Antagonistic Feedback Loops Involving Rau and Sprouty in the Drosophila Eye Control Neuronal and Glial Differentiation

See allHide authors and affiliations

Science Signaling  05 Nov 2013:
Vol. 6, Issue 300, pp. ra96
DOI: 10.1126/scisignal.2004651

This article has a correction. Please see:


During development, differentiation is often initiated by the activation of different receptor tyrosine kinases (RTKs), which results in the tightly regulated activation of cytoplasmic signaling cascades. In the differentiation of neurons and glia in the developing Drosophila eye, we found that the proper intensity of RTK signaling downstream of fibroblast growth factor receptor (FGFR) or epidermal growth factor receptor required two mutually antagonistic feedback loops. We identified a positive feedback loop mediated by the Ras association (RA) domain–containing protein Rau that sustained Ras activity and counteracted the negative feedback loop mediated by Sprouty. Rau has two RA domains that together showed a binding preference for GTP (guanosine 5′-triphosphate)–loaded (active) Ras. Rau homodimerized and was found in large–molecular weight complexes. Deletion of rau in flies decreased the differentiation of retinal wrapping glia and induced a rough eye phenotype, similar to that seen in alterations of Ras signaling. Further, the expression of sprouty was repressed and that of rau was increased by the COUP transcription factor Seven-up in the presence of weak, but not constitutive, activation of FGFR. Together, our findings reveal another regulatory mechanism that controls the intensity of RTK signaling in the developing neural network in the Drosophila eye.


The formation of the nervous system depends on a well-orchestrated assembly of many different neuronal and glial cell types. Cell fate determination often relies on the activity of complex signaling pathways such as those mediated by receptor tyrosine kinases (RTKs) (1). Generally, the binding of highly specific and selective ligands results in the phosphorylation of one or several tyrosine residues on the cytoplasmic domain of an RTK, a signal that often determines cell fate (2, 3). Thus, the duration and strength of this signal need to be carefully controlled during development. Prominent examples are signaling cascades mediated by the epidermal growth factor receptor (EGFR) that specify cell fate during vulva induction in Caenorhabditis elegans or during photoreceptor cell fate induction in Drosophila (47).

The compound eye of Drosophila comprises 750 single-eye units called ommatidia. Each ommatidium houses eight photoreceptor neurons (R1 to R8). Eye development is initiated during the third instar larval stage in the eye-antennal imaginal disc, where neurogenesis is triggered posteriorly to the morphogenetic furrow by the activation of the proneural gene atonal, which specifies the central photoreceptor cell R8. Starting from this founding photoreceptor cell, sequential rounds of EGFR activation induce the formation of the so-called outer photoreceptor cells R1 to R6. The R7 cell is then directed toward its fate by additional signaling mediated by the Sevenless RTK. Without either Sevenless or EGFR signaling, the cell acquires a nonneuronal fate. Feeding into the same signaling cascades, EGFR can compensate for loss of Sevenless, and thus, photoreceptor cell fate appears to depend on the intensity of RTK signaling transduction (4, 8, 9).

In addition to neurons, glial cells are important for retinal function. The specification and differentiation of the retinal glia provides another case where differential RTK activity is decisive (10). The eye imaginal disc is linked to the brain lobes through the optic stalk and harbors several glial progenitor cells (1114). From the optic stalk, retinal glial cells migrate onto the eye disc epithelium where, upon contact with photoreceptor axons, they eventually differentiate into wrapping glial cells. The number and the migratory potential of these progenitors, the perineurial glial cells, are adjusted to the growing eye imaginal disc by juxtacrine activation of the fibroblast growth factor receptor (FGFR) through the glial FGF8 (fibroblast growth factor 8)–like ligand Pyramus (10). In the anterior portion of the eye disc, the perineurial glia comes into contact with nascent photoreceptor axons. These axons secrete an additional FGF8-like molecule (Thisbe) and thereby induce the transformation of the perineurial glia toward a wrapping glial fate (10). Subsequently, axonally provided Thisbe controls the extent of glial wrapping around photoreceptor axons (10). Thus, the level of FGF-induced signaling must be carefully controlled in the developmental stage.

One mechanism that cells use to fine-tune RTK signaling is conveyed through the negative regulator Sprouty (2, 1520). Initially found to act downstream of FGFR, it is now clear that Sprouty is a more general inhibitor of RTK signaling. Sprouty is induced by strong RTK activation and subsequently silences RTK activity. However, it is unclear how RTK activity can be sustained in the presence of Sprouty. We identified the Drosophila gene rau, which encodes the Ras association (RA) domain–containing protein Rau, and found that Rau acts in a positive feedback loop downstream of FGFR- and EGFR-mediated signaling by preferentially binding GTP (guanosine 5′-triphosphate)–loaded Ras. Thus, Rau counteracts Sprouty to enable differential responses to RTK activation in the developing Drosophila eye.


rau is expressed by retinal glial cells

We had previously noted the induction of sprouty expression in response to FGFR activation at the onset of wrapping glial cell differentiation (10). To further understand the regulation of FGFR-mediated signaling during glial cell development, we reasoned that relevant genes might be similarly expressed with sprouty, and we therefore screened a large collection of transgenic enhancer trap lines for expression patterns that were identical to the one shown by sprouty (21). We identified the insertion rR125, which directs β-galactosidase expression in the sprouty-expressing wrapping glial cells. These cells also express the wrapping glial marker Mz97Gal4 (Fig. 1, A and B). As observed for sprouty, expression of the enhancer trap insertion rR125 was under the control of FGFR. Silencing of FGFR-mediated signaling after expression of a dominant-negative FGFR (Drosophila Heartless) construct in the wrapping glia using repoGal4 resulted in the loss of rR125 expression, whereas expression of a constitutively active FGFR in all glial cells using repoGal4 increased rR125 expression (Fig. 1, C to H).

Fig. 1 Expression of sprouty and rau in third instar eye imaginal discs.

(A and B) Staining for the wrapping glia marker (Mz97Gal4 UASLamGFP, green) and β-galactosidase (red), which was directed by the (A) sprouty (stylacZ) or (B) raulacZ enhancer trap insertion. Glial nuclei are stained for Repo expression (blue). (C to H) False color analysis of the intensity of sprouty (C to E) or rau (F to H) reporter abundance observed in flies upon pan-glial expression of a dominant-negative FGFR (htlDN) or a constitutively active FGFR (htlact). Blue, none; green, low; yellow, moderate; red, high. All images are representative of more than 20 eye imaginal discs stained in at least three independent experiments.

To test the breadth of RTK pathway activation, we expressed a Drosophila hemagglutinin (HA)–tagged Rolled protein [the Drosphila mitogen-activated protein kinase (MAPK)] (22) in all glial cells using repoGal4 (fig. S1). HA-tagged Rolled was most prominently detected in the nuclei of the wrapping and the subperineurial glia, whereas it was more evenly distributed between the cytoplasm and nucleus in the perineurial glial cells (fig. S1, A and B). Thus, activation of FGFR-mediated signaling, which occurs when wrapping glia differentiate, correlates with the activation of sprouty and rR125 expression.

rau promotes EGFR signaling

The rR125 enhancer trap carries a transposon insertion 5′ of the gene CG8965 (fig. S1C). In situ hybridization experiments and immunostaining for Rau in embryos and larval brains showed that the expression pattern detected by the rR125 enhancer (fig. S1, D to H) reflects CG8965 expression (23). This correlated with the expression of the FGFR Heartless and the activity of the sprouty enhancer in longitudinal glial cells (fig. S1, I and J). A transposon insertion, CG8965f04153 (available from the Exelixis stock collection), shows a rough eye phenotype when either homozygous or in trans to deficiencies uncovering this gene (Fig. 2). This phenotype led us to name the gene rau, because “rau” means rough in German.

Fig. 2 Analysis of the rau eye phenotype.

(A to F) Scanning electron microscopy (SEM) analyses of external eye phenotypes for the respective mutants. Below each eye, higher magnifications are shown to analyze the organization of ommatidia. (G) Quantification of misshaped ommatidia. ***P < 0.001; n = 5 eyes with the central 600 ommatidia each. (H and I) Rough eye phenotype in (H) wild-type flies or (I) flies with GMRGal4-mediated knockdown of rau. (J and K) Rough eye phenotypes in flies with (J) GMRGal4-mediated overexpression of rau or (K) GMRGal4-mediated knockdown of sprouty. All images are representative of eyes from five flies.

When we placed the rauf04153 mutation in trans to a 140-kb deficiency covering the entire region [Df(2L)Exel6014], we noted a rough eye phenotype. To further analyze the role of rau, we generated a smaller 60-kb deficiency [Df(2L)rauΔ] that affected only four other protein-encoding genes: Trissin receptor, which encodes a putative G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptor; CG31644, which encodes a cytochrome c oxidase subunit; CG43307, which encodes an 83–amino acid short protein that is present only in males; and CG11034, which encodes a peptidase. No mutant information is currently available for any of these genes. Homozygous rauΔ flies were viable (although males were sterile) and showed a prominent rough eye phenotype (Fig. 2, A and B). Similarly, trans-heterozygous rauf04153/rauΔ flies caused sterility in males and a weaker rough eye phenotype compared to homozygous rauΔ flies (Fig. 2, B and C). This may indicate the hypomorphic nature of rauf04153.

To test for the genetic relationship between rau and sprouty, we removed one copy of the RTK signaling antagonist sprouty in a rau mutant background (rauΔ/rauΔ;styΔ5/+ or rauf04153/rauΔ;styΔ5/+). In both double-mutant flies, the rough eye phenotype was alleviated (Fig. 2, D and E). In contrast, when we removed one copy of pointed, which encodes a transcription factor that is downstream of EGFR and FGFR signaling, the rough eye phenotype was worse (Fig. 2F). The rough eye phenotype is characterized by fused ommatidia and can be easily quantified. Quantitative analysis of the ommatidia in the rau mutant flies indicates that downstream RTK signaling effectors, Pointed and Rau, antagonize sprouty in the developing eye (Fig. 2G).

To further address whether Rau is critical to the developing eye, we performed loss- and gain-of-function experiments. Compared with wild-type flies (Fig. 2H), silencing rau expression with RNA interference using the GMRGal4 driver induced a rough eye phenotype (Fig. 2I). As well, increasing rau expression using GMRGal4 caused a rough eye phenotype (Fig. 2J) that resembled the rough eye phenotype of sprouty knockdown flies (Fig. 2K). In summary, these data indicate that the proper expression of rau is important for development of the eye.

rau is important for R7 photoreceptor development

During photoreceptor cell differentiation, each ommatidial eye unit houses eight photoreceptor neurons, which develop in a sequential manner depending in part on EGFR activation (9). The photoreceptor cell R7 forms last and requires twofold activation of Ras signaling by the EGFR and the Sevenless RTK. The rau enhancer trap insertion rR125 showed the most prominent expression in the R7 cell (Fig. 3A). This finding correlates with the reported strong RTK activation in this cell (4). Compared with eyes from wild-type flies (Fig. 3B), the R7 cell marker Prospero was frequently lost in the eye imaginal disc from rauf04153/rauΔ flies (Fig. 3C) and from homozygous rauΔ mutant flies (Fig. 3D). However, the “R7 cell loss” phenotype in rau mutants was suppressed by the removal of one copy of sprouty (Fig. 3E), highlighting the important role of RTK signaling maintenance in R7 cell formation. Similar to the phenotype seen in rauf04153/rauΔ flies (Fig. 3C), the expression of raudsRNA also induced loss of the R7 cell (Fig. 3, F and G). The greatest loss was observed in homozygous mutants, in which the R7 cell was missing in 16.4% of 3112 rau mutant ommatidia from 27 eye discs (Fig. 3H), and a corresponding proportion of ommatidial axon bundles in the optic stalk comprised only seven instead of the normal eight axons (fig. S2A). Correspondingly, we saw a similar phenotype in adult rau mutant ommatidia compared to wild type (fig. S2, B and C). This loss of R7 cells resembles the eye phenotype observed after impaired RTK signaling (4). Moreover, when we overexpressed rau in a wild-type background using GMRGal4, additional R7 cells were formed (fig. S2D). In conclusion, these data indicate that rau activates RTK signaling mediated by EGFR and Sevenless receptor during photoreceptor development.

Fig. 3 Rau is involved in R7 specification.

Confocal sections of third instar raurR125 eye discs stained for Prospero (red), Elav (blue), and β-galactosidase (green). (A) raurR125 expression in R7 cells (arrows). False color analysis of rau expression. (B to G) Loss of R7 in rau mutant third instar eye discs. Elav (red) and Prospero (gray) costaining identifies R7 cells [some are indicated by arrows (B)]; horseradish peroxidase (HRP) staining (blue) marks neuronal membranes. The asterisks indicate ommatidia with missing R7 cells. All images are representative of more than five imaginal discs from three different stainings. (H) Quantification of R7 loss in genotypes as indicated. n, number of eye discs counted; **P < 0.01.

rau promotes FGFR signaling in glial cells

To test whether Rau also functions in the eye glia, we induced loss and gain of function of Rau in wrapping glial cells. Trans-heterozygous rau mutants (rauf04153/rauΔ) showed impaired glial wrapping in the optic stalk (Fig. 4, A and B), indicating that Rau is important for wrapping glia development. Pan-glial overexpression of sprouty also induced glial wrapping defects (Fig. 4C), and concomitant knockdown of rau and overexpression of sprouty in glia further impaired the phenotype (Fig. 4D). When we silenced seven-up specifically in glial cells, we noted broad areas in the optic stalk devoid of glial processes (Fig. 4E), indicating glial wrapping defects as we had seen after pan-glial overexpression of sprouty (Fig. 4C). Transmission electron microscopy (TEM) analysis revealed that many of the ommatidial axon fascicles in rau mutants appeared fused compared with those in wild-type flies (Fig. 4, F and G). A similar phenotype is observed after overexpression of sprouty specifically in the wrapping glia using Mz97Gal4 (10). In contrast, a phenotype of hyper-wrapped ommatidial axon bundles was observed after specific overexpression of rau in only the wrapping glia using the Mz97Gal4 driver line (Fig. 4H), similar to sprouty loss-of-function mutants (Fig. 4I). The number of hyper-wrapped ommatidial axons was slightly increased by the overexpression of rau (Fig. 4J). In contrast, rau mutants showed an increased percentage of axons showing no wrapping (Fig. 4J), a phenotype that has been previously linked to reduced FGFR activity (10). The data therefore suggest that Rau and Sprouty can both exert antagonizing functions in the retinal glial cells and that Rau promotes signaling that mediates glial wrapping.

Fig. 4 Rau is required for wrapping glial development.

(A to E) Projections of confocal images of third instar eye discs; glial membranes in green, and glial nuclei in red. The bottom images show the optic stalk of the respective confocal image. The arrows indicate defective glial wrapping. (F to I) Electron micrographs of cross sections of third instar optic stalks. Axon clusters are marked in dark blue. Light blue indicates axon clusters with seven, instead of eight, axons. Asterisk notes nonwrapped axon clusters in rau mutants (G); arrows mark a hyperwrapped axon cluster upon overexpression of rau in wrapping glia (H) or pan-glial knockdown of sprouty (I). (J) Quantification of hypo- and hyperwrapped axon clusters. ***P < 0.001; n = 3 optic stalks for each genotype. (K to N) Confocal analyses of third instar eye imaginal discs stained for Seven-up (green), HRP (blue), and β-galactosidase expression (red) directed by (K and L) sproutylacZ or (M and N) raulacZ. (K and M) Basal projection showing glial cells; dotted lines indicate the position of the orthogonal section shown in (L) or (N), which show Seven-up and Sprouty (L) or Rau (N) abundance in the eye disc glia. Arrows mark weak Seven-up abundance in the basal perineurial glia; asterisks indicate apical perineurial glial cells. (O to R) Analysis of expression of sprouty (O and P) and rau (Q and R) after pan-glial seven-up knockdown (svp kd). Dicer2 expression was used to enhance the efficiency of RNA interference. (S) Quantification of sprouty-positive glial cells per eye disc. **P < 0.01; n = 3 eye discs with 18 to 20 photoreceptor rows. All images are representative of eye imaginal discs from more than 10 flies.

Seven-up regulates rau and sprouty expression

Given the importance of the two antagonistic signaling components, Sprouty and Rau, we wondered how the corresponding genes are activated during wrapping glial cell fate allocation in the eye disc. It has been described that sprouty expression in photoreceptor cells is suppressed by the orphan nuclear hormone receptor Seven-up (24, 25). A similar relation was observed in glial cells, in which Seven-up was found in glial cells that did not show abundance of the wrapping glial markers Sprouty (Fig. 4, K and L) or Rau (Fig. 4, M and N). The greatest Seven-up abundance was found in the apical-most perineurial glial cells that lack contact to descending photoreceptor axons and will not differentiate into wrapping glia (Fig. 4L, asterisks). Upon pan-glial overexpression of the activated FGFR Heartless, Seven-up abundance was markedly suppressed in all glia except the subperineurial glia, which are recognized by the size of their nuclei (fig. S3, A and B). In contrast, overexpression of FGFR increased Sprouty abundance in all glia except the subperineurial glia (Fig. 1 and fig. S3, A and B), suggesting that RTK activity differentially regulates seven-up and sprouty expression.

In agreement with the observation that silencing seven-up induced glial wrapping defects (Fig. 4E), the number of cells showing higher Sprouty abundance increased after silencing seven-up (Fig. 4, O and P). In turn, Sprouty abundance was reduced by overexpression of seven-up in single-cell clones (fig. S3C). Furthermore, larval lethality caused by pan-glial knockdown of seven-up was prevented by concomitant knockdown of sprouty. These findings demonstrate that Seven-up normally represses sprouty expression and indicates that the glial wrapping phenotype is induced by decreased FGFR signaling.

When we silenced seven-up, we saw reduced rau expression (Fig. 4, Q and R), in contrast to sprouty, whose expression increased, evident in the number of Sprouty-positive glial cells (Fig. 4S). The data suggest that the expression of sprouty and rau in the wrapping glia is under the common control of FGFR-mediated signaling, but that the Seven-up transcription factor affects expression of rau and sprouty differently, providing distinct feedback on FGFR activity during differentiation.

Rau binds RasGTP

rau encodes a 430–amino acid protein that has two RA domains (Fig. 5A). Antibodies directed against the Rau protein detected a weakly abundant cytoplasmic protein with a molecular weight of 51 kD (Fig. 5B), which corresponds to the calculated size. The protein was absent in mutant tissue and was easily detected upon ubiquitous Gal4-mediated expression (Fig. 5B). Because the RA domains bind members of the Rho–GTPase (guanosine triphosphatase) family (26), we tested the binding specificities of bacterially produced Rau protein. Pull-down experiments demonstrated a preferential interaction of Rau with GTP-loaded Ras (RasGTP) over Rap1 or Rac (Fig. 5C and fig. S4). To test whether both RA domains can mediate RasGTP binding independently, we generated Rau variants lacking either or both of the RA domains (Fig. 5A). Although both RA domains interacted with Ras, no binding preference for GTP-loaded Ras was seen when either the first or the second RA domain was deleted (Fig. 5C). Thus, cooperative effects appear to mediate the preference for GTP-loaded Ras.

Fig. 5 Rau binds GTP-loaded Ras1.

(A) rau encodes a protein with two RA domains and a C-terminal class II PDZ-binding motif (PDZ BM). Three deletion constructs were generated. The Myc tag was added at the N terminus; the His and HA tags were added at the C terminus. (B) Rau antibodies recognize two ~51-kD proteins in third instar brain and wing disc lysates (arrowheads). The lower band (black arrowhead) is absent in mutant rauΔ brains and wing discs. rau expression is induced by daGal4. (C) GST pull-down of His-tagged Rau protein, wild-type or variants lacking either or both RA domains, incubated with GST-tagged Ras1 coupled to glutathione-Sepharose beads and loaded with GDP (guanosine diphosphate) or GTP. Blots are representative of three independent experiments.

In a recent global protein-protein interaction study, it was reported that Rau can interact with itself (27). We therefore tested whether this involves the RA domains by performing GST (glutathione S-transferase) pull-down experiments. GST-tagged, full-length Rau protein interacted with His-tagged, full-length Rau (Fig. 6A). Reduced interaction was seen with the deletion mutants, in which either the RA1 or the RA2 domain was deleted. Full-length Rau also weakly interacted with the deletion construct lacking both RA domains (Fig. 6A). To further identify the dimerization domain, we generated a Rau variant lacking the N terminus, which was able to dimerize as the full-length construct. Unfortunately, we were not able to generate a C-terminal deletion because the corresponding proteins precipitated. In conclusion, the data indicate that RA domain clustering in the Rau protein favors the binding of RasGTP and, moreover, that Rau can interact with itself, which may further contribute to the binding preference for RasGTP. The ability of Rau to form multimers might account for the Rau-positive, large–molecular weight complexes that we identified in vivo by gel filtration experiments (Fig. 6B).

Fig. 6 Rau forms dimers and is found in large–molecular weight complexes.

(A) GST pull-down of His-tagged Rau protein variants, lacking either one or both RA domains (ΔRA) or the N terminus (ΔNterm). Purified His-tagged Rau variants were incubated with GST-tagged Rau coupled to glutathione-Sepharose beads. (B) Gel filtration was performed from embryo lysates ubiquitously expressing MycRau. Staining for the Myc epitope shows Rau localization in large molecular weight complexes. Blots are representative of two independent experiments.

Both RA domains are required for Rau function

The tandem organization of the RA domains suggests that Rau might cluster GTP-loaded Ras and, thus, promote its activity. To validate this assumption, we tested the functionality of the different constructs in rescue and gain-of-function experiments. For all experiments, we used phiC31-mediated transgene integration, and all transgenes analyzed in this study were inserted into the same landing site at 86Fb, ensuring comparable expression. As a rescue endpoint, we used the R7 cell phenotype, which allows an easier and more precise quantification compared with the external rough eye phenotype. In rau mutants, many ommatidia lacked the central R7 cell (Fig. 3H and fig. S5A). Upon ubiquitous, daGal4-driven, expression of a wild-type Rau protein, about 60% of the mutant ommatidia were rescued (fig. S5, B and C). A weaker rescue of only about 40% of the ommatidia was seen after expression of the HA-tagged Rau protein (fig. S5C). In this protein variant, the HA tag is added at the C terminus, disrupting the PDZ-binding motif, which indicates that this motif may help but is not essential for function. When we expressed HA-tagged Rau protein forms that lacked either the first or the second RA domain, no statistically significant rescue was observed (fig. S5C). Likewise, no rescue occurred when we expressed a Rau protein lacking both RA domains (fig. S5C).

In addition to rescue experiments, we overexpressed the different Rau deletion constructs in a wild-type background. Whereas expression of the untagged (Fig. 2J) as well as the HA-tagged Rau variant caused a clear rough eye phenotype (fig. S6A), no abnormal eye phenotype was seen when we expressed a form of Rau that lacked the first, second, or both RA domains (fig. S6, B to D). In conclusion, although both RA domains were able to interact with Ras in vitro, both domains were needed for in vivo function. Our results support a model where multimerization of RA domains favors the recruitment of GTP-loaded Ras, which subsequently facilitates the activation of the MAPK cascade.


During development, often single bursts of RTK activity suffice to direct important cellular decisions. In other cases, multiple rounds of RTK activation are required to trigger a certain reaction profile, and in yet other cases, such as in the developing eye imaginal disc, a sustained low level of EGFR activity is needed (2, 28, 29). Here, we identified the Drosophila RA domain containing protein Rau, which constitutes the first cell-autonomous positive feedback regulator acting on both EGFR- and FGFR-induced signaling. In the developing fly compound eye, we found that sustained RTK activity is modulated through a positive feedback loop initiated by Rau, which is counterbalanced by the negative regulator Sprouty. The balance of these two regulatory mechanisms ensures the correct activity of EGFR- and FGFR-dependent signaling pathways in the developing eye.

Within the RTK signaling pathway, different positive and negative feedback mechanisms have been identified (2, 28). A prominent negative feedback mechanism is triggered by the secreted protein Argos. Argos expression is induced by RTK activation, and secreted Argos protein can sequester the activating ligand Spitz (3033). In addition, intracellular proteins have been identified to exert a negative feedback function. Sprouty is the most prominent inhibitor of RTK activity and was shown in this study to act downstream of FGFR signaling as well as downstream of EGFR signaling. However, the precise point at which Sprouty intercepts RTK signaling is variable (17). In the developing fly eye, Sprouty acts upstream of Ras, whereas in the developing wing, Sprouty functions at the level of Raf (19, 34). An additional negative feedback loop is mediated by the cell surface protein Kekkon, which is specific to EGFR signaling (35, 36).

Positive feedback loops are less frequent and may act through the transcriptional activation of genes that encode activating ligands. This is found in the ventral ectoderm of Drosophila embryos or in follicle epithelium, where the activity of the EGFR pathway is amplified by induction of the expression of its ligand Vein (32, 37). In addition, the expression of rhomboid may be triggered, which subsequently facilitates the release of activating ligands (32, 38). Together, these mechanisms ensure a paracrine-mediated amplification of the RTK signal and are thus likely not as effective in regulating RTK activity in single cells.

Rau is a previously unidentified positive regulator of RTK signaling that acts within the cell. Here, we found that Rau function sustained both EGFR and FGFR signaling activity. Rau is a small 51-kD protein that harbors two RA domains, which are found in several RasGTP effectors such as guanine nucleotide–releasing factors (26, 39). Pull-down experiments demonstrated that Rau preferentially bound GTP-loaded (activated) Ras. The Rau protein is characterized by two RA domains. Although both RA domains are able to bind Ras individually, single RA domains do not show any selectivity toward the GTP-bound form of Ras. Thus, the clustering of two RA domains promotes the selection of RasGTP. In agreement with this notion, we found that Rau can form dimers or, possibly, multimers. In lysates from embryos, Rau is found in high–molecular weight protein complex (Fig. 6), suggesting that it could interact with other components of the RTK signalosome (1, 40). This way, RA domains are further clustered, and thus, GTP-loaded Ras may be sequestered. In addition, this may also contribute to the clustering of Raf, which is more active in a dimerized state (41, 42). Moreover, it was recently shown that Ras signaling depends on the formation of nanoclusters at the membrane (43). This local aggregation may further promote interaction of Ras with son of sevenless, which can trigger additional activation of the RTK signaling cascade (44). In addition, Rau harbors a class II PDZ-binding motif, suggesting that Rau can integrate further signals to modulate RTK signaling.

The activity of the EGFR and the FGFR signaling cascades is conveyed in part through the transcription factor Pointed (10, 4547). Heterozygous loss of pointed significantly increased the rough eye phenotype evoked by loss of Rau function. Moreover, upon overexpression of the constitutively active PointedP1 (48), rau expression was also increased. In line with this notion, CG8965/rau was also identified in a screen for receptor tyrosine signaling targets (49). Thus, the data suggest that Rau activation occurs after initial RTK stimulation through direct transcriptional activation through Pointed, which is similar to the activation of the secreted EGFR antagonist Argos.

Here, we have dissected the role of Rau in differentiating glial cells of the fly retina. These glial cells are borne out of the optic stalk and need to migrate onto the eye imaginal disc where some of these cells differentiate into wrapping glial cells upon contacting axonal membranes. The development of these glial cells is under the control of FGFR signaling. Initially, low activity of FGFR signaling in these glial cells is permissive for expression of seven-up, which encodes an orphan nuclear receptor of the COUP-TF (COUP transcription factor) family (50) that suppresses sprouty, but not rau, expression. Activation of Rau requires greater activity of FGFR, which is achieved only through interaction with axons (10). High activation of FGFR signaling also inhibits seven-up expression and thus relieves the negative regulation of sprouty. This negative regulation of COUP-TFII transcription factors by RTKs is also seen during photoreceptor development in the fly eye and appears to be conserved during evolution (51).

In conclusion, the Rau/Sprouty signaling module provides effective means to sustain a short RTK activation pulse, for example, during cellular differentiation. We propose that Rau dimers or multimers assemble a scaffold that favors the recruitment of RasGTP, which then could more efficiently activate the MAPK cascade. Thus, ultimately, Rau may promote the formation of Raf dimers, which might confer robustness and increased signaling intensity (41, 42). Future studies will reveal the precise conformations and complexes that enable Rau to modulate RTK signaling in fly development.

Materials and Methods


All crosses were performed on standard food at 25°C according to standard procedures. Genetic crosses were kept at 25°C unless indicated otherwise. The following fly strains were used: Mz97Gal4 (12), daGal4 (52), repoGal4 (53, 54), and UASHA-Rolled flies (22). All UASdsRNA flies were obtained from the collections at the Vienna Drosophila RNAi Centre or from Kyoto. Other stocks used in this study were obtained from the Bloomington Stock Center (Indiana University) or the Harvard collection. A glial-expressed flp source was used to induce glial clones (14). Larvae were dissected at the late third instar stage. Transgenes generated were made with phiC31-based integration in the landing sites 68E and 86Fb (55). A FRT/Flp-mediated recombination was induced between the transposons P[XP]d07408 and P[WH]f04153 to generate rauΔ.

Immunohistochemistry and in situ hybridization

Fixation and treatment of tissues for immunohistochemistry were performed as previously described (56). The antibody for Rau was custom-made, directed against a small-peptide, CAEAVNNNPAKGLGHFVYL (Pineda). The following other antibodies were used: mouse anti-Repo (Developmental Studies Hybridoma Bank), anti–β-galactosidase (1:1000; Cappel), rabbit anti-Repo (generated in-house against bacterial fusion proteins), anti-Heartless (10), anti-GFP (green fluorescent protein) (1:1000; Molecular Probes), anti–Seven-up (24), and anti-HRP Dylight 647 (1:500; Dianova GmbH). Fluorescently labeled specimens were analyzed with a Zeiss 510 or Zeiss 710 laser scanning microscope; orthogonal sections were taken with the Zeiss image browser. In situ hybridization was performed with digoxigenin-labeled antisense RNA probes according to standard procedures.

Electron microscopic analysis

All TEM analyses were performed as previously published (14). Ultrathin sections were imaged with a Zeiss EM900 with a SIS Morada digital camera. For SEM, flies were fixed in 4% formaldehyde, 2% Triton X-100 in phosphate-buffered saline (PBS) (PBT) for 20 min. After PBT washes, heads were dried in an ethanol series, further dried overnight in a desiccator, and sputter-coated with gold. Specimens were viewed with a Hitachi S-3000N scanning electron microscope.

Molecular assays

rau complementary DNA was obtained from the Drosophila Genomics Research Center, Indiana University. The coding region of rau was cloned into the pIVEX2.3d plasmid (Addgene) using oligonucleotides with Not I and Sma I restriction sites and into the pDest15 plasmid using the Gateway cloning technology (Life Technologies) for bacterial expression. For expression in flies, the rau coding region was cloned into the pUAST attB rfA plasmids, using the Gateway cloning technology (Life Technologies), containing either N-terminal 10× Myc epitopes or C-terminal 3× HA epitopes. rau deletion variants were generated with the QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies).

Protein assays

To express tagged proteins, we used the vectors pQE80L (Qiagen), pDEST15 (Life Technologies), and pIVEX2.3 (Addgene). Pull-down experiments were essentially performed as described (57). In brief, tagged proteins were expressed in BL21-AI bacteria (Life Technologies). For purification, cells of 500-ml culture were lysed in PBS by sonification. Purification of the different fusion proteins was performed with glutathione-coupled Sepharose (GE Healthcare) or Ni-NTA agarose (Qiagen) according to the manufacturer’s instructions. Sepharose-coupled small GTPases were incubated with either 0.5 mM GTPγS or 0.5 mM GDP for 10 min at 30°C. The reaction was stopped by adding 10 μM MgCl2. The different proteins were mixed and incubated for 1 hour at room temperature. After washing of the beads, the probes were separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed after Western blot according to standard procedures. For gel filtration, 6 mg of protein lysate from embryos expressing 10× Myc-tagged Rau ubiquitously was separated according to the molecular weight with a Superose 6 10/300 GL column. The column was calibrated with protein standards. The probes were separated by SDS-PAGE and analyzed after Western blot according to standard procedures.

Statistical analysis

All statistical analyses were performed with a two-tailed Student’s t test. For quantification of the wrapping phenotype, we counted only the inner ommatidia of the optic stalk, omitting the outermost two rows. Quantification of the rough eye phenotype was done with SEM images. The inner 600 ommatidia of the eyes were counted.

Supplementary Materials

Fig. S1. HA-tagged Rolled MAPK is found in the nuclei of the rau-expressing wrapping glia.

Fig. S2. Expression of rau is required for R7 photoreceptor cell development.

Fig. S3. The FGFR, sprouty, and seven-up expression network.

Fig. S4. Rau prefers to bind Ras1.

Fig. S5. Both RA domains are required for the function of Rau in vivo.

Fig. S6. Overexpression of rau deletion variants in the eye.

References and Notes

Acknowledgments: We are thankful to the Bloomington and the Harvard stock centers, J. Chung, M. Leptin, M. Hoch, T. Hummel, and M. Mlodzik for flies and antibodies; A. Aho for help in the initial characterization of the rR125 insertion and F. Babatz for help during the TEM analyses; I. Bunse, A. Fleige, and E. Naffin for sequencing and for help in biochemical experiments and the generation of transgenes; and B. Z. Shilo, S. Limmer, S. Bogdan, and A. Bauke for comments. Funding: This work was supported by grants from the Deutsche Forschungsgemeinschaft to B.A. (AL 640/2) and C.K. (SFB629 B6). Author contributions: F.S. performed all experiments, except the TEM and SEM analyses, and wrote the manuscript. T.M. performed all the TEM analyses. Y.Y.-A. did the first characterization of the rau enhancer trap insertion. H.N. performed all the SEM analyses. B.A. generated the antibodies. C.K. isolated the rau enhancer trap insertion and wrote the manuscript. All materials are available upon request. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article