BIOLOGICAL OVERVIEW

Mechanisms that regulate axon branch stability are largely unknown. Genome-wide analyses of Rho GTPase activating protein (RhoGAP) function in Drosophila using RNA interference has identified p190 RhoGAP as essential for axon stability in mushroom body neurons, the olfactory learning and memory center. RhoGAP inactivation leads to axon branch retraction, a phenotype mimicked by activation of GTPase RhoA and its effector kinase Drok and modulated by the level and phosphorylation of myosin regulatory light chain (see Zipper). Thus, there exists a retraction pathway from RhoA to myosin in maturing neurons, which is normally repressed by RhoGAP. Local regulation of RhoGAP could control the structural plasticity of neurons. Indeed, genetic evidence supports negative regulation of RhoGAP by integrin (see Myospheroid) and Src, both implicated in neural plasticity (Billuart, 2001).

Recent studies suggest that Rho, a GTPase that regulates cell shape and motility through modulation of the actin cytoskeleton, regulates the stability of dendritic branches and spines in relatively mature neurons (see for example Nakayama, 2000). The activities of Rho are regulated positively by guanine nucleotide exchange factors (GEFs) and negatively by GTPase activating proteins (GAPs). In turn, RhoGEFs and RhoGAPs can be regulated by upstream cell surface receptors for guidance cues or adhesion proteins. RhoGEFs and RhoGAPs far outnumber Rho GTPases. The Drosophila genome contains six Rho GTPases, but at least 20 predicted RhoGEFs and as many RhoGAPs. The human genome is predicted to contain 59 to 77 RhoGAPs. While it is interesting to speculate on why so many Rho regulators are in the genome, their importance in the function of the human nervous system is highlighted by recent findings that mutations in a RhoGAP and a RhoGEF cause X-linked nonsyndromic mental retardation (Billuart, 2001).

Several RhoGEFs have been shown to be important for axon guidance. Less is known about the cellular function of RhoGAPs in the development and function of the nervous system. One RhoGAP whose function has been investigated is mammalian p190 RhoGAP (p190). Originally identified as a binding partner for p120 RasGAP in Src-transformed cells (Ellis, 1990; Settleman, 1992), p190 preferentially regulates the GTPase RhoA (Ridley, 1993) and is a substrate for Src tyrosine kinase both in vitro (Ellis, 1990) and in vivo (Brouns, 2001). Upon growth factor stimulation or integrin activation, p190 is recruited to the actin cytoskeleton (Burbelo, 1995; Chang, 1995; Sharma, 1998). p190A is highly expressed in the developing and adult mammalian CNS, and knockout mice exhibit defects in a number of developmental processes, including neural tube closure, axon outgrowth, guidance, and fasciculation (Billuart, 2001).

To understand the functions of RhoGAPs in the nervous system, a comprehensive loss-of-function analysis of this class of proteins was undertaken using transgenic double-stranded RNA interference (RNAi), focusing on the Drosophila mushroom body (MB) neurons. The MBs are the insect center for olfactory learning and memory. MB neurons have complex axonal and dendritic developmental programs, allowing various aspects of neuronal development to be studied. Of the 18 RhoGAPs examined, three show distinct loss-of-function phenotypes in MB neurons. In particular, inactivation of the Drosophila homolog of mammalian p190, the subject of this report, results in retraction of axonal branches. This phenotype is mimicked by activation of RhoA or of its effector kinase Drok, and is modified by the level and phosphorylation state of myosin regulatory light chain. These experiments indicate that a previously established pathway from RhoA to Drok to the regulation of myosin (Winter, 2001) also functions in maturing neurons to cause axon branch retraction. Under physiological conditions, this pathway is repressed by p190 to maintain axon branch stability (Billuart, 2001).

Twenty RhoGAPs were defined in the Drosophila genome by the presence of a putative catalytic GAP domain with three characteristic motifs. Sequence analysis of the 20 RhoGAP domains showed a low degree of similarity amongst pairs, except for RhoGAP-84C8 and RhoGAP-50C14, which share 48% amino acid identity. To systematically study the function of all RhoGAPs in neuronal morphogenesis, the transgenic RNAi approach was used, making use of the GAL4-UAS binary expression system. To test if MB neurons are sensitive to transgenic RNAi perturbation, UAS-RhoA inverted repeat transgenic flies were created capable of expressing double-stranded RNA (hereafter refer to as UAS-dsRNA. Then GAL4-OK107, which is highly expressed in all MB neurons and their neuroblasts, was introduced into these flies. Previous work indicated that RhoA is required for MB neuroblast proliferation and cytokinesis. Loss-of-function RhoA clones in the MB result in reduction in neuronal number produced by mutant neuroblasts and the presence of large multinucleated cells. This phenotype is mimicked by RhoA dsRNA expression. The cytokinesis defect is evident from the double labeling of spectrin (marking the cell cortex) and mCD8-GFP (marking the internal and cytoplasmic membrane). Phenotypes induced by RNAi are insertion-dependent. Lines with strong dsRNA expression consistently cause severe loss-of-function phenotypes. This strategy was therefore applied to inactivate the RhoGAP genes (Billuart, 2001).

More than 3 independent insertions of UAS-RhoGAP dsRNA were examined for 17 of the 20 predicted Drosophila RhoGAPs. The following RhoGAPs were excluded from this analysis: (1) a RhoGAP from the rotund (rn) region (RhoGAP-84C8) not expressed in the nervous system; (2) the fly homolog of vertebrate RLIP (RhoGAP-93B3), a putative effector of Ral GTPase involved in receptor-mediated endocytosis; and (3) RhoGAP-71E1, which has P element insertions in its 5' UTR that allowed an analysis of the loss-of-function phenotypes using the MARCM system. Positively labeled MB neuroblast clones for RhoGAP-71E1 exhibited reduction of cell number and misguidance of MB axons. Of the 17 different RhoGAPs subjected to RNAi analysis, six resulted in lethality when expressed ubiquitously using tubulin-GAL4, suggesting that they correspond to genes essential for viability. Three exhibited visible phenotypes or lethality when dsRNA was expressed in the imaginal discs using GAL4-T80. Only two genes gave significant and distinct defects in adult MBs as a result of strong dsRNA expression in MB neurons with GAL4-OK107. DRacGAP (RhoGAP-50C14) (Sotillos, 2000) dsRNA expression resulted in reduction of neuronal number, abnormally large cells, and overextension of the axons. The second gene, RhoGAP-16B12, is the focus of the remainder of this study (Billuart, 2001).

The adult mushroom body is composed of gamma, alpha'/ß', and alpha/ß neurons, totaling about 2500 per brain hemisphere. The axons of later-born alpha'/ß' and alpha/ß neurons are bifurcated, each projecting one branch dorsally and the other medially. The axons of the early-born gamma neurons project only medially in adult. MB neurons expressing p190 dsRNA exhibit truncation or loss of dorsal branches. No obvious defect was observed in cell number or morphology of the calyx where the dendrites are confined. The phenotype is dependent on the insertion site of the transgenes. Of the fifteen p190 dsRNA lines analyzed, three showed very weak or no phenotypes. Seven (e.g., line 14.1) showed intermediate phenotypes, including a mixture of normal, pointed dorsal lobes ('weak'), shortened dorsal lobes ('medium'), or a complete lack of dorsal lobes ('strong)'. Five lines (e.g., line 5.2) gave reproducibly strong phenotypes (~70% belonging to the 'medium' or 'strong' categories). Use was made of the dorsal axon branch phenotype to study the genetic interaction of p190 with other genes. The presence of two transgene copies of an intermediate line markedly enhanced the RNAi-induced phenotype, confirming the interpretation of phenotypic strengths among different categories and supporting that the RNAi-induced phenotype can serve as a sensitive assay for genetic interactions (Billuart, 2001).

Although inhibition of p190 in MB neurons preferentially affects the dorsal branch, the specificity is not absolute. As the truncation of the dorsal lobe becomes more severe, defects are observed in medial lobe fasciculation. Anti-Fasciclin II (Fas II) staining -- which strongly labels alpha/ß neurons, weakly labels gamma neurons, and does not label alpha'/ß' neurons in wild-type organisms -- reveals that alpha/ß neurons contribute to most fasciculation defects in the medial bundles. In extreme cases, truncated ß lobes were observed (Billuart, 2001).

To test if mammalian p190 could complement the RNAi-induced phenotype of Drosophila p190, transgenic flies were generated expressing rat p190A under the control of the UAS promoter. Expression of rat p190A in MB neurons almost completely rescues the p190 dorsal lobe truncation phenotype, demonstrating the functional conservation between fly and mammalian p190 (Billuart, 2001).

The spectrum of the p190 RNAi-induced phenotypes suggests that they are caused by MB axon retraction. To confirm this interpretation, the p190 phenotypes were analyzed at different stages of development using a strong RNAi line. At the end of the third instar larval stage, MBs are composed of gamma and alpha'/ß' neurons; each forms a dorsal and a medial branch. No RNAi-induced phenotypes were found at this stage. At 18 hr after puparium formation (APF), wild-type MB gamma neurons have pruned their axons in the dorsal and medial lobes, but alpha'/ß' neurons retain their axon branches in the lobes. A large proportion of MB neurons expressing p190 dsRNA also possessed the dorsal alpha' lobes, indicating that most, if not all, dorsal branches of alpha'/ß' neurons were originally present. At this stage pointed dorsal lobes and very thin processes were observed punctuated by dots at the tip of these lobes, resembling aspects of axon retraction. By 36 hr APF, most p190 dsRNA-expressing MB neurons exhibit medium or strong phenotypes. Quantification indicated a time-dependent reduction of dorsal branches in p190 dsRNA-expressing MB neurons. These experiments indicate that p190 is required for the stability of the dorsal axon branches of at least alpha'/ß', and possibly alpha/ß, neurons during development (Billuart, 2001).

To test if axon retraction persists in adult life, the p190 phenotype of an intermediate line was quantified from 0 to 6 weeks after adult eclosion; reduction in the dorsal lobe is progressive. This result indicates that p190 is also required for axon stability throughout adult life (Billuart, 2001).

Microinjection of the GAP domain of p190 into fibroblasts results in actin stress fiber disassembly, suggesting that it primarily acts on RhoA (Ridley, 1993). Consistent with this observation, overexpression of RhoA in MB neurons significantly enhances the p190 phenotype. If p190 acts on RhoA, one would further predict that activation of the RhoA pathway would mimic p190 loss-of-function phenotypes. To explore this possibility, active RhoA (RhoA V14) was expressed constitutively in MB neurons. OK107-driven RhoA V14 expression results in adult lethality. When reared at 18°C, a few escapers were recovered that exhibited complex MB axon defects. It was difficult to determine if the escapers shared qualitatively similar phenotypes to those caused by p190 RNAi. However, in pupae, a selective dorsal lobe reduction similar to the p190 RNAi phenotype was often observed (Billuart, 2001).

RhoA transduces signals to both the nucleus and the cytoskeleton. To address which downstream signaling pathway mediates axon retraction, activated RhoA mutants (RhoA V14) were used with additional 'effector domain' mutations. The F39V mutation blocks RhoA's function in inducing stress fiber formation without affecting nuclear signaling, whereas the E40L mutation allows both nuclear and cytoskeletal pathways weakly. When RhoA V14(E40L) was expressed in MB neurons, a dorsal lobe phenotype similar to the p190 phenotype was found. However, RhoA V14(F39V) expression had almost no phenotype, suggesting that a cytoskeletal pathway is responsible for RhoA's effect on axon branch retraction (Billuart, 2001).

The results implicate an axon branch retraction pathway in MB neurons involving RhoA, Drok, and MRLC; this pathway is repressed by p190 to maintain axon branch stability. What are the consequences of preventing its activation? Insight first came from analyzing MB neurons overexpressing rat p190A, which almost completely rescues the Drosophila p190 RNAi phenotype. Moreover, MB neurons overexpressing rat p190A alone or in the p190 RNAi background exhibit other phenotypes including overextension of dorsal axon branches, the opposite of the p190 RNAi phenotype. Double labeling of MB axons and rat p190A reveals that rat p190A is preferentially located at the tip of axon terminals and is concentrated in overextended axons (Billuart, 2001).

Overexpression of RhoGap p190 in MB neurons also causes axon overextension. A single point mutation (R1389L) in the GAP domain of D-p190, that was predicted to interfere with the GAP activity, largely abolished this phenotype, demonstrating that axon overextension is dependent on the GAP activity of p190 (Billuart, 2001).

The simplest interpretation for the overextension phenotype is that overexpression of p190 inhibits normal RhoA signaling, suggesting that RhoA signaling is required for preventing axon overextension. Loss-of-function mutants in the RhoA pathway would be predicted to lead to axon overextension. Using the MARCM system, MB neuroblast clones homozygous for RhoA, Drok, and sqh were generated to critically test this hypothesis. Loss of RhoA activity leads to severe neuroblast proliferation defects such that neurons contributing to the adult dorsal lobes are never born. Similar proliferation defects are found in sqh^AX3 neuroblast clones. Thus, the effects of RhoA and MRLC could not be evaluated directly using loss-of-function mutants (Billuart, 2001).

By inhibiting the activity of RhoGAP p190, this study has uncovered a 'retraction signaling pathway' from RhoA and Drok to the regulation of myosin II activity via modulation of myosin regulatory light chain (MRLC) phosphorylation. This pathway is largely dormant, such that loss-of-function RhoA does not cause detectable phenotypes (Lee, 2000) and loss-of-function Drok gives rise to a subtle axon overextension phenotype. Yet, this pathway is extant, since activation of RhoA or Drok or inactivation of p190 by RNAi results in robust axon branch retraction phenotypes. Further, the p190 phenotype is modulated by changing the endogenous level or the phosphorylation state of MRLC, a critical output of Drok signaling in vivo (Winter, 2001). Thus, it appears that all components in the pathway are present in MB neurons and participate in mediating the axon retraction when p190 function is inhibited (Billuart, 2001).

Studies of mammalian hippocampal pyramidal neurons have provided an analogous example of an extant signaling pathway maintained in a dormant state. While activation of RhoA or ROCK (mammalian homolog of Drok) in maturing pyramidal neurons results in dendritic branch retraction and elimination, inhibition of RhoA or ROCK activity does not lead to detectable phenotypes. However, the effect of RhoA activation can be completely suppressed by ROCK inhibition, indicating that endogenous ROCK is used to mediate the action of the activated RhoA (Nakayama, 2000). It is possible that under physiological conditions, the RhoA pathway in pyramidal neurons is actively repressed by mammalian p190 RhoGAP to stabilize dendritic branches. Regulation of a repressor of RhoA signaling is likely to be a widely used mechanism to regulate stability of neuronal processes (Billuart, 2001).

Why are the dorsal and medial axon branches of the same neurons differentially sensitive to the retraction pathway? It is possible that some signaling components are differentially distributed between these two branches. Since p190 inhibition and activation of RhoA or Drok all selectively affect dorsal branches, differential distribution of signaling components downstream of Drok, such as MRLC, MyoII, or MRLC phosphatase could account for this difference (Billuart, 2001).

An alternative explanation is that the relative levels of RhoA and Rac activity may differ in dorsal and medial branches. This hypothesis is based on the general scheme in which Rac plays a positive role in process outgrowth, whereas RhoA acts negatively. If Rac activity were higher in the medial branches, then medial branches would be expected to be more resistant to retraction upon activation of RhoA. Two lines of evidence support this Rac/RhoA antagonism hypothesis. (1) The MB gamma neurons, which only send medial branches after metamorphosis, appear to resist the retraction under all experimental conditions tested, even though GAL4-OK107-driven dsRNAs is highly expressed in gamma neurons. Of the three types of MB neurons, gamma neurons are the only ones that express Trio, a RacGEF, during the critical period for axon retraction of MB development. (2) RNAi-induced DRacGAP loss-of-function results in overextension of the axons at the tip of the dorsal lobe, a phenotype similar to overexpression of p190 but opposite that of the axon branch retraction resulting from p190 RNAi. These opposite loss-of-function phenotypes of p190 and DRacGAP suggest that the relative balance of RhoA and Rac activation regulates axon branch stability (Billuart, 2001).

Identifying upstream signals that regulate p190 activity may provide further insight into the physiological signals that derepress the RhoA-mediated retraction pathway. Two upstream regulators of mammalian p190 are Src family protein tyrosine kinases and integrin family extracellular matrix adhesion molecules. Whether p190 is positively or negatively regulated by integrin and Src remains controversial; indeed, a biphasic regulation of RhoA by integrin through Src and p190 has been proposed. Given the functional similarities of Drosophila and mammalian p190 by both genetic complementation and overexpression assays, it seems likely that Drosophila p190 and Src or integrin work in the same pathway, as mammalian biochemical studies have suggested. If that is the case, then the genetic results of this study support the notion that p190 is negatively regulated by both Src and integrin. However, genetic interaction studies alone cannot rule out the possibility that integrin and Src positively regulate RhoA through an as-of-yet unidentified mechanism independent of p190, or that signals are integrated downstream of RhoA (Billuart, 2001).

Negative regulation of p190 by Src could result from the fact that there is partial conservation of Src phosphorylation sites between mammalian and Drosophila p190. Src phosphorylation of mammalian p190 at Y1105 appears to be responsible for p190 activation (Haskell, 2001). Src phosphorylation of tyrosines in the GTP binding domain of mammalian p190 results in the loss of its GTP binding activity (Roof, 2000), and hence a disruption of the RhoGAP activity in vivo (Tatsis, 1998; Brouns, 2000). Thus, phosphorylation of the GTP binding domain is predicted to negatively regulate p190. Since only Src-phosphorylation sites in the GTP binding domain of mammalian p190 are conserved in Drosophila, Src phosphorylation of D-p190 likely negatively regulates its activity (Billuart, 2001).

The negative regulation of p190, and hence positive regulation of RhoA, by integrins is further supported by mutant phenotypes in Drosophila neurons. Previously studies have shown that loss of Drosophila ßPS integrin (myospheroid) or alphaPS3 (volado) causes excessive synaptic sprouting and morphological growth in the neuromuscular junction. Homozygous loss of ßPS integrin in MB neurons results in axon overextension phenotypes strikingly similar to those caused by p190 overexpression and Drok inactivation, but are the opposite that of phenotypes caused by loss-of-function p190 or activation of RhoA and Drok. Taken together with the genetic interaction data, these experiments support a model that integrin derepresses the retraction pathway by negatively regulating p190 RhoGAP. Future biochemical studies are required to elucidate the mechanisms by which integrin and Src regulate p190 in Drosophila (Billuart, 2001).

These findings suggest a link between a molecular pathway that regulates axon branch stability to Src and integrin, both implicated in neural plasticity and memory formation. Mice lacking the Src family kinase Fyn have impaired long-term potentiation (LTP) and spatial learning, phenotypes that are separable from developmental defects. Src has also been implicated in regulating LTP induction. Likewise, inhibiting integrin function results in LTP defects in rats, synaptic plasticity in the Drosophila neuromuscular junction, and MB-mediated short-term memory in Drosophila. The molecular mechanisms by which Src and integrin regulate plasticity are largely unknown. Mammalian p190 is highly expressed in the adult brain (Brouns, 2001). In p190 dsRNA-expressing MB neurons, axon branch retraction continues over the course of adult life, suggesting that p190 RhoGAP continues to function in regulating axon stability in this olfactory learning and memory center in adult brain. Taken together, this study raises the intriguing possibility that regulation of p190 RhoGAP activity, and hence the structural changes of subcellular compartments of neurons, may contribute to the morphological plasticity essential for learning and memory (Billuart, 2001).

PROTEIN STRUCTURE

Amino Acids - 1564

Structural Domains

Sequence analysis has revealed that RhoGAP-16B12 is the Drosophila homolog of mammalian p190 RhoGAP. In addition to C-terminal GAP domains, Drosophila and rat p190A contain near their N termini a predicted GTP binding domain most similar to the Rab subfamily of small GTPases. Specific tyrosine residues of R-p190A are phosphorylated by the nonreceptor tyrosine kinase Src, including two residues (Y1087 and Y1105) involved in the binding of p120 RasGAP. Those sites are not conserved in Drosophila. However, two tyrosine residues in the GTP binding domain of rat p190A are conserved in the Drososphila protein. Phosphorylation of these residues in rat p190A by c-Src abolishes its capability to bind GTP (Roof, 2000), which in turn is necessary (Tatsis, 1998) for the GAP activity (Billuart, 2001).

date revised: 2 January 2023

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

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