Rac1

DEVELOPMENTAL BIOLOGY

Embryonic

Strong and ubiquitous expression of Rac1 mRNA is evident in precellular as well as cellular blastoderm stages. The transcript is concentrated at the basal part of the cellular blastoderm. After gastrulation, Rac1 transcripts become highly enriched in mesoderm between stages 10 and 12. At stage 13, Rac1 transcripts start to appear in the nervous system and the gut. Later in development, somatic mesoderm expression vanishes, whereas the nervous system and gut expression persists. The expression pattern of Cdc42, another Rho family GTPase, is qualitatively similar to that of Rac1 in all of the stages described for Rac1 (Luo, 1994).

The Rho GTPases Rac1 and Cdc42 have been implicated in the regulation of axon outgrowth and guidance. However, the downstream effector pathways through which these GTPases exert their effects on axon development are not well characterized. Axon outgrowth defects within specific subsets of motoneurons expressing constitutively active Drosophila Rac1 largely persist even with the addition of an effector-loop mutation to Rac1 that disrupts its ability to bind to p21-activated kinase (Pak) and other Cdc42/Rac1 interactive-binding (CRIB)-motif effector proteins. While hyperactivation of Pak itself does not lead to axon outgrowth defects as when Rac1 is constitutively activated, live analysis reveals that Pak can alter filopodial activity within specific subsets of neurons similar to constitutive activation of Cdc42. Moreover, the axon guidance defects induced by constitutive activation of Cdc42 persist even in the absence of Pak activity. These results suggest that (1) Rac1 controls axon outgrowth through downstream effector pathways distinct from Pak, (2) Cdc42 controls axon guidance through both Pak and other CRIB effectors, and (3) Pak's primary contribution to in vivo axon development is to regulate filopodial dynamics that influence growth cone guidance (Kim, 2003).

These studies support the idea that Rac1 mediates axon outgrowth through downstream effector pathways that are distinct from those that mediate Cdc42-dependent guidance. This is based on the following: (1) by using mutants of Rac1 and Cdc42 that render them constitutively active, the downstream cellular events that each is responsible for during axon development can be effectively isolated. Rac1 and Cdc42 likely work through different effector pathways, since constitutive activation of each leads to distinct phenotypes. (2) An effector-loop mutation that disrupts Cdc42's ability to bind CRIB-motif proteins can effectively suppress axonal defects resulting from constitutively active Cdc42. In contrast, the parallel mutation in Rac1 only partially suppresses defects resulting from constitutive activation of Rac1. Finally, activation of Pak, a CRIB effector for both Rac1 and Cdc42, leads to guidance defects and increased filopodial activities similar to those seen in activated Cdc42 mutants, but not activated Rac1 mutants (Kim, 2003).

A model is favored in which Rac1 mediates axon outgrowth mainly through effector proteins that do not possess the CRIB-motif, while Cdc42 mediates growth cone guidance primarily through CRIB proteins. Since activation of the CRIB protein Pak does not lead to axon outgrowth, but rather, guidance defects, Pak does not seem to play a direct role in Rac1-dependent outgrowth. This model is consistent with reports demonstrating that Pak is not required for axon outgrowth, at least during its initial phase, but is required during proper guidance and targeting of axons in a later phase. This, however, does not necessarily exclude Pak from playing a significant role in other Rac signaling pathways that mediate axon development (Kim, 2003).

Closer examination of growth cone behavior begins to reveal nonlinear signaling events. The ability of activated Pak to enhance filopodial activity similar to what activated Cdc42 achieves suggests that Pak likely plays a role in Cdc42-mediated growth cone guidance that depends on filopodial activity control. In this context, however, Pak's contribution to filopodial regulation is thought necessary, but not entirely sufficient, to dictate proper guidance. Also, Pak may not be the sole effector responsible for Cdc42-dependent filopodial activity. CRIB proteins affected by the Y40C mutation in constitutively active Cdc42 do not play a role in filopodial regulation, suggesting that other effectors are likely playing redundant roles in mediating filopodial activity. In addition, the persistence of guidance defects in motoneurons expressing Dcdc42V12 in a Pak null genetic background further suggests that Cdc42 is likely working through other CRIB effectors, in addition to Pak, to mediate its effects on growth cone behavior and guidance (Kim, 2003).

By isolating cellular events downstream of Rac1 and Cdc42 through the use of constitutively active mutants, the effector pathways responsible for axon development have become amenable to dissection. Since constitutively active mutants lock the GTPases in an active GTP-bound state, downstream events can be examined irrespective of GTPase activation. Furthermore, by coupling effector-loop mutations with the constitutively active mutation, a role for specific effectors becomes more evident. More studies that isolate effector pathways are necessary to complement existing studies (Kim, 2003).

In conclusion, studies using in situ analysis support the general view that cytoskeletal dynamics, outgrowth, and guidance are not necessarily coupled directly under the activation of Rac1 or Cdc42. Instead, these studies imply that a complex repertoire of growth cone behaviors are being mediated in parallel as a meshwork of events, and that the workings of Pak and other Rac1 and Cdc42 effector proteins, many of which are yet to be examined in situ, collectively contribute to proper axon development (Kim, 2003).

Oogenesis

Similar to the Drosophila Egfr and to the mammalian PDGFR family, stimulation of PDGF- and VEGF-receptor related (Pvr) activates the MAP-kinase pathway in Schneider cells as well as in border cells. However, it has been shown, by loss-of-function and gain-of-function experiments, that MAP-kinase signaling does not affect border cell migration. In addition, no effect of phospholipase C-gamma (PLC-gamma) or phosphatidylinositol 3' kinase (PI3K) has been demonstrated on this migration, using loss-of-function mutants (PLC-gamma) or border cell expression of dominant negative and dominant activated forms (PI3K). This was somewhat unexpected, since PLC-gamma and PI3K have been implicated in motility and guidance effects of RTKs (in particular PDGFR) in tissue culture cells. To address how Pvr signaling might be affecting cell migration in vivo, the effect of Pvr signaling on cell morphology and cytoskeleton was tested. In border cells as well as in other follicle cells, expression of lambda-Pvr has a dramatic effect on the actin cytoskeleton. Massive F-actin accumulation, actin-rich extensions, and changes in cell shape were produced in lambda-Pvr expressing follicle cells. The normal cells have modest cortical F-actin accumulation. This result was likely to be relevant to the guidance function of Pvr, because direct control of F-actin accumulation would allow receptor activation to control cell migration (Duchek, 2001).

The actin cytoskeleton has been shown to be affected by small GTPases of the Rho superfamily in many systems, with the exact effects depending on the cellular context. In the border cell migration system, Rac is an attractive candidate for mediating the effect of activated Pvr, since dominant negative Rac (RacN17) has been shown to inhibit border cell migration. Epistasis experiments could not be done by quantifying border cell migration because activated Pvr and dominant negative Rac have the same effect. Instead, whether Rac is required for the effect of Pvr on the actin cytoskeleton in follicle cells was tested. Coexpression of dominant negative Rac suppresses the effect of activated Pvr on the actin cytoskeleton. In addition, follicle cells expressing activated Rac (RacV12) have dramatic accumulation of F-actin, resembling that caused by activated Pvr. Finally, if Rac were directly downstream of Pvr, one would expect activated Rac to inhibit border cell migration, as observed for the activated receptor. Although a previous study reported that activated Rac does not affect border cell migration (Murphy, 1996), this was reexamined using the slboGal4 driver and it was found that activated Rac completely blocks border cell migration. These results are consistent with a role of Rac in the guidance pathway downstream of Pvr (Duchek, 2001).

In mammalian tissue culture cells, PDGF stimulation can cause Rac-dependent F-actin accumulation, suggesting that the effect observed in follicle cells may reflect a conserved pathway. PI3K has been implicated as a mediator of the effect of PDGFR on Rac in Swiss 3T3 cells. However, PI3K does not appear to play a key role in guidance of border cell migration as discussed above. To investigate how Pvr might lead to activation of Rac, two groups of Drosophila mutants were tested for their effect on border cell migration: mutants in genes shown to be downstream of receptor tyrosine kinases in other contexts, and mutants linked to Rac activation. Most mutations were homozygous lethal, so their effect in border cells was tested by generating mutant clones in a heterozygous animal (mosaic analysis). Of the 8 genes tested, only myoblast city (mbc) has a detectable effect on border cell migration. Mbc is homologous to mammalian DOCK180 and C. elegans CED-5. Mbc/DOCK180/CED-5 acts as an activator of Rac (Duchek, 2001 and references therein).

mbc has been independently identified in a screen for gain-of-function suppressors of the slbo mutant phenotype. slbo mutant border cells migrate poorly. Increased expression of mbc in slbo mutant border cells improves their migration, suggesting that mbc has a positive role in promoting border cell migration. Mbc protein is detected in follicle cells, including border cells, and is overexpressed upon induction of the EP element EPg36390 located upstream of mbc. Removing mbc function from border cells by generating mutant clones causes severe delays in their migration. At stage 10, when 100% of control (GFP) clones have reached the oocyte, only 10% of mbc mutant border cell clusters have done so, and these are the oldest egg chambers. Thus, mbc is not absolutely required for border cell migration, but, contrary to the other genes implicated in RTK and Rac signaling, loss of mbc function severely impairs this cell migration (Duchek, 2001).

A signaling network for patterning of neuronal connectivity in the Drosophila brain

The precise number and pattern of axonal connections generated during brain development regulates animal behavior. Therefore, understanding how developmental signals interact to regulate axonal extension and retraction to achieve precise neuronal connectivity is a fundamental goal of neurobiology. This question was investigated in the developing adult brain of Drosophila. Extension and retraction is regulated by crosstalk between Wnt, fibroblast growth factor (FGF) receptor, and Jun N-terminal kinase (JNK) signaling, but independent of neuronal activity. The Rac1 GTPase integrates a Wnt-Frizzled-Disheveled axon-stabilizing signal and a Branchless (FGF)-Breathless (FGF receptor) axon-retracting signal to modulate JNK activity. JNK activity is necessary and sufficient for axon extension, whereas the antagonistic Wnt and FGF signals act to balance the extension and retraction required for the generation of the precise wiring pattern (Srahna, 2006).

Based on the observation that blocking Fz2 results in decreased numbers of dorsal cluster neuron (DCN) axons in the medulla, it was reasoned that Fz2 could be a receptor for a putative stabilization signal. Since Fz2 and Fz are partially redundant receptors for the canonical Wnt signaling pathway, expression of the canonical Wnt ligand Wingless (Wg) was investigated in the brain during pupation. However, no Wg expression was detected in the pupal optic lobes, suggesting that Wg is unlikely to be involved in regulating DCN axon extension. Therefore, the expression of Wnt5, which has been shown to be involved in axon repulsion and fasciculation in the embryonic CNS, was investigated. Anti-Wnt5 staining revealed widely distributed Wnt5 expression domains beginning at PF and lasting throughout pupal development and into adult life. Wnt5 is strongly expressed in the distal medulla and is also present on axonal bundles crossing the second optic chiasm.The number of DCN axons crossing to the medulla was examined in wnt5 mutant flies. The number of DCN axons crossing the optic chiasm is reduced from 11.7 to 7.9 in the absence of wnt5, suggesting that it may play a role in stabilizing DCN axons (Srahna, 2006).

Next, the requirement of the Wnt signaling adaptor protein Dsh was tested. In animals heterozygous for dsh⁶, a null allele of dsh, the average number of DCN axons crossing between the lobula and the medulla is reduced from 11.7 to 7.6 with 78.5% showing less than eight axons crossing. Signaling through Dsh is mediated by one of two domains. Signaling via the DIX (Disheveled and Axin) domain is thought to result in the activation of Armadillo/β-Catenin. DEP (Disheveled, Egl-10, Pleckstrin) domain-dependent signaling results in activation of the JNK signaling pathway by regulation of Rho family GTPase proteins during, for example, convergent extension movements in vertebrates. To uncover which of these two pathways is required for DCN axon extension the dsh¹ mutant, deficient only in the activity of the DEP domain, was tested. Indeed, in brains from dsh¹ heterozygous animals the number of extending axons was reduced from 11.7 to 7.4. In flies homozygous for the dsh¹ allele the average number of axons crossing was further reduced to 4.7, with all the samples having less than six axons crossing. In contrast, the DCN-specific expression of Axin, a physiological inhibitor of the Wnt canonical pathway, did not affect the extension of DCN axons. Similarly, expression of a constitutively active form of the fly β-Catenin Armadillo also had no apparent effect on DCN extension. Finally, whether Wnt5 and Dsh interact synergistically was tested. To this end, wnt5, dsh¹ trans-heterozygous animals were generated. These flies show the same phenotype as flies homozygous for dsh¹, suggesting that Wnt5 signals through the Dsh DEP domain (Srahna, 2006).

To determine if dsh is expressed at times and places suggested by its genetic requirement in DCN axon outgrowth, the distribution of Dsh protein during brain development was examined. Dsh protein is ubiquitously expressed during brain development. High expression of Dsh is detected in the distal ends of DCN axons at about 15% PF shortly before they extend across the optic chiasm toward the medulla. In general, higher levels of Dsh were observed in the neuropil than in cell bodies (Srahna, 2006).

In summary, these data indicate that the stabilization of DCN axons is dependent on the Dsh protein acting non-canonically via its DEP domain. Importantly, the axons that do cross in dsh mutant brains do so along the correct paths. This suggests that, like JNK signaling, Wnt signaling regulates extension, but not guidance, of the DCN axons (Srahna, 2006).

Wnt signaling to Dsh requires the Fz receptors. To examine if the effect of Wnt5 on DCN axon extension is also mediated by Fz receptors, the number of DCN axons crossing the optic chiasm in was counted fz, fz2, and fz3 mutants. There was no significant change in the number of axons crossing in the brain of fz3 homozygous animals. In contrast, in brains heterozygous for fz and fz2, the number of the axons crossing was reduced from 11.7 to 6.6 (fz) and 6.9 (fz2), with 71% and 85.7%, respectively, showing less than eight axons crossing. These data suggest that DCN axons respond to Wnt5 using the Fz and Fz2 receptors, but not Fz3. To determine whether the Fz receptors act cell-autonomously in individual DCNs, single-cell clones doubly mutant for fz and fz2 were generated and the number of DCN axons crossing the optic chiasm was counted. In contrast to wild-type cells, where 37% of all DCN axons cross, none of the fz, fz2 mutant axons reach the medulla. To test whether wnt5, fz, and fz2 genetically interact in DCNs, flies trans-heterozygous for wnt5 and both receptors were examined. Flies heterozygous for both wnt5 and fz mutations show a strong synergistic loss of DCN axons (11.7 to 3.7) and in fact have a phenotype very similar to that of flies homozygous for dsh¹. Flies doubly heterozygous for wnt5 and fz2 also show a significant decrease in DCN axons (5.7), compared with either wnt5 (~8) or fz2 (8.5) mutants. These data indicate that the genetic interaction between wnt5 and fz is stronger than the interaction between wnt5 and fz2 (Srahna, 2006).

Examination of the expression domains of Fz and Fz2 in the developing brain supports the possibility that they play roles in stabilizing DCN axons. Both Fz and Fz2 are widely expressed in the developing adult brain neuropil. In addition, Fz is expressed at higher levels in DCN cell bodies (Srahna, 2006).

The observation that the wnt5 null phenotype can be enhanced by reduction of Fz, Fz2, or Dsh suggests that another Wnt may be partially compensating for the loss of Wnt5. To test this possibility, flies heterozygous for either wnt2 or wnt4 were examined. wnt2 heterozygotes display a reduction of DCN axon crossing from 11.7 to 7.3, whereas no phenotype was observed for wnt4. Thus, wnt2 and wnt5 may act together to stabilize the subset of DCN axons that do not retract during development. In summary, these results support the model that Wnt signaling via the Fz receptors transmits a non-canonical signal through Dsh resulting in the stabilization of a subset of DCN axons (Srahna, 2006).

Data is provided that supports the hypothesis that the regulation of JNK by Rac1 modulates DCN axon extension. As such attempts were made to determine how Wnt signaling might interact with Rac1 and JNK. The opposite phenotypes of dsh and Rac1 loss-of-function suggest that they might act antagonistically. To determine if Rac1 is acting upstream of, downstream of, or in parallel to Dsh in DCN axon extension, dominant-negative Rac1 was expressed in dsh¹ mutant flies. If Rac1 acts upstream of Dsh, the dsh¹ phenotype (i.e., decreased numbers of axons crossing the optic chiasm) is expected. If Rac1 acts downstream of Dsh, the Rac1 mutant phenotype (i.e., increased number of axons crossing) would be expected If they act in parallel, an intermediate, relatively normal phenotype is expected. Increased numbers of axon crossing were observed, suggesting that Rac1 acts downstream of Dsh during DCN axon extension and that Dsh may repress Rac1 (Srahna, 2006).

Next, whether Dsh control of DCN axon extension is mediated by the JNK signaling pathway acting downstream of Wnt signaling was tested, as the similarity of their phenotypes suggests. If this were the case, activating JNK signaling should suppress the reduction in Dsh levels. Conversely, reducing both should show a synergistic effect. Therefore the JNKK hep was expressed in dsh¹ heterozygous flies and it was found that the hep gain-of-function is epistatic to dsh loss-of-function. Furthermore, reducing JNK activity by one copy of BSK-DN in dsh¹ mutant animals results in a synergistic reduction of extension to an average of 0.8 axons with 60% showing no axons crossing and no samples with more than three axons. In summary, the results of genetic analyses suggest that Wnt signaling via Dsh enhances JNK activity through the suppression of Rac1 (Srahna, 2006).

Dsh appears to promote JNK signaling and to be expressed in DCN axons prior to their extension toward the medulla early in pupal development. Since JNK signaling is required for this initial extension, it may be that Dsh also plays a role in the early extension of DCN axons. To test this possibility, DCN axon extension was examined at 30% pupal development in dsh¹ mutant brains. In wild-type pupae, essentially all (~40) DCN axons extend toward the medulla. In contrast, in dsh¹ mutant pupae, a strong reduction in the number of DCN axons crossing the optic chiasm between the lobula and the medulla was observed (Srahna, 2006).

Although the genetic data indicate that Dsh- and Rac-mediated signaling have sensitive and antagonistic effects on the JNK pathway, they do not establish whether the Dsh-Rac interaction modulates JNK's intrinsic activity. To test this, the amount of phosphorylated JNK relative to total JNK levels in fly brains was evaluated by Western blot analysis using phospho-JNK (P-JNK) and pan-JNK specific antibodies. Then it was determined if Dsh is indeed required for increased levels of JNK phosphorylation. Dsh¹ mutant brains showed a 25% reduction in P-JNK consistent with a stimulatory role for Dsh on JNK signaling. The reduction caused by loss of Dsh function is reversed, when the amount of Rac is reduced by half, consistent with a negative effect of Rac on JNK signaling downstream of Dsh. These data support the conclusion that Dsh and Rac interact to regulate JNK signaling by modulating the phosphorylated active pool of JNK (Srahna, 2006).

Taken together, these data suggest that during brain development DCN axons extend under the influence of JNK signaling. A non-canonical Wnt signal acting via Fz and Dsh ensures that JNK signaling remains active by attenuating Rac activity. In contrast, activation of the FGFR activates Rac1 and suppresses JNK signaling. These data support a model whereby the balance of the Wnt and FGF signals is responsible for determining the number of DCN axons that stably cross the optic chiasm. To test this model, FGFR levels were reduced, using the dominant-negative btl transgene, in dsh¹ heterozygous flies. It was found that simultaneous reduction of FGF and Wnt signaling restored the number of axons crossing the optic chiasm to almost wild-type levels (10.2, with 33% of the samples indistinguishable from wild-type, suggesting that the two signals in parallel, act to control the patterning of DCN axon connectivity (Srahna, 2006).

These data suggest the following model of DCN axon extension and retraction. DCN axons extend due to active JNK signal. These axons encounter Wnt5 and probably Wnt2 as well, resulting in activation of Disheveled. Disheveled, via its DEP domain, has a negative effect on the activity of the Rac GTPase, thus keeping JNK signaling active. After DCN axons cross the second optic chiasm they encounter a spatially regulated FGF/Branchless signal that activates the FGFR/Breathless pathway. Breathless in turn activates Rac, which inhibits JNK signaling in a subset of axons. These axons then retract back toward the lobula. The wide expression of the different components of these pathways and the modulation of JNK phosphorylation by Dsh and Rac in whole-head extracts strongly suggests that this model may apply to many neuronal types (Srahna, 2006).

TGF-β signals regulate axonal development through distinct Smad-independent mechanisms

Proper nerve connections form when growing axons terminate at the correct postsynaptic target. Transforming growth factor β (TGFβ) signals regulate axon growth. In most contexts, TGFβ signals are tightly linked to Smad transcriptional activity. Although known to exist, how Smad-independent pathways mediate TGFβ responses in vivo is unclear. In Drosophila mushroom body (MB) neurons, loss of the TGFβ receptor Baboon (Babo) results in axon overextension. Conversely, misexpression of constitutively active Babo results in premature axon termination. Smad activity is not required for these phenotypes. This study shows that Babo signals require the Rho GTPases Rho1 and Rac, and LIM kinase1 (LIMK1), which regulate the actin cytoskeleton. Contrary to the well-established receptor activation model, in which type 1 receptors act downstream of type 2 receptors, this study shows that the type 2 receptors Wishful thinking (Wit) and Punt act downstream of the Babo type 1 receptor. Wit and Punt regulate axon growth independently, and interchangeably, through LIMK1-dependent and -independent mechanisms. Thus, novel TGFβ receptor interactions control non-Smad signals and regulate multiple aspects of axonal development in vivo (Ng, 2008).

Once growing axons reach the correct postsynaptic target, axon outgrowth terminates and synaptogenesis begins. These studies suggest that TGFβ signals play a role. When Babo is inactivated, MB axon growth does not terminate properly and overextends across the midline. Consistent with this, CA Babo expression results in precocious termination, forming axon truncations. How Babo is spatially and temporally regulated remains to be determined. Analogous to the Drosophila NMJ, MB axon growth might be terminated through retrograde signalling. Target-derived TGFβ ligands could signal to Babo (on MB axon growth cones) and stop axons growing further. In an alternative scenario, TGFβ ligands might act as a positional cue that prevents MB axons from crossing the midline. Recent data have shown that Babo acting through Smad2 restricts individual R7 photoreceptor axons to single termini. Loss of Babo, Smad2, or the nuclear import regulator Importin α3 (Karyopherin α3 - FlyBase), results in R7 mutant axons invading neighbouring R7 terminal zones. With the phenotype described in this study, Babo could similarly be restricting MB axons to appropriate termination zones, its loss resulting in inappropriate terminations on the contralateral side (Ng, 2008).

In contrast to MB neurons, Babo inactivation in AL and OL neurons resulted in axon extension and targeting defects. This might reflect cell-intrinsic differences in the response in different neurons to a common Babo signalling program. This may be the case for MB axon pruning and DC axon extension, which require Babo/Smad2 signals. Whether these differences derive from cell-intrinsic properties, or from Babo signal transduction, they underline the importance of Smad-independent signals in many aspects of axonal development (Ng, 2008).

The results suggest that Smad-independent signals involve Rho GTPases. One caveat in genetic interaction experiments is that the loss of any given gene might not be dosage-sensitive with a particular assay. Nevertheless, all the manipulations together suggest that Babo-regulated axon growth requires Rho1, Rac and LIMK1. How Babo signals involve Rho GTPases remains to be fully determined. In addition to LIMK1, which binds to Wit, one possibility, as demonstrated for many axon guidance receptors, is that the RhoGEFs, RhoGAPs and Rho proteins might be linked to the Babo receptor complex. Thus, ligand-mediated changes in receptor properties would lead to spatiotemporal changes in Rho GTPase and LIMK1 activities (Ng, 2008).

The data suggest that a RhoGEF2/Rho1/Rok/LIMK1 pathway mediates Babo responses. Whether Rac activators are required is unclear, as tested RacGEFs do not genetically interact with babo. In this respect, rather than through GEFs, Babo might regulate Rac through GAPs, by inhibiting Tumbleweed (Tum) activity (Ng, 2008).

Do mutations in Rho1 and Rac components phenocopy babo phenotypes? β lobe overextensions are observed in Rok, Rho1 and Rac mutant neurons. In MB neurons, Rac GTPases also control axon outgrowth, guidance and branching. Rho1 also has additional roles in MB neurons. Although Rho1 mutant neuroblasts have cell proliferation defects, single-cell αβ clones do show β lobe extensions. RhoGEF2 strong loss-of-function clones do not exhibit axon overextension. As there are 23 RhoGEFs in the Drosophila genome, there might well be redundancy in the way Rho1 is activated. LIMK1 inactivation in MB neurons was reported previously. Axon overextensions were not observed as LIMK1 loss results in axon outgrowth and misguidance phenotypes. This suggests that LIMK1 mediates multiple axon guidance signals, of which TGFβ is a subset in MB morphogenesis (Ng, 2008).

Although their phenotypes are similar, several lines of evidence indicate that CA Babo does not simply reflect LIMK1 misregulation in MB neurons. First, whereas LIMK1 genetically interacts with most Rho family members and many Rho regulators, CA babo is dosage-sensitive only to Rho1 and Rac and specific Rho regulators, suggesting that Babo regulates LIMK1 only through a subset of Rho signals (Ng, 2008).

Second, the LIMK1 misexpression phenotype is suppressed by expression of wild-type cofilin (Twinstar Tsr), S3A Tsr, or the cofilin phosphatase Slingshot (Ssh). By contrast, only wild-type Tsr, but not S3A Tsr or Ssh, suppresses CA Babo. The suppression by wild-type Tsr might reflect a restoration of the endogenous balance or spatial distribution of cofilin-on (unphosphorylated) and -off (phosphorylated) states within neurons. Indeed, optimal axon outgrowth requires cofilin to undergo cycles of phosphorylation and dephosphorylation. Since S3A forms of cofilin cannot be inactivated and recycled from actin-bound complexes, wild-type cofilin is more potent in actin cytoskeletal regulation (Ng, 2008).

CA Babo might not simply misregulate LIMK1 but also additional cofilin regulators. Recent data suggest that extracellular cues (including mammalian BMPs) can regulate cofilin through Ssh phosphatase and phospholipase Cγ activities. In different cell types, cofilin phosphorylation and phospholipid binding (which also inhibits cofilin activity) states vary and potently affect cell motility and cytoskeletal regulation. Whether a combination of LIMK1, Ssh and phospholipid regulation affects cofilin-dependent axon growth remains to be determined (Ng, 2008).

Third, by phalloidin staining, LIMK1, but not CA Babo, misexpression results in a dramatic increase in F-actin in MB neurons. Thus, CA Babo does not in itself lead to actin misregulation. Fourth, Babo also regulates axon growth independently of LIMK1 (Ng, 2008).

This study differs significantly from the canonical model of Smad signalling, in which type 1 receptors function downstream of the ligand-type 2 receptor complex. In this study, the gain- and loss-of-function results suggest that type 2 receptors act downstream of type 1 signals. Since ectopic only Wit and Put suppress the babo axon overextension phenotype, this implies that Smad-dependent and -independent signals have distinct type 1/type 2 receptor interactions. How these interactions propagate Smad-independent signals remains to be fully determined. Babo could act as a ligand-binding co-receptor with Wit and Put. In addition, Babo kinase activity could regulate type 2 receptor or Rho functions. The results suggest, however, that provided that Wit or Put signals are sufficiently high, Babo is not required. Whatever the mechanism(s), it is likely that Babo requires the Wit C-terminus-LIMK1 interaction to relay cofilin phosphoregulatory signals. How Put functions is unclear. Since the put¹³⁵ allele (used in this study) carries a missense mutation within the kinase domain, this suggests that kinase activity is essential. put does not genetically interact with LIMK1. Since Put lacks the C-terminal extension of Wit that is necessary for LIMK1 binding, this suggests that Put acts independently of LIMK1. One potential effector is Rac, which, in the context of Babo signalling, also appears to be Pak1- and thus LIMK1-independent (Ng, 2008).

In MB neurons, Wit and Put can function interchangeably. In other in vivo paradigms, type 2 receptors are not interchangeable. However, since the Wit C-terminal tail is required to substitute for Put, this suggests that Wit axon growth signals are independent of its kinase activity. Together, this suggests that Smad-independent signals involve LIMK1-dependent and -independent mechanisms (Ng, 2008).

This study shows that Babo mediates two distinct responses in related MB populations. How do MB neurons choose between axon pruning and axon growth? The babo rescue studies suggest that whereas Babo_a or Babo_b elicits Smad-independent responses, only Babo_a mediates Smad-dependent responses. Since Babo isoforms differ only in the extracellular domain, differences in ligand binding could determine Smad2 or Rho GTPase activation. However, it is worth noting that in DC neurons, either isoform mediates axon extension through Smad2 and Medea. In addition, although expressed in all MB neurons, CA babo misexpression (which confers ligand-independent signals) perturbs only αβ axons. Thus, cell-intrinsic properties might also be essential in determining Babo responses (Ng, 2008).

Many TGFβ ligands signal through Babo. For example, Dawdle, an Activin-related ligand, patterns Drosophila motor axons, whereas Activin (Activin-β, FlyBase) is required for MB axon pruning. Whether these ligands regulate Babo MB, AL and OL axonal morphogenesis is unclear. Taken together, the evidence suggests that Babo signalling is varied in vivo and is involved in many aspects of neuronal development (Ng, 2008).

TGFβ signals are responsible for many aspects of development and disease and, throughout different models, Smad pathways are closely involved. Although Smad-independent pathways are known, their mechanisms and roles in vivo are unclear. TGFβ signals often drive cell shape changes in vivo. During epithelial-to-mesenchymal transition (EMT), cells lose their epithelial structure and adopt a fibroblast-like structure that is essential for cell migration during development and tumour invasion. TGFβ-mediated changes in the actin cytoskeleton and adherens junctions are necessary for EMT. Although Smads are crucial, TGFβ signals also involve the Cdc42-Par6 complex, resulting in cell de-adhesion and F-actin breakdown through Rho1 degradation. In other studies, however, TGFβ-mediated EMT has been shown to require Rho1, which can be regulated by Smad activity (Ng, 2008).

Many TGFβ-driven events in Drosophila are Smad-dependent. Whether Smad-independent roles exist beyond those identified in this study remains to be tested. This study therefore provides a framework to understand how non-Smad signals regulate cell morphogenesis during development (Ng, 2008).

Effects of Mutation or Deletion

Actin and microtubule cytoskeletons have overlapping, but distinct roles in the morphogenesis of epidermal hairs during Drosophila wing development. The function of both the actin and microtubule cytoskeletons appears to be required for the growth of wing hairs, since treatment of cultured pupal wings with either cytochalasin D or vinblastine is able to slow prehair extension. At higher doses, a complete blockage of hair development is seen. The microtubule cytoskeleton is also required for localizing prehair initiation to the distalmost part of the cell. Disruption of the microtubule cytoskeleton results in the development of multiple prehairs along the apical cell periphery. The multiple hair cells are a phenocopy of mutations in the inturned group of tissue polarity genes, which are downstream targets of the frizzled signaling/signal transduction pathway. The actin cytoskeleton also plays a role in maintaining prehair integrity during prehair development, since treatment of pupal wings with cytochalasin D, which inhibits actin polymerization, led to branched prehairs. This is a phenocopy of mutations in crinkled, and suggests mutations that cause branched hairs will be in genes that encode products that interact with the actin cytoskeleton. Several other mutant genotypes produce branched or split bristles or hairs. For example, mutations in singed, chickadee and capping protein produce bristles and/or hairs that are split, bent or stunted in ways that partially resemble cytochalasin D treatment. However, the phenotypes associated with these mutations do not resemble those seen with CD treatment as closely as the phenotype associated with crinkled (e.g. there is not hair splitting in sn mutants). The recent finding that mutations in the small G-protein rho result in an inturned-like phenotype and that the expression of a dominant negative form of rac also results in multiple hair cell phenotype is interesting with regard to the interaction of the actin and microtubule cytoskeletons. Small G-proteins of the rho and rac families are thought to interact with the actin cytoskeleton, yet they produce a wing hair phenotype that is similar to what is seen with the disruption of the microtubule cytoskeleton. This could be due to both the small G-proteins and the micotubule cytoskeleton being required for localizing a common component or activity to the vicinity of the distal vertex, or to the small G-proteins affecting the structure of the microtubule cytoskeleton, or to the microtubular cytoskeleton functioning in the localization of the small G-proteins or, alternatively, these two classes of proteins could be functioning in parallel pathways that function independently to restrict prehair initiation to the distal region of the cell. The observation that the expression of a dominant negative form of rac1 causes a disruption of the microtubule array suggests the possibliity that the phenotypes associate with G-protein loss could be due to their disrupting the structure/function of the microtubule cytoskeleton and not to their being part of the frizzled signaling/signal transduction pathway (Turner, 1998).

Frizzled family proteins have been described as receptors of Wnt signaling molecules. In Drosophila, the two known Frizzled proteins are associated with distinct developmental processes. Genesis of epithelial planar polarity requires Frizzled, whereas Dfz2 affects morphogenesis by wingless-mediated signaling. Dishevelled is required in both signaling pathways. Genetic and overexpression assays have been used to show that Dishevelled activates JNK cascades. In contrast to the action of wingless-pathway components, mutations in rhoA, hemipterous, basket, and jun as well as deficiencies removing the Rac1 and Rac2 genes show a strong dominant suppression of a Dishevelled overexpression phenotype in the compound eye. In an in vitro assay, expression of Dsh has been shown to induce phosphorylation of Jun, indicating that Dsh is a potent activator of the JNK pathway. Whereas the PDZ domain of Dsh, known to be required in the transduction of the wingless signal, is dispensable for signal-independent induction of Jun phosphorylation, the C-terminal DEP domain of Dsh is found to be essential. The planar polarity-specific dsh1 allele is found to be mutated in the DEP domain. These results indicate that different Wnt/Fz signals activate distinct intracellular pathways, and Dishevelled discriminates among them by distinct domain interactions (Boutros, 1998).

How can Fz/Dsh signaling be linked to small GTPase and JNK/MAPK pathways? Recent studies provided evidence that links G protein-coupled receptors, which share structural features with Fz proteins, to MAPK signaling through heterotrimeric G proteins and PI-3 kinases. It is intriguing to speculate that a subset of Fz proteins might signal through a similar pathway. It was also shown recently that XWnt5A and rFz2, in a heterologous assay, increase intracellular calcium via G proteins and phosphoinositol signaling. A mutation in the beta-subunit of a heterotrimeric G protein in C. elegans prevents correct spindle orientation, a process that is believed to be dependent on a Wnt and a Fz receptor, but not on Arm. Further studies regarding a possible involvement of PI-3K and G proteins in planar polarity signaling may provide additional insight to the diversity of Fz-related signaling pathways (Boutros, 1998 and references).

Cdc42 and Rac1 contribute differently to the organization of epithelial cells in the Drosophila wing imaginal disc. Drac1 is required to assemble actin at adherens junctions. Failure of adherens junction actin assembly in Drac1 dominant-negative mutants is associated with increased cell death. Dcdc42, on the other hand, is required for processes that involve polarized cell shape changes during both pupal and larval development. In the third larval instar, Dcdc42 is required for apico-basal epithelial elongation. Whereas normal wing disc epithelial cells increase in height more than twofold during the third instar, cells that express a dominant-negative version of Dcdc42 remain short and are abnormally shaped. Dcdc42 localizes to both apical and basal regions of the cell during these events, and mediates elongation, at least in part, by effecting a reorganization of the basal actin cytoskeleton. These observations suggest that a common Cdc42-based mechanism may govern polarized cell shape changes in a wide variety of cell types (Eaton, 1995).

The wing of Drosophila is covered by an array of distally pointing hairs. A hair begins as a single membrane outgrowth from each wing epithelial cell, and its distal orientation is determined by the restriction of outgrowth to a single distal site on the cell circumference. The roles of Cdc42 and Rac1 were examined in the formation of wing hairs. Cdc42 is required for localized actin polymerization in the extending hair. Interfering with Cdc42 activity by expression of a dominant negative protein abolishes both localized actin polymerization and hair outgrowth. In contrast, Rac1 is important for restricting the site at which hairs grow out. Cells expressing the dominant negative Rac1N17 fail to restrict outgrowth to a single site and give rise to multiple wing hairs. This polarity defect is associated with disturbances in the organization of junctional actin and also with disruption of an intricate microtubule network that is intimately associated with the junctional region. Apical junctions and microtubules are also involved in structural aspects of hair outgrowth. During hair formation, the apical microtubules that point distally elongate and fill the emerging wing hair. As the hair elongates, junctional proteins are reorganized on the proximal and distal edges of each cell (Eaton, 1996).

The Drosophila myoblast city (mbc) locus was previously identified on the basis of a defect in myoblast fusion. The mbc transcript and its encoded protein are expressed in a broad range of tissues, including somatic myoblasts, cardial cells, and visceral mesoderm. The gene is also expressed in the pole cells and in ectodermally derived tissues, including the epidermis. Consistent with this latter expression, mbc mutant embryos exhibit defects in dorsal closure and cytoskeletal organization in the migrating epidermis. Both the mesodermal and ectodermal defects are reminiscent of those induced by altered forms of Drac1 and suggest that mbc may function in the same pathway. MBC bears striking homology to human DOCK180, which interacts with the SH2-SH3 adapter protein Crk and may play a role in signal transduction from focal adhesions. Taken together, these results suggest the possibility that MBC is an intermediate in a signal transduction pathway from the rho/rac family of GTPases to events in the cytoskeleton and that this pathway may be used during myoblast fusion and dorsal closure (Erickson, 1997).

mbc mutants have been identified as suppressors of a rac1 overexpression phenotype. To determine whether Rac and Cdc42 GTPases can disrupt eye development, transgenic flies were generated in which wild-type Rac1, Rac2, and Cdc42 GTPases were expressed in the developing eye under the control of the synthetic glass multimer reporter (GMR). Flies harboring a single copy of the rac1 transgene exhibit an externally rough eye, and retinal sections reveal a loss of pigment cells and a disruption of the normal ommatidial morphology, with occasional loss of photoreceptors. With two copies of the GMR-rac1 transgene, a complete disruption of normal eye structure is observed; a similar phenotype is observed in GMR-rac2 transgenic flies. The GMR-cdc42 transgenic flies exhibit externally rough eyes distinct from those seen in the rac1 and rac2 transgenics. Retinal sectioning reveals missing photoreceptors and a disruption of ommatidial morphology. Although the cdc42-induced eye phenotype somewhat resembles the GMR-rho1 phenotype, the cdc42 transgenics also exhibit an abnormal rhabdomere morphology. The postmitotic elongation event that establishes the depth of the retina was also examined. Overexpression of either rho1 or cdc42, but not rac1, disrupts the normal elongation of all retinal cells. Thus, each of these members of the Rho GTPase family, when overexpressed, induces distinct alterations of normal eye development (Nolan, 1998).

To identify specific components of a Rac1 signaling pathway in Drosophila, rac1 transgenic flies were used to screen for dominant mutations that specifically suppress the rac1-induced rough eye phenotype but not that caused by GMR-cdc42 or GMR-rho1. Chromosomal deficiences that cover either rac1 or rac2 each suppress the GMR-rac1 eye defect, confirming that the phenotype is sensitive to the levels of endogenous Rac activity and that Rac1 and Rac2 are normally expressed during eye development. To identify rac1-suppressing mutations, mutagenized wild-type males were mated with GMR-rac1 females and the resulting F₁ progeny were examined for suppression of the rough eye phenotype. A total of 23,000 F₁ flies were screened, and 36 dominantly suppressing mutations were identified. Three complementation groups were established on the basis of lethality, and a single complementation group of 11 alleles has been termed Suppressor of rac1 [Su(rac)1]. Each of the Su(rac)1 alleles dominantly suppresses the GMR-rac1-induced rough eye surface as well as the underlying retinal morphology, rescuing the percentage of normal appearing ommatidia from 3% in GMR-rac flies to 97% in GMR-rac1/Su(rac)1 flies. Each of these alleles also suppresses the GMR-rac2-mediated defect, although none suppresses a GMR-rho1 phenotype. These data suggest that Su(rac)1 encodes a specific component of a Rac-mediated signaling pathway (Nolan, 1998).

Because a specific requirement for Rac activity, but not that of Cdc42, has been demonstrated in the fusion of myoblasts during muscle development (Luo, 1994), the musculature of Su(rac)1 mutants was examined. Myoblast fusion is normally completed by stage 15; however, in stage 15 Su(rac)1 mutants, myoblasts are largely unfused. Meiotic mapping localized Su(rac)1 alleles to a chromosomal region similar to that of a previously reported gene, mbc, that is also associated with a loss-of-function myoblast fusion defect. Null alleles of mbc fail to complement the lethality and myoblast fusion phenotype of several alleles of Su(rac)1. Moreover, mbc alleles also suppress the GMR-rac1 phenotype. Together, these results indicate that Su(rac)1 is allelic to mbc. Although the role of Rac in myoblast fusion is unknown, these results suggest the Mbc mediates the activity of Rac in this morphogenetic process in which actin rearrangements have been implicated previously (Nolan, 1998).

Other phenotypes were examined that would be consistent with aberrant Rac signaling. Drosophila Rac1 has been implicated in axonal outgrowth (Luo, 1994), and mbc mutants exhibit a low penetrance defect in the fasciculation of axons of the ventral nerve cord neurons. Specifically, some mbc mutant embryos exhibit improper spacing between commissures and, in extreme cases, a lack of fasciculation of the longitudinal connectives, possibly because of abnormal migration of the central nervous system (CNS) neurons across the ventral surface. In support of a role for Mbc in cell migration is the recent observation that mutations in ced-5, the C. elegans homolog of mbc, result in defective migration of the distal tip cells of the gonad. Additionally, mutations in mig-2, a C. elegans gene encoding a Rac-like GTPase, also affect distal tip cell migration and axon outgrowth. Moreover, the mammalian Rac GTPase appears to regulate the motility of cultured fibroblasts. It is possible that a pathway mediated by both Rac and Mbc regulates neuronal migration and axon growth, and may explain the CNS defects observed in mbc mutant embryos (Nolan, 1998 and references).

Cell transfection and in vitro nucleotide exchange assays with each DH domain of human Trio have suggested that GEF1 preferentially activates Rac, whereas GEF2 activates Rho (Debant, 1996; Bellanger, 1998a). To explore the relationship between trio and Rac in Drosophila, the compound eye was used as an established system to test for genetic interactions in GTPase signaling pathways. Rac overexpression under the control of the eye-specific promoter GMR creates a mispatterened 'rough' eye in which individual ommatidia are misshapen. However, removal of a single copy of trio causes a dramatic suppression of this Rac gain-of-function phenotype. This is true for additional trio alleles. In contrast, overexpression of Rho also generates a rough eye, but this phenotype is not significantly altered by reduction in trio activity. This suggests that in the Drosophila retina, trio functions to activate one or more of the Drosophila Rac-like genes but not Rho (Bateman, 2000).

In the embryo, previous studies have shown that the same nerve branches affected by trio mutations are also most sensitive to Rac perturbation. Although occasional ISNb stop short phenotypes are observed, the predominant ISNb and SNa bypass phenotypes induced by Drac1^N17 overexpression are distinct from phenotypes caused by loss of trio function. This difference likely reflects Drac1^N17 interference with multiple neural activators of Rac GTPases. However, it was reasoned that if trio is involved in Rac activation in the embryonic motor nervous system, the penetrance of the Drac1^N17 phenotype should be sensitive to changes in the genetic dose of trio. Consistent with this hypothesis, removal of a single copy of trio in embryos expressing Drac1^N17 causes a distinct increase in the penetrance of ISNb bypass. This was true for all alleles of trio tested. Moreover, coexpression of Drac1^N17 and a wild-type trio transgene results in a dramatic suppression of the ISNb bypass phenotype, consistent with the model that trio is an activator of Rac GTPases in the embryonic motor nervous system (Bateman, 2000).

Despite phenotypic differences between Drac1^N17 and trio in the motor nervous system, analysis of the CNS in embryos lacking Rac function reveals defects identical to those observed in trio mutants. Specifically, expression of the dominant-negative Drac1^N17, under the control of the neural-specific GAL4 driver C155, causes a failure of the lateralmost Fas II-positive longitudinal pathway to properly connect at stage 17 (13.7%). In contrast, neural expression of either Dcdc42^N17 or DRho1^N19 does not cause defects in longitudinal pathfinding, indicating that the CNS phenotype is specific to interference with Rac-like GTPase function. Thus, in the CNS, mutant phenotypes of Drac1^N17 and trio are consistent with disruption of a common pathway (Bateman, 2000).

Rhodopsin is essential for photoreceptor morphogenesis; photoreceptors lacking rhodopsin degenerate in humans, mice, and Drosophila. Transgenic expression of a dominant-active Drosophila Rho guanosine triphosphatase, Rac1, rescues photoreceptor morphogenesis in rhodopsin-null mutants; expression of dominant-negative Rac1 results in a phenotype similar to that seen in rhodopsin-null mutants. Rac1 is localized in a specialization of the photoreceptor cortical actin cytoskeleton, which is lost in rhodopsin-null mutants. Thus, rhodopsin appears to organize the actin cytoskeleton through Rac1, contributing a structural support essential for photoreceptor morphogenesis (Chang, 2000).

Sensory neurons present a challenge for morphogenesis: to harness the generic mechanisms of the cytoskeleton to shape a cell to the needs of its specific sensory protein. For photoreceptors, it is clear that morphogenesis and maintenance of the photosensitive organelle (in Drosophila, the rhabdomeres; in vetrebrates, the outer segments rods and cones) depends on the organelle's sensory protein: rhodopsin. Rhabdomeres and outer segments are orderly stacks of photosensitive plasma membrane organized from enormously expanded apical cell surfaces. The forces that constrain this expansion and organize it into a dense stack are incompletely understood, but the cortical actin cytoskeleton and its associated proteins are substantial contributors. It has been suggested that in addition to its sensory role, Drosophila rhodopsin organizes the cortical actin cytoskeleton into an essential morphogenetic constraint: the rhabdomere terminal web (RTW). The RTW defines the regular, curving base of the rhabdomere that partitions the rhabdomere from the photoreceptor cytoplasm. In rhodopsin-null mutants, the rhabdomere base fails to organize correctly, and the rhabdomere collapses deep into the photoreceptor cytoplasm in convoluted sheets of apposed membrane (Chang, 2000 and references therein).

A chimeric protein that decorates F-actin with green fluorescent protein (GFP) reveals the RTW as bundled microfilaments extending from the rhabdomere base deep into the photoreceptor. Before rhodopsin expression, the RTW of developing photoreceptors shows less microfilament bundling, resembling a house painter's brush. At about 90% of pupal development (pd), after the onset of rhodopsin expression at 75% pd, RTW microfilaments elongate commensurate with the increasing microvillar length and gathered into bundles (Chang, 2000).

RTW maturation and rhabdomere morphogenesis fail in photoreceptors lacking rhodopsin. Paralleling the normal initiation of microvillar organization observed in rhodopsin-null mutants, the RTW of mutant photoreceptors appears normal before the time when rhodopsin expression would normally commence. The RTW growth and bundling that normally follow rhodopsin expression fail in rhodopsin-null photoreceptors. Unlike wild-type rhabdomeres, the smaller, flattened rhabdomeres formed in the rhodopsin-null mutant collapse into the photoreceptor cytoplasm in convoluted sheets of apposed membrane during the first day after eclosion. The actin cytoskeleton becomes thoroughly disorganized in the absence of rhodopsin (Chang, 2000).

Although rhodopsin contributes about 50% of rhabdomere membrane protein, it is unlikely to support morphogenesis by a simple mass effect. Smaller, but ultrastructurally normal rhabdomeres form in mutants in which Rh1 is reduced by over 99%. Furthermore, a pulse of rhodopsin expression restricted to a narrow window of development is sufficient to rescue rhabdomere morphogenesis in photoreceptors otherwise lacking rhodopsin. It is proposed that an additional role for rhodopsin is to contribute an activity required to organize the RTW into an effective subapical barrier (Chang, 2000).

Drosophila Rac1, localizes to the rhabdomere base beginning with the onset of microvillar organization during midpupal development; it remaines subapical in adult eyes. To explore potential Rac1 functions in rhabdomere morphogenesis, dominant-negative N17Rac1 was expressed at defined stages of eye development. N17Rac1 expression during rhabdomere morphogenesis leads to reduced, disordered rhabdomeres. Fewer microvilli are seen in cross section, and a well-defined rhabdomere base is not formed; apposed sheets of rhabdomere membrane involute into the photoreceptor cytoplasm. Although these defects are reminiscent of those seen in rhodopsin-null mutants, the phenotype is not a consequence of a failure of rhodopsin delivery to the rhabdomeres. The actin cytoskeleton, however, appears diffuse and disordered as a result of transgene expression (Chang, 2000).

The resemblance of the rhabdomere base defects caused by N17Rac1 to those seen in rhodopsin-null mutants suggests that rhodopsin might exert its structural effect through Rac1. If so, it was reasoned that expression of constitutively active V12Rac1 might rescue rhabdomere morphogenesis in photoreceptors lacking rhodopsin. To test this idea, V12Rac1 was expressed during rhabdomere morphogenesis in ninaEI17 mutants that lack rhodopsin in photoreceptors R1 to R6. Substantial rescue of rhabdomere morphogenesis was observed. Occasional loops of rhabdomere membrane intrude into the photoreceptor, but most terminate at a well-defined base. The RTW is more tightly organized in V12Rac1-expressing animals. Similar to rhodopsin-null rhabdomeres rescued by a pulse of rhodopsin expression, V12Rac1-expressing animals show substantial rescue 5 days after eclosion. Thus, V12Rac1 appears to supply a durable organizing activity lost in rhodopsin-null mutants (Chang, 2000).

Similar to the requirement of small amounts of rhodopsin for normal morphogenesis, rescue appears quite sensitive to Rac1V12. About 18% of R1 to R6 rhabdomeres are rescued in non-heat-shocked hsGAL/SM1;UAS-Rac1V12 ninaEI17 eyes, rising to 90% in animals heat-shocked at 80% pd. Substantial rhabdomeres and tighter organization of the RTW are evident. To examine rescue specificity among Rho small GTPases, constitutively active V12Cdc42 and V14Rho were overexpressed in ninaE-null mutants. V12Cdc42 rescues rhodopsin-null morphogenesis, but V14Rho does not. Neither Rho nor Cdc42 immunolocalizes to the RTW of normal flies, but Cdc42 has been found to activate Rac in other systems and may do so here (Chang, 2000).

The observations reported here suggest that Rac1 links rhodopsin to photoreceptor morphogenesis: Targeted delivery of rhodopsin to the developing rhabdomere promotes localized Rac1 activity that, in turn, orchestrates assembly of the RTW. In rhodopsin-null photoreceptors, failure to correctly organize the RTW, likely including a failure of microfilament cross linking, would allow sheets of self-adhesive rhabdomere membrane to intrude unopposed into the photoreceptor cytoplasm. How rhodopsin contributes to Rac1 activity, as well as its downstream effectors, remains to be determined. Two attractive effector candidates are nonmuscle myosin II and moesin, which localize to the base of the developing rhabdomere and which, in other systems, lie downstream of small GTPases (Chang, 2000).

An actin barrier may also shape vertebrate photoreceptors, constraining newly added photosensitive membrane to the outer segment. Actin and actin-associated proteins localize to the site of outer segment disc membrane evagination, and nascent outer segment disc membrane intrudes into the cytoplasm of rabbit photoreceptors exposed to cytochalasin D. Given the several parallels between vertebrate and Drosophila retinal development and the highly conserved mechanisms of the cytoskeleton, it is interesting to speculate that vertebrate rhodopsin may also regulate the photoreceptor cytoskeleton. It is possible that some mutant rhodopsins, including those causing human retinitis pigmentosa, may result in photoreceptor degeneration because of an inability to correctly organize the actin cytoskeleton (Chang, 2000).

Rac and axon guidance

Putative constitutively active and dominant-negative Drac1 proteins were expressed in PNS tissues. In wild-type embryos, each segment contains three highly stereotyped clusters of PNS neurons connected by axon bundles. When expressed in neurons, Drac1 dominant negative mutant proteins cause axon outgrowth defects in peripheral neurons without affecting dendrites. Loss of axons between the dorsal and lateral clusters were observed. In addition to defects in axonal initiation Rac1 mutation also causes defects in axon elongation. Dominant negative Rac1 also causes abnormal neuronal accumulation of filamentous actin. When expressed in muscle precursors, altered Rac proteins cause complete failure of, or abnormality in, myoblast fusion. Expressions of analogous mutant Dcdc42 proteins cause qualitatively distinct morphological defects, suggesting that similar GTPases in the same subfamily have unique roles in morphogenesis (Luo, 1994).

Previous genetic studies of intersegmental nerve b (ISNb) development have identified several cell-surface proteins required for correct axon guidance to appropriate target muscles. Of all the proteins currently known to control ISNb guidance, Drac1 and Dlar are most likely to mediate target entry as opposed to defasciculation, because both display a parallel bypass phenotype. The small GTPase Drac1 plays a key role in this guidance process. Neuronal expression of the dominant negative mutation Drac1(N17) causes axons to bypass and extend beyond normal synaptic partners. GTPase mutations were placed under the control of the yeast transcriptional activator GAL4. Combination of a neuronal GAL4 'driver' with a GTPase cDNA 'reporter' under the control of the GAL4 upstream activator sequence (UAS) results in specific cDNA expression. Neuronal-specific expression of either Drac1(V12) or Dcdc42(V12) causes ISNb motor growth cones to arrest outgrowth just as axons begin to explore the periphery. This highly penetrant motor phenotype is analogous to the growth cone arrest observed in sensory neurons when they express the same GTPase mutations. SNa axons also require Rac to reach correct targets. The growth cone arrest seen in the V12 backgrounds suggests that hyperactivation of different GTPase pathways disrupts leading edge motility. This phenotype is consistently reproduced by pharmacological blockade of actin assembly, carried out by cytochalasin D treatment. Genetic interactions between Drac1(N17) and the receptor-tyrosine phosphatase Dlar suggest that ISNb guidance requires the integration of multiple, convergent signals. Double mutant Drac1(N17-Dlar compound mutants display a penetrance two- to three-fold higher than would be expected if the defects were simply additive. Thus, Rac function in ISNb axons is quite sensitive to the dosage of Dlar protein. This synergistic genetic interaction suggests that Rac and Dlar function together, though not in a simple linear pathway. The existence of multiple inputs during target entry may explain why null mutations in Dlar (and other choice point genes) are not completely penetrant on their own (Kaufmann, 1998)

Correct pathfinding by Drosophila photoreceptor axons requires recruitment of p21-activated kinase (Pak) to the membrane by the SH2-SH3 adaptor Dock. The guanine nucleotide exchange factor (GEF) Trio has been identified as another essential component in photoreceptor axon guidance. Regulated exchange activity of one of the two Trio GEF domains is critical for accurate pathfinding. This GEF domain activates Rac, which in turn activates Pak. Mutations in trio result in projection defects similar to those observed in both Pak and dock mutants, and trio interacts genetically with Rac, Pak, and dock. These data define a signaling pathway from Trio to Rac to Pak that links guidance receptors to the growth cone cytoskeleton. It is proposed that distinct signals transduced via Trio and Dock act combinatorially to activate Pak in spatially restricted domains within the growth cone, thereby controlling the direction of axon extension (Newsome, 2000).

The small GTPases Rac and Rho act as cellular switches in many important biological processes. In the fruit fly Drosophila, RhoA participates in the establishment of planar polarity, a process mediated by the receptor Frizzled (Fz). Thus far, analysis of Rac in this process has not been possible because of the absence of mutant Rac alleles. The roles of Rac and Rho in establishing the polarity of ommatidia in the Drosophila eye have been investigated. By expressing a dominant negative or a constitutively activated form of Rac1, Rac signaling is interfered with specifically and ommatidial polarity is disrupted. The resulting defects are similar to the loss/gain-of-function phenotypes typical of tissue-polarity genes. Through genetic interaction and rescue experiments involving a polarity-specific, loss-of-function dishevelled (dsh) allele, Rac1 has been found to act downstream of Dsh in the Fz signaling pathway, but upstream of, or in parallel to, RhoA. Rac signals to the nucleus through the Jun N-terminal kinase (JNK) cascade in this process. By generating point mutations in the effector loop of RhoA, it has been found that RhoA also signals to the nucleus during the establishment of ommatidial polarity. Nevertheless, Rac and RhoA activate transcription of distinct target genes. Thus Rac is specifically required downstream of Dsh in the Fz pathway. It functions upstream or in parallel to RhoA and both signal to the nucleus, through distinct effectors, to establish planar polarity in the Drosophila eye (Fanto, 2000).

Random mutagenesis of activated mammalian Rho^V14 has led to the identification of mutations in the effector loop (a portion of the GTPase responsible for interaction with several effectors) that block either its action on cytoskeletal dynamics or on transcriptional activation of SRF. The F39V mutation impedes the formation of actin stress fibers but does not interfere with the activation of SRF-mediated transcription, separating the two effects of Rho^V14. The mutation E40L interferes with both SRF activation and the formation of stress fibers (Fanto, 2000).

The relevant mutations were recapitulated in the activated Drosophila Rho^V14 protein and they were expressed under the control of sev-gal4 in the eye disc (sev;Rho^V14 F39V and sev;Rho^V14 E40L). The sev;Rho^V14 F39V flies display a phenotype that is indistinguishable from that of sev;rho^V14 alone, with loss of photoreceptors and misorientation of otherwise wild-type clusters. This was evident even when the transgene was expressed at lower levels. Increasing the expression levels of sev;Rho^V14 F39V (two copies) led to an enhancement of both the polarity and the photoreceptor recruitment phenotypes. In contrast, sev;Rho^V14 E40L flies never displayed polarity defects, both when the transgene was expressed at low and at very high levels. Nevertheless, this mutant maintained the ability of sev;Rho^V14 to cause photoreceptor loss: although a large number of ommatidia had lost several photoreceptors, all the remaining ommatidia with wild-type complement had the correct polarity. This indicates that removing the function required for nuclear signaling (equivalent to SRF activation in cell culture) eliminated the ability of sev;Rho^V14 to induce polarity defects, suggesting that nuclear signaling by RhoA is critical for ommatidial polarity determination (Fanto, 2000).

To better characterize nuclear signaling by Rac and RhoA, the expression of puckered (puc) and Delta (Dl) were studied. Dl is the only known transcriptional target of Fz signaling in R3, and puc-lacZ expression serves as a measure of JNK activity in vivo. The puc gene is a transcriptional target of JNK signaling in Drosophila, and encodes a dual specificity protein phosphatase that acts as a negative regulator of JNK itself in a feedback loop. In the wild type, very weak beta-galactosidase expression from the puc enhancer trap line is detectable in all photoreceptor precursors. Expression of sev;rac^V12 lead to strong upregulation of puc-lacz in one or, more frequently, two cells of the cluster, identified as R3/R4 precursor cells, consistent with the expression pattern of sev;Rac^V12. These data resemble the upregulation of puc-lacz when the JNK pathway has been activated in the same cells (Fanto, 2000).

In contrast, Rho^V14 affects puc-lacz expression differently. Although in sev;Rho^V14 eye discs puc-lacz expression is upregulated in some cells at a later stage, these were not identifiable as the R3/R4 pair, but were often found in the position of the R2/R5/R8 precursors (where sev is not expressed). This suggests that the effect seen is not a direct consequence of Rho activation, but more likely a secondary effect (RhoA^V14 E40L fails to induce significant puc-lacz expression). Thus, the direct transcriptional activation of puc-lacz in R3/R4 correlates with the genetic interactions with the JNK module, suggesting a difference in the action of Rac and RhoA (Fanto, 2000).

An important aspect of R3/R4 cell fate and ommatidial polarity determination is the upregulation of Dl expression in the R3 precursor by Fz. Dl then signals to Notch on the R4 precursor, resulting in the choice of the R4 cell fate. In addition to Fz, other components of the Fz/planar-polarity pathway have also been found to upregulate Dl transcription. Thus, whether Rac and RhoA also regulate Dl transcription was investigated by monitoring Dl-lacZ expression in sev;Rac^V12 and sev;Rho^V14 eye discs (Fanto, 2000).

In the wild type, Dl is expressed dynamically in photoreceptor precursors behind the furrow. Within the R3/R4 pair, it is expressed in R3 from rows 4 to 8, whereas it remains at lower levels in R4. In contrast to the difference in puc expression, both sev;Rac^V12 and sev;Rho^V14 upregulated Dl-lacz expression in both R3/R4 precursors. The RhoA^V14 E40L isoform that is impaired in nuclear signaling does not affect Dl expression, confirming the importance of nuclear signaling by RhoA. These effects are very similar to those of sev;Fz, supporting the idea that Rac and RhoA act downstream of Fz in the regulation of the R3/R4 cell fate. Their different effects on puc-lacz indicate that their downstream effectors in nuclear signaling are distinct (Fanto, 2000).

Recent studies have shown the small GTPases, Rac1, Rho, and CDC42, to have a role in axon guidance. To assess their participation in synapse assembly and function various forms of Drac1 have been expressed in the giant fiber system of Drosophila. Overexpression of wild-type Drac1 in the giant fiber (GF) leads to a disruption in axonal morphology; axons often terminate prematurely in a large swelling in the target area but lack the normal lateral bend where the synapse with the jump motor neuron would normally be found. Electrophysiological assays reveal longer latencies and lowering following frequencies indicating defects in the synapse between the GF and the tergotrochanteral motor neuron (TTMn). Thickened abnormal GF dendrites are also observed in the brain. Overexpression of the dominant-negative form of Drac1, (N17), results in axons that produced extra branches in the second thoracic neuromere (T2); however, the synaptic connection to the TTMn is present and functions normally. Conversely, expression of the constitutively active form, Drac1(V12), results in a complete lack of neurite outgrowth and this was also seen with overexpression of Dcdc42(V12). In the absence of a GF, these flies show no response in the jump (TTM) or flight (DLM) muscles upon brain stimulation. Taken together these results show that the balance of actin polymerization and depolymerization determines local process outgrowth and thereby synapse structure and function (Allen, 2001).

Growth, guidance and branching of axons are all essential processes for the precise wiring of the nervous system. Rho family GTPases transduce extracellular signals to regulate the actin cytoskeleton. In particular, Rac has been implicated in axon growth and guidance. Loss-of-function phenotypes of three Rac GTPases have been analyzed in Drosophila mushroom body neurons. Progressive loss of combined Rac1, Rac2 and Mtl activity leads first to defects in axon branching, then guidance, and finally growth. Expression of a Rac1 effector domain mutant that does not bind Pak rescues growth, partially rescues guidance, but does not rescue branching defects of Rac mutant neurons. Mosaic analysis reveals both cell autonomous and non-autonomous functions for Rac GTPases, the latter manifesting as a strong community effect in axon guidance and branching. These results demonstrate the central role of Rac GTPases in multiple aspects of axon development in vivo, and suggest that axon growth, guidance and branching could be controlled by differential activation of Rac signaling pathways (Ng, 2002).

The Drosophila genome has two Rac genes that share 92% amino acid sequence identity and have overlapping expression patterns. A highly related Mig-2-like (Mtl) gene, the ortholog of Caenorhabditis elegans mig-2, is present on the same chromosome. To isolate loss-of-function mutants of Rac1 and Rac2, small deficiencies were generated by means of imprecise excision of nearby P-elements. The Rac2^Delta excision disrupts only the Rac2 open reading frame (ORF), and hence is a Rac2-specific null mutation, but is homozygous viable. The Df(3)Rac1 excision disrupts the Rac1 ORF and two adjacent genes. Rac1 point mutations were then recovered from an ethylmethane sulphonate (EMS) screen for mutations that were lethal over Df(3)Rac1 in a Rac2^Delta homozygous background, but viable in a Rac2^Delta background. Sequence analysis and transgenic rescue have established that these mutants are recessive, loss-of-function Rac1 alleles (Ng, 2002).

Three Rac1 missense mutations were recovered, each altering an amino acid conserved within Rho GTPases. The strongest allele, Rac1^J11, changes glycine 60 to glutamate (Gly60Glu). Structural, biochemical and genetic criteria all indicate that this is a null allele. Structurally, Gly 60 forms a hydrogen bond with the gamma-phosphate of GTP, and is invariant in all members of the GTPase superfamily. Introduction of a glutamate is predicted to disrupt this interaction. Indeed an analogous Gly60Asp mutation in H-Ras disrupts the activity of both wild-type and constitutively active proteins. Biochemically, the Rac1^J11 mutation markedly impairs binding of GTP to Rac1 in vitro, reducing it to less than 10% compared with wild type. Genetically, homozygous Rac1^J11 phenotypes are indistinguishable from Rac1^J11/Df(3)Rac1 in both the mushroom bodies and visual system (Ng, 2002).

Mushroom body (MB) neurons of the Drosophila brain were used to examine the role of Rac in axon development. Adult MB neurons derive from four neuroblasts per brain hemisphere. Each neuroblast sequentially generates three classes of neurons with distinct patterns of axon projection. Each MB neuron sends a single primary neurite that gives rise to both dendritic branches and an axon. MB axons fasciculate tightly in the anteriorly projected axon peduncle. In the anterior brain, each early born class of gamma-neuron has one principal medial branch, whereas each neuron of the later-born alpha'/ ß' -or alpha/ß-classes has bifurcated axons with one dorsal (alpha' or alpha) and one medial (ß' or ß) projection. These axonal lobes can be distinguished using the fasciclin II (FasII) marker, which stains alpha/ß' -axons strongly, gamma-axons weakly, but does not stain alpha'/ß'-axons, cell bodies or dendrites of MB neurons (Ng, 2002).

Three sets of experiments were carried out to analyze Rac function in MB axon development. In the first set, by examining viable adults, gross axon defects were found in animals in which the whole brain was isogenic for particular Rac mutant combinations. In the second and third sets, the MARCM system was used to generate neuroblast and single-cell homozygous mutant clones that are positively labelled by MB Gal4-OK107 driven marker expression. These mosaic brains are unlikely to contain unlabelled clones outside the MB lineage, since clones were induced in newly hatched larvae, when proliferation is largely confined to the MB neuroblasts. Progressive defects were found in MB axon branching, guidance and growth as wild-type copies of Rac1, Rac2 and Mtl, referred to hereafter as Rac GTPases, were removed. Although all three Rac GTPases contribute to the fidelity of MB axon development, there is a differential dependence, with loss of Rac1 having the largest effect and loss of Rac2 the smallest effect (Ng, 2002).

Axon branching is most sensitive to loss of Rac GTPase activity. Defective branching included the generation of alpha- but not ß-axonal branches and vice versa. This was revealed either by FasII immunostaining or expression of a murine (m)CD8-green fluorescent protein (GFP) fusion protein in the whole MB or in neuroblast clones. Rac1^J11 heterozygotes exhibit significant branching defects. The percentage of branching defects increases as additional wild-type copies of Rac genes are removed. The absence of a specific axonal lobe might be caused either by a failure of individual axons to bifurcate, or by misguidance of the bifurcated branches. To distinguish between these possibilities, single-cell clones were generated in the later-born alpha/ß'-neurons. In all cases where FasII staining reveals a branching defect in the MB as a whole, only a single unbranched alpha-or ß-axon was detected for every alpha/ß'-neuron labelled. These axons extend normally and follow appropriate trajectories in the remaining lobe. These observations indicate that the lack of one particular axonal lobe, as observed in isogenic or largely isogenic mutant brains or neuroblast clones, is caused by a failure in axon branching (Ng, 2002).

Axon guidance displays an intermediate sensitivity to loss of Rac GTPases. Whereas Rac1^J11 heterozygotes exhibit mainly branching defects, in Rac1^J11 homozygotes, most MB axons are misguided. In mosaic animals, 21% of neuroblast clones homozygous for Rac1^J11 exhibited defective guidance. This percentage increased to 55% if the neuroblast clone was additionally homozygous for Rac2^Delta, and to 73% if the whole brain was also heterozygous for Mtl^Delta. The predominant guidance defects were caused by accumulation of most or all axons in a ball-like structure at the beginning of the axon peduncle that was intensely stained for FasII. By labelling single mutant gamma-neurons, it was found that individual axons tended to wind around the FasII-positive axon 'balls', similar to those seen in the neuroblast clones. Therefore, these axon balls, whether they are observed in isogenic mutant brains, neuroblast clones, or single neuron clones, are caused by misguidance of MB axons. Axon growth is least sensitive to loss of Rac GTPases. Defects in axon growth were found mainly in MB neurons homozygous for Rac1^J11;Rac2^Delta;Mtl^Delta. Ball-like axon accumulations were only rarely observed in this genotype. Instead, these axons terminated prematurely or failed to enter FasII-positive regions altogether. Most of the remaining axons exhibited severe guidance defects, extending their axons in a non-stereotypical fashion. To confirm this interpretation of a growth defect, single-cell clones homozygous for Rac1^J11;Rac2^Delta;Mtl^Delta were examined in brain hemispheres in which most of the FasII-positive gamma-axons (representing non-clonal tissue) were correctly patterned as in wild type. It was found that 55% of homozygous Rac1^J11;Rac2^Delta;Mtl^Delta single-cell clones exhibited axon-stalling defects, mostly at the peduncle, as compared with less than 5% in Rac1^J11;Rac2^Delta single-cell clones. These data indicate that Rac1, Rac2 and Mtl collaborate to control axon growth in a cell-autonomous manner (Ng, 2002).

The dendritic region of MB neuroblast clones homozygous for Rac1^J11;Rac2^Delta;Mtl^Delta still possessed mCD8-GFP-positive neurites, which were presumably contributed by the initial neurite outgrowth from the cell body and elaboration of MB dendrites. Quantitative analysis of single-cell Rac1^J11;Rac2^Delta;Mtl^Delta clones revealed significant reduction in both total dendritic length and number of dendritic segments per neuron, indicating that Rac GTPases are also required for dendritic growth and branching (Ng, 2002).

The differential sensitivity of axon growth, guidance and branching to loss of Rac function could in principle reflect the fact that axons must grow in order to be assayed for guidance and branching defects, and may need correct guidance to reach appropriate branching points. However, such a simple hierarchical model cannot explain the data. If Rac were equally required for growth, guidance and branching, and these processes were simply epistatic to one another as proposed by this heirarchical model, then growth defects should always be more frequent than guidance defects, which in turn should always be more frequent than branching defects. For example, if growth, guidance and branching were equally reduced by 20%, then in a population of 100 axons, 20 would show a growth defect, 16 would show a guidance defect (20% of the 80 axons that grow), and 13 would show a branching defect (20% of the 64 axons that grow and navigate correctly). However, this is not what was observed; instead, a shift was seen from branching to guidance to growth defects as the combined level of wild type Rac was progressively reduced. That axon branching can be selectively perturbed is best exemplified in genotypes involving a hypomorphic allele Rac1^J10. Furthermore, axon branching defects could be disrupted without any growth or guidance defects, whereas guidance could be disrupted without obvious growth defects. These data strongly suggest that axon growth, guidance and branching are separable events requiring increasing amounts of combined Rac GTPase activity in vivo (Ng, 2002).

One model that could account for these observations is that growth, guidance and branching use different Rac effector pathways. To test this idea, use was made of Rac effector domain mutants. The Rac Phe37Ala mutation (RacF37A) abolishes the ability of activated Rac to induce lamellipodia formation without affecting the Pak/JNK pathway. However, the Rac Tyr40Cys mutation (RacY40C) blocks Rac binding to effectors containing the 'CRIB' motif, including Pak, but does not affect lamellipodia formation. Overexpression of analogous Rac1F37A and Rac1Y40C mutants, or wild-type Rac1, did not disrupt MB axon patterning in a wild-type background. Therefore the MARCM system could be used to express these effector mutants specifically in Rac mutant clones to determine which effector pathways are required for MB axon growth, guidance and branching (Ng, 2002).

MB axon growth defects in single-cell Rac1^J11;Rac2^Delta;Mtl^Delta clones were largely rescued by transgenic expression of wild-type Rac1 or Rac1Y40C, but not Rac1F37A, indicating that direct binding of CRIB effector proteins is not required for Rac function in axon growth. To assess the effector pathways involved in guidance and branching, these same transgenes were expressed in Rac1^J11;Rac2^Delta neuroblast clones. In this background, 81% of axons show mutant phenotypes, predominantly guidance (55%) and branching (24%) defects. Expression of wild-type Rac1 markedly rescues these defects, reducing the fraction of abnormal axon phenotypes to 53%. The remaining branching and guidance defects were probably caused by influences of nearby non-clonal MB neurons heterozygous for Rac1^J11;Rac2^Delta, which exhibit a similar degree of branching and guidance defects. Neither Rac1Y40C nor Rac1F37A expression reduces the percentage of total axonal defects (78% and 88%, respectively). However, expression of Rac1Y40C (but not Rac1F37A) results in a marked shift in the distribution of axonal defects, with most showing branching (45%) rather than guidance (31%) defects. Thus, compared with wild-type Rac1, expression of Rac1Y40C in a Rac mutant background is able to rescue growth, partially rescue guidance, but is unable to rescue branching defects (Ng, 2002).

These results suggest that different downstream effector pathways mediate axon growth, guidance and branching. In particular, Rac binding of CRIB-domain effectors such as Pak is not required for axon growth, but may contribute to axon guidance and branching. This is consistent with genetic analyses indicating a requirement for Drosophila Pak in axon guidance but not growth. The mosaic analysis revealed an unexpected degree of cell non-autonomous effects in axon guidance and branching caused by defective Rac activity. If every MB axon were to choose its pathway independently, then in a brain hemisphere containing one homozygous Rac1^J11;Rac2^Delta neuroblast and three heterozygous neuroblasts, one would expect to observe a mixture of wild-type and mutant trajectories. Remarkably, all FasII-positive axons in the same hemisphere as the mutant neuroblast clone invariably exhibited the same guidance defect. This cannot be explained simply by the guidance defect caused by Rac1^J11;Rac2^Delta heterozygous neurons, since only 8% of such hemispheres exhibited this defect. Such a cell non-autonomous effect is also observed in axon branching. If every branching-defective alpha/ß-neuron were to make an independent decision to form either a single dorsal or medial branch, then in neuroblast clones where hundreds of axons are examined together, one would expect to see alpha thinning of both axonal lobes rather than the absence of a single lobe. Instead, in Rac1^J11;Rac2^Delta neuroblast clones that exhibited branching defects, all MB axons in the neuroblast clone projected either dorsally (one-third) or medially (two-thirds). Most non-clonal axons (as revealed by FasII staining) would always make the same choice as the mutant clone. Whereas homozygous mutant axons could induce their heterozygous neighbours to make the same errors, it is possible that heterozygous axons could reduce the error rate of nearby homozygous mutant axons. This could be one explanation for why the phenotypes of homozygous mutant animals are of higher penetrance than those of homozygous mutant neuroblast clones (Ng, 2002).

These observations suggest a marked community effect in MB axon guidance and branching: axons of mixed genotypes make their choices together. In any given animal, the collective choice of a normal versus mutant projection is likely to be influenced by the severity of the genotype, and the relative number of homozygous versus heterozygous axons. Such collective decision making is probably a result of tight fasciculation among MB axons, which is not disrupted in Rac mutants. It will be interesting to test whether this community effect reflects a general feature of axon development in a complex central nervous system environment (Ng, 2002).

This analysis of Rac GTPases in MB axon development suggests a mechanistic link between axon growth, guidance and branching. Although there is evidence that axon growth and guidance have different cytoskeletal requirements, their connections are not well understood. Little is known about intracellular signaling mechanisms that regulate axon branching. This study shows that axon growth, guidance and branching require increasing amounts of combined Rac GTPase activity in vivo. The requirement of Rac GTPases for axon outgrowth and guidance in C. elegans has also been reported recently. It is proposed that axon branching, guidance and growth specified by extracellular cues require different amount of Rac GTPase activation in the growth cone, which in turn engage different downstream pathways to specify distinct cytoskeletal changes (Ng, 2002).

Rho family GTPases are ideal candidates to regulate aspects of cytoskeletal dynamics downstream of axon guidance receptors. To examine the in vivo role of Rho GTPases in midline guidance, dominant negative (dn) and constitutively active (ct) forms of Rho, Drac1, and Dcdc42 are expressed in the Drosophila CNS. When expressed alone, only ctDrac and ctDcdc42 cause axons in the pCC/MP2 pathway to cross the midline inappropriately. Heterozygous loss of Roundabout enhances the ctDrac phenotype and causes errors in embryos expressing dnRho or ctRho. Homozygous loss of Son-of-Sevenless (Sos) also enhances the ctDrac phenotype and causes errors in embryos expressing either dnRho or dnDrac. CtRho suppresses the midline crossing errors caused by loss of Sos. CtDrac and ctDcdc42 phenotypes are suppressed by heterozygous loss of Profilin, but strongly enhanced by coexpression of constitutively active myosin light chain kinase (ctMLCK), which increases myosin II activity. Expression of ctMLCK also causes errors in embryos expressing either dnRho or ctRho. These data confirm that Rho family GTPases are required for regulation of actin polymerization and/or myosin activity and that this is critical for the response of growth cones to midline repulsive signals. Midline repulsion appears to require down-regulation of Drac1 and Dcdc42 and activation of Rho (Fritz, 2002).

Thus, when expressed alone, only ctDrac and ctDcdc42 cause midline crossing errors. However, the mutant GTPases interact genetically with mutations in robo, Sos, and chic and with overexpression of ctMLCK. The interactions are surprisingly specific. Midline crossing errors caused by expression of ctDrac or ctDcdc42 are suppressed by heterozygous loss of Profilin and enhanced by expression of ctMLCK. These results indicate that Drac1 and Dcdc42 encourage axons to cross the midline by regulating actin polymerization and/or myosin activity. CtRho and dnRho interact strongly with expression of ctMLCK or heterozygous loss of Robo, which suggests that regulation of myosin activity by Rho is crucial for midline repulsion. This work demonstrates that Rho, Drac1, and Dcdc42 are involved in dictating which axon may cross the midline, presumably by aiding in the transduction of attractive and/or repulsive cues operating at the midline. By using mutations in signaling molecules known to prevent pCC/MP2 axons from crossing the midline, this analysis concentrates on how Rho, Drac1, and Dcdc42 may regulate cytoskeletal dynamics in response to midline repulsive cues (Fritz, 2002).

The Rho family of GTPases was first studied in fibroblasts where activation of Cdc42, Rac, or Rho results in production of filopodia, lamellapodia, and stress fibers, respectively. In wound-healing assays, Rac appears to control actin polymerization to provide the protrusive force needed for movement, while Cdc42 determines cell polarity to localize Rac activity to the leading edge of the cell. Rho seems to play a role in adhesion and spreading during cell migration. These same processes are involved in growth cone motility, which makes the Rho GTPases candidates for regulation of cytoskeletal dynamics during axon guidance. Experiments in neurons, both in vitro and in vivo, indicate that activation of Rac and/or Cdc42 increases axon outgrowth and this is opposed by activation of Rho, which leads to growth cone collapse or retraction. This is consistent with findings that expression of ctDcdc42 or ctDrac allow axons to ignore repulsive signals at the midline and continue extending across the midline (Fritz, 2002).

Rho family GTPases activate a number of effectors that may affect axon outgrowth by regulating adhesion, myosin force generation, and/or actin polymerization. The ctDrac- and ctDcdc42-induced midline crossing errors are suppressed by heterozygous loss of Profilin, an actin-binding protein, which stimulates actin polymerization. Since reducing actin polymerization partially rescues the ctDrac and ctDcdc42 phenotypes as well as errors caused by heterozygous loss of Robo, it is likely that the midline crossing errors are caused by excessive actin polymerization. Increased actin polymerization may produce more filopodia to explore the midline, which leads to midline crossing. There are several pathways through which Drac1 and Dcdc42 might affect actin polymerization. The Cdc42/ Rac effector p21-activated kinase (PAK) activates LIM kinase to phosphorylate cofilin, an actin-depolymerizing factor required for neurite outgrowth. Cdc42 also activates actin polymerization through WASP, which stimulates polymerization by binding to the Arp2/3 complex. The activation of WASP by Cdc42 is enhanced by Profilin, which may explain why the suppression of the ctDcdc42 phenotype is stronger than that of the ctDrac-induced errors. However, actin polymerization may not be the only process regulated by Rho family GTPases to increase outgrowth (Fritz, 2002).

The interactions between the Drac1 and Dcdc42 and ctMLCK indicate that misregulation of myosin activity may contribute to ctDrac- and ctDcdc42-induced axon guidance errors. Coexpression of ctMLCK with ctDrac or ctDcdc42 results in a strong enhancement of midline crossing errors, while expression of dnDrac or dnDcdc42 suppresses the defects caused by increased myosin activity. This suggests that Drac1 and/or Dcdc42 activate myosin activity in the growth cone to increase outgrowth. One mechanism may be through activation of PAK, which leads to phosphorylation of myosin regulatory light chains (MLC) to increase myosin activity. However, it has been shown that PAK also phosphorylates and inactivates MLCK, resulting in less myosin activity. In vitro, PAK phosphorylates MLCK at serine 439, which is present in ctMLCK, and serine 991, which has been removed from ctMLCK, so the impact of this pathway on the truncated ctMLCK protein is uncertain. Alternatively, it is possible that the interaction of Drac1 or Dcdc42 and ctMLCK is a secondary effect to increased actin polymerization. If increased actin polymerization is causing more filopodial exploration of the midline, increasing myosin activity through ctMLCK expression could cause axons to cross the midline before they can retract filopodia encountering repulsive signals. Separating the relative contributions of Drac1 and Dcdc42 to actin polymerization and myosin activity will require more specific experiments involving the effectors of Drac1 and Dcdc42 (Fritz, 2002).

The data suggest that Drac1 and Dcdc42 activation must be prevented or reduced for axons to respond to repulsive signals at the midline. The midline crossing errors seen in Drac1 mutants are strongly enhanced by a partial loss of Robo, which suggests that midline repulsion requires a down-regulation of Drac1 activity. Down-regulation of Rac activity occurs in response to other repulsive signals, such as Ephrin and Semaphorin. A mechanism for this is suggested by experiments showing that Plexin-B, the receptor for Semaphorin, binds specifically to activated Rac, most likely to prevent it from activating effectors. Experiments in cell culture systems have confirmed that Robo-mediated signaling involves down-regulation of Cdc42. Activation of Robo by Slit recruits srGAP1 to the CC3 domain of Robo’s cytoplasmic tail, where it interacts with and inactivates Cdc42. Although srGAP1 does not affect the activity of Rac, srGAP2 and srGAP3 also bind to Robo, and one of these may regulate Rac activity. Down-regulation of Cdc42 and Rac by Robo-dependent repulsive signals is consistent with recent experiments showing that activation of DCC by chemoattractive Netrins stimulates neurite outgrowth and results in activation of Cdc42 and Rac1. Together, these data and the literature led to the hypothesis that Robo prevents axons from crossing the midline by decreasing Drac and Dcdc42 activity so that actin polymerization and myosin force generation are reduced (Fritz, 2002).

Down-regulation of Dcdc42 and Drac1 by Robo may also repel axons by preventing coupling of the actin cytoskeleton to the substrate. Rac is required for localization of E-Cadherin to cell-cell contacts and recruiting actin to Cadherin binding sites. Cdc42 and Rac promote Cadherin-mediated adhesion by preventing IQGAP, a CaM-binding Ras GAP, from interfering with the interaction of ß-catenin with alpha-catenin. Integrin-mediated adhesion also involves signaling through Rho family GTPases. By reducing actin and myosin dynamics and decoupling the cytoskeleton from the substrate, downregulation of Drac and Dcdc42 by repulsive guidance receptors would prevent axons from extending across the midline (Fritz, 2002).

The role of Rho in midline repulsion is more difficult to determine since both dnRho and ctRho enhance the midline crossing phenotype of heterozygous robo mutants. This is consistent with the data in which both dnRho and ctRho enhance the ctMLCK phenotype. Similar complexities are seen in the literature; expression of a Rho GEF, which is expected to increase Rho activity, leads to increased attraction to the midline, even though activation of Rho usually leads to growth cone collapse or retraction. The complexity of the Rho interactions is understandable when the dual role of myosin activity during axon guidance is considered. The most documented connection between myosin activity and Rho is through the effector Rho Kinase (RhoK). RhoK phosphorylates MLC and also inactivates myosin phosphatase by phosphorylating its myosin binding subunit, leading to increased phosphorylation of MLC and therefore increased myosin activity. Myosin activation is needed both for the retrograde flow of actin that retracts filopodia and for the force that propels the growth cone forward. Repulsive guidance signals are expected to increase retrograde flow while preventing forward movement (Fritz, 2002).

Clearly, regulation of Rho family GTPase activity is necessary to prevent axons from crossing the midline inappropriately. Midline repulsive signaling involves regulation of all three GTPases; Drac1 and Dcdc42 are likely downregulated, while Rho seems to be activated downstream of repulsive signals. The Rho family GTPases influence actin polymerization and/or myosin force generation to regulate the processes of growth cone motility that are required for proper response to axon guidance signals (Fritz, 2002).

Rho family small GTPases are thought to be key molecules in the regulation of cytoskeletal organization, especially for actin filaments. In order to examine the functions of Rac1 and Cdc42 in axon guidance at the midline of the central nervous system in Drosophila embryos, Rac1 and Cdc42 were either activated or inactivated in all postmitotic neurons. The phenotypes of Cdc42 activation and Rac1 inactivation were similar to those of roundabout mutants, in that many extra axons crossed the midline. Rac1 inactivation is dominant over Roundabout receptor activation. These observations indicate that Rac1 and Cdc42 have distinct functions in downstream signalling events triggered by Roundabout receptors. In order to further examine the functional difference between Rac1 and Cdc42 in the growth cone morphogenesis, primary embryonic cultures were used to closely observe neurite formation. Activation of Rac1 and Cdc42 is shown to have distinct effects on neurite formation, particularly on growth cone morphology and the actin filaments within. Both Rac1 and Cdc42 activation induced large growth cones and long filopodia, but Cdc42 did so more efficiently than Rac1. Only Rac1 activation, however, induced thick actin bundles in the filopodia. A clear difference was found between Rac1 and Cdc42 in terms of the response to an inhibitor of actin polymerization. These results suggest that Cdc42 is specifically involved in the regulation of actin filaments in growth cones, whereas Rac1 is involved in additional functions (Matsuura, 2004).

The Drosophila HEM-2/NAP1 homolog KETTE interacts with Dock and Rac to control axonal pathfinding and cytoskeletal organization.

In Drosophila, the correct formation of the segmental commissures depends on neuron-glial interactions at the midline. The VUM midline neurons extend axons along which glial cells migrate in between anterior and posterior commissures. The gene kette (correctly termed Hem-protein, or simply Hem) is required for the normal projection of the VUM axons and interference with kette function disrupts glial migration. In spite of the fact that glial migration is disrupted in kette mutants, both the axon guidance and glial migration phenotypes have their origin in midline neuron expression and not in midline glial expression. Axonal projection defects are found for many moto- and interneurons in kette mutants. In addition, kette affects the cell morphology of mesodermal and epidermal derivatives, which show an abnormal actin cytoskeleton. The Hem/Kette protein is homologous to the transmembrane protein HEM-2/NAP1 (Nck-associated protein) evolutionary conserved from worms to vertebrates. In the CNS, the membrane protein Kette could be participating directly in the neuron-glial interaction at the midline, where it could act as a signal to direct glial migration. Alternatively, Kette could serve as a receptor of possibly glial-derived signals during VUM growth cone guidance. The experimental data suggest that Kette transduces information to the neuronal cytoskeleton, which is in agreement with a receptor function (Hummel, 2000).

The vertebrate homolog of KETTE has been shown to interact with the first SH3 domain of the Nck adapter protein (Kitamura, 1996). The Drosophila homolog of Nck is encoded by dreadlocks (Garrity, 1996). dock was identified in a screen for mutations affecting axonal pathfinding and targeting of the adult photoreceptor neurons. In wild-type third instar larvae, the different photoreceptor cells stop their axonal growth in two distinct layers of the optic lobe, the lamina and the medulla. In contrast, dock mutant photoreceptor cells fail to establish this specific targeting, leading to a disruption of the lamina neuropile organization. In ~70% of the third instar larvae homozygous for the hypomorphic kette^Delta2-6 allele (n = 25), a weak disorganization was found of the lamina plexus and an abnormal bundling of R-cell axons in the medulla. The remaining larvae showed a stronger disorganization of the R-cell axons (20%) or were indistinguishable from wild type (10%). Further reduction of the kette gene function results in an enhancement of this axonal phenotype in 50% of the analyzed transheterozygous mutant kette larva. If one copy of dock is removed in the background of a hypomorphic kette mutation, a considerable enhancement of the larval projection phenotype is observed in 60% of the individuals. In addition, a significant enhancement of the homozygous dock phenotype is observed when removing one copy of kette in a dock mutant background (Hummel, 2000).

A reduction in the size of the longitudinal connectives in the embryonic CNS is observed in dock mutants. This phenotype resembles a hypomorphic kette connective phenotype. In mutant dock^P2 embryos, commissure separation is also affected, comparable with the hypomorphic phenotype seen in kette^J1-70 embryos. In correlation with the commissural phenotype, the VUM axons do not project properly in mutant dock embryos (Hummel, 2000).

In summary, these data show that both kette and dock mutants genetically interact and share a number of phenotypic traits. This suggests that these genes might be acting in the same genetic pathway during axonal pathfinding (Hummel, 2000).

The Rho family of small GTPases constitutes important regulatory factors also interacting with Nck. To analyze the functional interaction of KETTE with members of the Rho family, both the activated as well as the dominant-negative versions of Cdc42 and Rac1 were expressed in the midline cells of wild-type and kette mutant embryos. Expression of both of these mutant proteins in all midline cells using the simGAL4 driver line results in similar axonal defects. The projection of the VUM neurons resembles the phenotype observed in kette mutant embryos. In addition, the cell bodies of the VUM neurons appear sometimes displaced. In stage 16 embryos, the segmental commissures appear fused, which again indicates the importance of the midline neurons for the migration of the midline glia. Only a weak commissural disorganization is observed when the different Cdc42 or Rac1 proteins are expressed in the midline glial cells only. In all experiments, the expression of Rac1 appears to have more pronounced effects on the axonal morphology (Hummel, 2000).

To further test the interaction of kette and Rac1, activated Rac1 was expressed in all midline cells of mutant kette^J4-48 embryos. The commissures appeared separated, indicating that the midline glial cells are able to migrate between anterior and posterior commissures. Concomitantly, the connectives are further distant from the midline), indicating that expression of activated Rac1 can partially rescue the kette phenotype (Hummel, 2000).

Among others, the Nck adapter protein transduces signals via CDC42 and Rac1 to the Actin cytoskeleton. A GFP-moesin transgene was used to analyze the Actin cytoskeleton of mutant kette embryos. This protein binds to the F-actin fibers and thus allows a determination of their subcellular distribution using confocal microscopy. In wild-type embryos, F actin is found in axonal processes that are arranged in the typical ladder-like pattern. Prominent expression is also detected in the epidermis and the somatic musculature. In similar focal planes, kette embryos appear very different. Within the CNS, the typical fused commissure phenotype of mutant kette embryos is evident. Frequent intense granular staining is observed in the CNS and in the lateral body wall. Furthermore, the regular appearance of the cytoskeleton is disrupted in both mesoderm and ectoderm. In a tangential section of the dorsal epidermis, individual cells can be seen in wild-type embryos. Some cells form hairs, characterized by thin F-actin bundles. In mutant kette^C3-20 embryos, pronounced F-actin bundles are found, which often have a wavy appearance. In addition, the cortical actin cytoskeleton appears to stain weaker compared with wild-type embryos (Hummel, 2000).

Thus, mutations in kette affect the organization of the cytoskeleton. kette is expressed in neurons and is needed for correct axonal pathfinding. The KETTE protein seems to interact with the SH2-SH3 adapter Dock and at least part of the kette function might be mediated via small GTPases such as Rac1 (Hummel, 2000).

In addition to Kette function in axonal pathfinding, defects are observed in the morphology of trichomes and bristles in flies homozygous for the weak hypomorphic kette^Delta2-6 allele. Around 10% of the bristles appear wavy or do bend sharply and wing trichomes are enlarged and sometimes split. These phenotypes resemble those observed for mutations affecting the organization of the F-actin bundles or following expression of mutated Cdc42 or Rac1. Similarly, elevated levels of GTPase function in the developing eye cause late developmental defects as observed in hypomorphic kette mutations. cdc42 mutations have been isolated, but, presumably due to maternal contribution, loss of cdc42 function does not lead to an embryonic CNS phenotype. Both Cdc42 and Rac1 are important regulators of the actin cytoskeleton. The transduction of extracellular signaling to small GTPases is believed to involve Nck-type adapter proteins. Several phenotypic traits of kette are shared by mutations in the Drosophila gene dock, which encodes a Nck homolog. Furthermore, dock and kette genetically interact. The genetic data in combination with the kette loss-of-function and kette overexpression phenotypes led to the proposal of a model relating Dhem2/NAP1 function to cytoskeleton organization (Hummel, 2000).

The vertebrate KETTE homolog is HEM-2/NAP1 with 86% amino acid identity over the entire ORF, indicating that presumed protein-protein interactions are also conserved. To date, no hem-2 mutation has been described in vertebrates. The first SH3 domain of Nck was used to isolate Nck-associated proteins (NAP) and led to the identification of HEM-2/NAP1. Binding of HEM-2/NAP1 to Nck appears to be mediated by a 140-kD protein. Interestingly, in a screen for proteins interacting with activated Rac1, a complex consisting of HEM-2/NAP1 and a 140-kD protein was isolated. Thus, the 140-kD protein might be a novel adapter linking HEM-2/NAP1 signaling along two routes to the small GTPases. It will be of interest to identify Drosophila genes interacting with kette (Hummel, 2000 and references therein).

The Drosophila Nck homolog is encoded by dock. dock function appears highly specialized for growth-cone guidance since no mutant phenotypes have been reported in the mesoderm or the ectoderm. Because kette shows more pleiotropic defects, other adapter proteins may interact with the Kette protein (Hummel, 2000).

During axonal pathfinding, coordinated cytoskeletal remodeling occurs at the tip of the extending neurites, the growth cone. The Rho family of GTPases mediates the regulation of the reorganization of the actin cytoskeleton induced by extracellular signals: Cdc42, Rac1, and RhoA. In fibroblast cells, the different GTPases induce different cellular responses. Similarly, different functions appear to be associated with the different Drosophila GTPases. Rho as well as Cdc42 function is needed for cell shape changes during gastrulation, dorsal closure, bristle, and hair formation. Bristle and hair formation are similarly affected by kette (Hummel, 2000).

These data suggest that Kette provides a novel mechanism linking extracellular signals to the neuronal cytoskeleton. Central relay proteins are SH2-SH3 adapter proteins that control the organization of the actin cytoskeleton via a number of downstream proteins. Biochemical data suggest that additional proteins (p140 kD) may bypass the function of SH2-SH3 adapter proteins, but a detailed analysis awaits its isolation. The Kette protein might interact with extracellular signals, which, in the CNS, might possibly be presented by glial cells. To gain further insight in the neuron-glial interaction at the midline, future work will be directed toward the identification of these components (Hummel, 2000).

The role of Rac in dorsal closure

Rac1 Effects of mutation part 2/2

Home page: The Interactive Fly © 1997 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.