immersion fixed in 4 % PFA at 4°C for 2 h and washed three times in PBS.
For immunostaining of 14-μm-thick cryosections and 120-μm-thick vibratome sections, sections were blocked in blocking solution [PBS containing 0.3 % Triton X-100 and 5 % normal goat serum (NGS) or normal donkey serum (NDS)] at room temperature for 1 h and incubated with primary antibody in blocking solution at 4°C overnight. Sections were then washed five times in PBST (PBS containing 0.3 % Triton X-100) and incubated with secondary antibody in blocking solution at room temperature for 1 h (cryosections) or at 4°C overnight (vibratome sections). Sections were then washed five times in PBST and mounted with Fluoromount-G (Southern Biotech; Birmingham, AL).
For immunostaining of whole-mount embryos and limbs, samples were first incubated in Dent's Bleach [10 % H 2 O 2 , 13.3 % dimethyl sulfoxide (DMSO), 53.3 % methanol] at 4°C for 24 h, washed with methanol five times, and fixed in Dent's Fix (20 % DMSO, 80 % methanol) at 4°C overnight. Samples were washed in PBS three times, incubated with primary antibody in blocking solution (20 % DMSO, 75 % PBST, 5 % NGS, 0.025 % sodium azide) at room temperature for 5-7 days with gentle end-over-end rotation, and then washed five times in PBST. Samples were incubated with secondary antibody in blocking solution at room temperature for 2 days with gentle end-over-end rotation and then washed five times in PBST. Samples were dehydrated in 50 % methanol/PBS and then 100 % methanol and cleared in BBBA.

Dissociated spinal cord neuron cultures
Individual spinal cords from E11.5 embryos were dissected in Hank's balanced salt solution (HBSS), and genotyping was performed on the remaining embryonic tissue. Spinal cords were diced, washed in HBSS, and incubated in 0.05 % trypsin-EDTA for 12 min at 37°C, washed in DMEM with 10 % heat-inactivated normal horse serum, resuspended in the same media, triturated with a fire-polished Pasteur pipette, and then filtered through a 70-μm filter. Cells were plated on sterile glass coverslips coated in poly-D-lysine (20 μg/ml) and laminin (20 μg/ml) at a density of 50,000 cells/coverslip in 24-well plates. For immunohistochemistry, cells were washed twice in PBS with 0.5 mM MgCl 2 and 1 mM CaCl 2 and fixed for 10 min in 4 % PFA. GFP was visualized with chicken anti-GFP (GFP-1020; Aves Lab; Tigard, OR). Axon lengths were analyzed using the NeuronJ plug-in for ImageJ.

Cholera toxin subunit B (CTB) anterograde tracing of optic tract axons
Adult mice were deeply anesthetized with ketamine/xylazine, and~1-2 μl 2 mg/ml Alexa Fluor 594-conjugated cholera toxin subunit B (C-22842; Life Technologies) was intraocularly injected into the left eye of the mouse with a glass micro-needle. Four days later, mice were sacrificed by trans-cardiac perfusion, and brains were dissected out and further fixed in 4 % PFA overnight at 4°C. Brains were embedded in 3 % low melting point agarose and sectioned at 100 μm on a vibratome. Sections were mounted on slides with Fluoromount-G and imaged by confocal microscopy.

Microscopy and image analysis
Samples processed for AP histochemistry were imaged using a Zeiss Stemi V11 microscope with a color Axiocam CCD in combination with Openlab software. Immunostained samples were imaged using a Zeiss LSM700 confocal microscope with Zen software. Images of whole mounts were acquired with a 10× air objective at 10-μm intervals in the Z dimension, and the entire Z stack was either collapsed using maximum intensity projection or color-coded based on Z-dimension depth. BBBA-cleared embryos were positioned in custom-built metal holders consisting of a shallow triangular trough (sides: 2 cm × 2 cm × 1 cm; and depths: 1, 2, 3, or 4 mm). The trough was filled with BBBA and coverslipped during imaging.

Quantification of motor and sensory neurons, measurement of nerve diameter, and statistical analysis
The number of neurons on each cryosection was manually quantified using ImageJ software. To measure the diameter of limb motor nerves, a line perpendicular to the nerve at the designated site was drawn in Adobe Illustrator and the width of the nerve along this line was then measured. Statistical comparisons were performed in Microsoft Excel.