Drosophila cbl Is Essential for Control of Cell Death and Cell Differentiation during Eye Development

Background Activation of cell surface receptors transduces extracellular signals into cellular responses such as proliferation, differentiation and survival. However, as important as the activation of these receptors is their appropriate spatial and temporal down-regulation for normal development and tissue homeostasis. The Cbl family of E3-ubiquitin ligases plays a major role for the ligand-dependent inactivation of receptor tyrosine kinases (RTKs), most notably the Epidermal Growth Factor Receptor (EGFR) through ubiquitin-mediated endocytosis and lysosomal degradation. Methodology/Principal Findings Here, we report the mutant phenotypes of Drosophila cbl (D-cbl) during eye development. D-cbl mutants display overgrowth, inhibition of apoptosis, differentiation defects and increased ommatidial spacing. Using genetic interaction and molecular markers, we show that most of these phenotypes are caused by increased activity of the Drosophila EGFR. Our genetic data also indicate a critical role of ubiquitination for D-cbl function, consistent with biochemical models. Conclusions/Significance These data may provide a mechanistic model for the understanding of the oncogenic activity of mammalian cbl genes.


INTRODUCTION
Normal cellular function and tissue homeostasis is dependent on the precise regulation of several signal transduction pathways that control cell proliferation, cell differentiation and cell survival. Each cell integrates an array of extracellular signals into appropriate cellular responses. Deregulation of these processes causes developmental abnormalities and human diseases including cancer. However, we still lack a clear understanding of how these processes are integrated in the context of a developing organism.
The development of the retina in the Drosophila compound eye has long been a model system to study how extra-cellular signaling generates precise cellular differentiation patterns (reviewed by reference [1]). The compound eye is composed of ,800 ommatidia, repetitive units each containing a precise number of different cell types. The adult fly eye develops from a monolayer epithelium-the eye imaginal disc. In early larval stages cells in the eye imaginal disc proliferate to provide the cellular mass for eye development. During mid-third instar larval stage, cellular differentiation starts at the posterior end of the eye imaginal disc, which coincides with formation of the morphogenetic furrow (MF) that sweeps across the disc from posterior to anterior [1,2]. As the MF progresses towards the anterior, cells located behind the MF start differentiating into distinct cell types in a strict sequence to form the ommatidium. Each ommatidium has eight photoreceptor neurons or ''R'' cells (R1-R8). R8 is the first R cell to be specified, and serves as the founder cell for recruitment of the other R cells in the order R2/R5RR3/R4RR1/R6RR7, followed by four nonneuronal cone cells during late third instar larval stage, and three classes of pigment cells during early pupal stages [3]. Finally, after specification of these cell types has been completed, all surplus undifferentiated cells are removed by apoptosis [3,4]. This occurs between 26-30 hours after puparium formation [5].
The specification of cell fate in the developing Drosophila retina is controlled by combinatorial signaling. Two receptor tyrosine kinases (RTKs), the epidermal growth factor receptor (EGFR) and Sevenless (Sev), contribute to retinal development [6,7]. Activation of EGFR by the secreted ligand Spitz (sSpi), a transforming growth factor (TGF-a) homologue, regulates the specification of all R cells in the developing eye, except R8 [6,8,9]. Over-expression of sSpi causes an over-recruitment of all cell types, while expression of dominant negative EGFR (EGFR DN ), or shifting a temperaturesensitive EGFR allele to the non-permissive temperature leads to an impairment of differentiation [6,10,11]. Ommatidia mutant for argos, gap1 and sprouty, three negative regulators of EGFR, contain extra R and cone cells surrounded by more secondary and tertiary pigment cells in the lattice [12][13][14][15][16][17][18]. In addition, EGFR signaling is utilized for cell survival during Drosophila eye development, due to its negative regulation of hid, a cell death-inducing gene [19][20][21]. In contrast to the EGFR which controls the development of all R cells in the ommatidium except R8, sev is required only for R7 differentiation [22].
As important as the activation of cell surface receptors is their inactivation for appropriate control of cell number and differentiation. The proto-oncogene Casitas B-lineage lymphoma (Cbl) was first identified as a retroviral transforming gene product that induces pre-B cell lymphomas and myeloid leukemia [23]. Cbl is involved in many signaling events through its function as a multidomain adaptor protein and has been best characterized as a negative regulator of RTKs, mostly EGFR (reviewed by [24,25]). This concept grew out of genetic studies performed in C.elegans in which Sli-1, the Cbl ortholog, attenuates the activity of Let-23, the EGFR equivalent, in vulval development. [26]. Mammals contain three Cbl genes known as c-Cbl, Cbl-b and Cbl-3, which function as negative regulators of EGFR [25,27,28]. Knock-out mice of ccbl, cbl-b and cbl-3 have no obvious developmental phenotypes except in the immune system suggesting that they are functionally redundant [29][30][31][32]. Drosophila has only one cbl gene, referred to as D-cbl [33][34][35], eliminating the problem of redundancy, and the genetic characterization of D-cbl mutants may reveal more information about its oncogenic role. For example, an isoform of D-cbl, which mimicked the oncogenic viral cbl (v-cbl), demonstrated that v-cbl acts in a dominant negative manner [35]. Furthermore, consistent with studies in C.elegans and mammalian cell culture, D-Cbl has been shown to function as a negative regulator of EGFR during dorsoventral patterning in oogenesis and guided migration of border cells [36,37]. A loss-of-function analysis of D-cbl for eye development in Drosophila has not been reported.
Here, we present the characterization of the D-cbl mutant phenotype for eye development. D-cbl mutants display overgrowth and lack of developmental apoptosis. Mutant ommatidia contain increased numbers of photoreceptors (mostly R7), cone and pigment cells. Genetic interaction tests indicate that D-cbl regulates the EGFR pathway during eye development consistent with its proposed role as negative regulator of EGFR. Our genetic data indicate a critical role of ubiquitination for D-cbl function, in accord with biochemical models. In summary, these data provide a genetic model for the understanding of the oncogenic activity of mammalian cbl genes.

Isolation and characterization of D-cbl mutants
In a mutagenesis screen, we isolated five mutant D-cbl alleles as recessive suppressors of the small eye phenotype caused by expression of the pro-apoptotic gene hid under control of the eyespecific GMR enhancer (GMR-hid; Fig. 1A-D). For details about the GMR-hid suppressor screen see Material and Methods, and references [44][45][46]. Because D-Cbl is a known negative regulator of EGFR, and because increased EGFR activity inhibits the proapoptotic function of Hid [19][20][21], the isolation of D-cbl mutants as suppressors of GMR-hid can be explained by its effect on EGFR. However, what sparked our interest in characterizing D-cbl for eye development are the mutant phenotypes without GMR-hid expression. The eyes appear rough and bulgy with larger ommatidia (Fig. 1E, F), and D-cbl mutant heads are overgrown (Fig. 1G, H). The overgrowth phenotype is already visible in eye-antennal imaginal discs of 3 rd instar larvae (Fig. 1I, J). Both, the strong rough eye and the overgrowth phenotype cannot be solely explained for by inhibition of apoptosis. Thus, we characterized the D-cbl mutant phenotype during eye development in more detail.
DNA sequencing revealed missense and non-sense mutations (Fig. 1K). D-cbl 7 and D-cbl L31 affect the same residue, the highly conserved Arg406 residue in the RING domain. The remaining alleles, D-cbl K26 , D-cbl 4 and D-cbl 8 introduce premature STOP codons at positions 60, 116 and 178, respectively (Fig. 1K). D-cbl 4 is identical to a previously isolated allele, D-cbl F165 [36]. At least Dcbl K26 can be considered a null allele of D-cbl. Interestingly, all isolated alleles affect both the large and the small isoform of D-cbl. We did not recover mutant alleles that affect only the large isoform. All experiments in this study were performed with at least two alleles, the null allele D-cbl K26 and the RING domain mutant D-cbl 7 , both of which show identical results.
D-Cbl regulates the EGFR pathway in the Drosophila eye D-Cbl has previously been shown to be a negative regulator of EGFR signaling during dorsoventral patterning in oogenesis and border cell migration [36,37]. A similar loss-of-function analysis of D-cbl has not been done for eye development. We have performed several genetic interaction tests to determine whether D-Cbl controls EGFR signaling during eye development. First, heterozygosity of Dcbl considerably rescued the rough eye phenotype caused by overexpression of dominant negative ras N17 (Fig. 2A, D). Second, Dcbl K26 dominantly suppresses the eye phenotype caused by misexpression of the active form of the repressor yan (yan act ) (Fig. 2B, E), a target gene negatively regulated by EGFR signaling [47,48]. Third, to more directly assess a role of D-cbl for the regulation of EGFR, we analyzed the effect of D-cbl mutants on the small eye phenotype caused by expression of a dominant negative allele of EGFR, EGFR DN [6], under control of the eye-specific enhancer GMR (GMR-EGFR DN ) (Fig. 2C). EGFR DN lacks the intracellular tyrosine kinase domain, but leaves the transmembrane and extracellular domains intact [6]. EGFR DN is able to dimerize with endogenous EGFR, but trans-phosphorylation upon ligand binding does not take place and thus the dimer is unable to signal. However, the inhibition of endogenous EGFR by EGFR DN is not complete as some R cells still survive and differentiate [6] (data not shown), which is not observed in strong EGFR mutant clones [49]. Thus, the small eye phenotype of GMR-EGFR DN is caused by partial inhibition of endogenous EGFR. In D-cbl K26 mutant clones, the GMR-EGFR DN phenotype is moderately strongly suppressed (Fig. 2F) implying that loss of D-cbl partially restores the activity of endogenous EGFR inhibited by EGFR DN . Combined, these data suggest that D-cbl mutants contain increased EGFR activity suggesting that D-cbl negatively controls EGFR activity.
To further confirm this notion we tested three molecular markers in loss-of-function and gain-of-function analyses of D-cbl. First, phospho-tyrosine labeling (p-Tyr) as a marker of RTK activity is increased in D-cbl clones in third instar larval eye discs (Fig. 3A). This is well visible in D-cbl clones crossing the MF and posterior to the MF (Fig. 3A). In the reverse experiment, overexpression of D-cbl, p-Tyr labeling is significantly reduced (Fig. 3B). Because EGFR is the only known RTK acting posterior to the MF in eye development (except Sev which we can exclude as a target of D-cbl, see below), the increased p-Tyr labeling is mainly caused by increased EGFR activity.
Second, in wild-type third instar larval eye discs, immunolabeling with dpERK, an antibody that recognizes activated MAPK acting downstream of EGFR, is detectable in one ommatidial column immediately posterior to the MF [50] (Fig. 3E). Further posteriorly, dpERK is not detectable suggestive of MAPK inactivation. In D-cbl mutant clones, dpERK labeling persists further posteriorly to the MF (Fig. 3C) suggesting lack of MAPK inactivation. Furthermore, the reverse experiment, overexpression of D-cbl, results in loss of dpERK labeling (Fig. 3D).
The third molecular marker used is Yan, a transcriptional repressor in the nucleus. In response to EGFR signaling, yan transcription is inhibited and Yan protein is proteolytically degraded [47,48]. In pupal eye discs 35 hours after puparium formation (APF) Yan protein is strongly reduced in D-cbl clones (Fig. 3F) suggesting that they contain increased EGFR activity. Taken together, these data suggest that D-cbl negatively regulates EGFR signaling during eye development.

D-cbl mutants block apoptosis and cause overrecruitment of all cell types in the eye
In the Drosophila eye, EGFR signaling is utilized for cell survival and cell differentiation. Because D-cbl mutants cause increased EGFR signaling, we tested whether this has consequences for cell survival and cell differentiation in the fly eye. EGFR function is anti-apoptotic due to its negative regulation of hid [19][20][21]. Developmental cell death during eye development is maximal between 26 and 30 hours after puparium formation (APF) when surplus, undifferentiated cells are eliminated [5]. This elimination requires the pro-apoptotic function of hid [20,21,51]. To determine whether D-cbl mutants affect cell death, we labeled 28 hours APF eye discs with an antibody that recognizes cleaved and thus activated Caspase-3 (Cas3). In D-cbl K26 mutant clones, developmental cell death is significantly blocked (Fig. 4A, B). This finding is consistent with the isolation of D-cbl mutants as suppressors of GMR-hid (Fig. 1C, D).
Next, we tested whether D-cbl mutant clones display differentiation defects. The eye disc is fully differentiated by 42 hrs APF  [3]. Labeling of 42 hrs APF eye discs with Elav antibody, a R cell marker, revealed that each D-cbl mutant ommatidium contains on average 11.4561.03 (n = 35) R cells compared to eight in wildtype ommatidia (Fig. 4C, D). In addition, D-cbl mutant ommatidia contain up to eight cone cells compared to four in wild-type ommatidia (Fig. 4E, F) and about three primary pigment cells instead of two in wild-type (Fig. 4G, H). The total numbers of secondary and tertiary pigment cells has increased from nine in wild-type to an average of fourteen per D-cbl ommatidium (Fig. 4G, H). In summary, these data show that Dcbl mutants cause increased cellular survival and over-recruitment of all cell types disrupting the regular array of the ommatidia in the developing Drosophila eye.

D-cbl specifically affects R7 development
We determined whether the over-recruitment of R cells in D-cbl clones affects all photoreceptors randomly, or specific types of photoreceptors. Labeling with antibodies directed against Rough (R2,5,3,4-specific) and Seven-up (Svp, R3,4,1,6-specific) did not reveal significant differences between D-cbl and wild-type tissue (Fig. 5A-D). However, labeling with an antibody against Prospero (Pros), a R7 marker, reveals 4.560.53 (n = 35) cells per D-cbl mutant ommatidium whereas in wild-type only one cell is detected (Fig. 5E, F). To determine whether the additional Pros-positive cells have photoreceptor character, we performed Pros and Elav double labelings. In D-cbl mutant clones, all Pros-positive cells are Elav-positive, suggesting that most, if not all, additional photoreceptors in D-cbl ommatidia are R7 cells (Fig. 5G, H). Thus, among the R cells, D-cbl specifically affects the specification of R7. A similar and specific increase in the number of R7 cells has been observed in eye tissue mutant for other negative regulators of EGFR signaling including argos, gap1, and sprouty [12,[16][17][18]. Thus, the over-recruitment phenotype in D-cbl mutants is typical for increased EGFR activity.

Loss of D-cbl can compensate for loss of sevenless
In addition to EGFR, Sevenless (Sev), a second RTK, is involved in eye development. sev is only required for R7 specification and consequently sev mutants do not contain R7 cells [22]. However, expression of dominant active sev constructs gives rise to the recruitment of multiple R7 cells [52], similar to the D-cbl phenotype. Thus, it is formally possible that D-cbl also regulates Sev.
We tested this possibility. If the additional R7 cells in D-cbl clones result from hyper-activity of Sev, then these cells should require sev for their specification. However, genetic removal of sev using the null allele sev d2 [53] has no effect on R7 specification in D-cbl K26 mosaic background (Fig. 6). The average number of total R cells in sev d2 , D-cbl K26 mutant ommatidia is 11.1260.74 (n = 35), containing more than four R7 cells per ommatidium (Fig. 6). The GFP-positive area is single mutant for sev d2 and hence does not form R7 (Fig. 6B, D). Thus, in D-cbl K26 mutants, the R7 cells can complete their differentiation program even in the absence of Sev, suggesting that the increase of EGFR signaling in D-cbl K26 can compensate for loss of Sev. A similar sev-independent mode of R7 specification has been observed for other negative regulators of the EGFR pathway such as gap1 and sprouty [16][17][18].

DISCUSSION
The phenotypic characterization of D-cbl mutants for eye development in Drosophila allows making four important conclu-  sions: First, D-cbl plays an important role for the negative regulation of EGFR activity. Second, loss of D-cbl causes severe mis-specification and over-recruitment defects. Third, loss of D-cbl blocks developmental apoptosis. Fourth, loss of D-cbl causes tissue overgrowth at the organismal level, providing a model to study the oncogenic activity of mammalian cbl genes.

D-Cbl negatively regulates EGFR
D-cbl mutant ommatidia contain increased numbers of R7, cone and pigment cells. Similar phenotypes have been observed in mutants of other negative regulators of the EGFR pathway such as gap1, argos and sprouty [12,[16][17][18]. We also confirmed a regulatory role of D-cbl for EGFR activity in genetic interaction studies (Fig. 2). Furthermore, using the molecular markers p-Tyr, dpERK and Yan as readouts for EGFR activity we showed that D-cbl clones contain increased RTK and MAPK activity. Likewise, overexpression of D-cbl blocks p-Tyr and dpERK labeling. Thus, the D-cbl mutant phenotypes in the eye are consistent with increased EGFR activity and suggest that D-Cbl negatively regulates EGFR, in accord with previous reports [34,36,37].
In contrast to argos, gap1 and sprouty, D-cbl does not appear to regulate all RTKs. For example, D-cbl does not influence Torso [36], a RTK involved in specification of the termini in the Drosophila embryo [54]. Here, we have demonstrated that D-cbl does not control the Sev RTK. This difference likely reflects the direct mode of EGFR regulation by D-cbl, while argos, gap1 and sprouty act downstream in the Ras/MAPK pathway which is shared by all RTKs. Biochemical data has demonstrated that mammalian Cbl proteins directly bind to tyrosine-phosphorylated EGFR and ubiquitylates it for endocytosis and lysosomal degradation [38][39][40][41][42][43]. Although we have not verified a similar biochemical mechanism for the interaction between Drosophila EGFR and D-Cbl, it is likely that the mechanism is similar. This notion is supported by the isolation of two D-cbl alleles affecting the RING domain (Fig. 1K). The RING domain contains an E3 ubiquitin ligase activity which targets the EGFR for ubiquitylation [39]. The mutant phenotype of D-cbl 7 affecting the RING domain is indistinguishable from the null allele D-cbl K26 (data not shown), further supporting an essential role of ubiquitylation for D-cbl function.
It is unclear why only the number of R7 cells is affected whereas the remaining R cells are normal in number although R1-R6 also require the EGFR for specification. However, it suggests that the sequence of events during R cell specification is normal in D-cbl clones. The fact that D-cbl clones contain up to four additional R7 cells is likely due to the fact that R7 and the four cone cells are developmentally equivalent. These five cells express sev and all have the capacity to become R7 if Sev or downstream components are activated [52]. Thus, the additional R7 cells in D-cbl clones likely represent transformed cone cells.
However, this transformation does not mean that the cone cells are lost in D-cbl clones. In contrast, we even observe an overrecruitment of cone cells. Interestingly, the cone cell overrecruitment in D-cbl mutants does not occur during pupal stages as suggested for gap1 [16]. It occurs at the correct developmental time in late third instar eye development (data not shown). Thus, the over-recruitment of several different cell types in D-cbl clones follows the same rules of reiterative use of the EGFR as compared to wild-type.

EGFR-independent phenotypes of D-cbl
Despite the fact that the D-cbl mutant phenotypes are similar to the ones described for argos, gap1 and sprouty, we noticed at least two phenotypes which appear to be specific for D-cbl. First, D-cbl mutant heads and imaginal discs are overgrown (Fig. 1). Second, the spacing between the ommatidial clusters is increased (see examples in Fig. 4D, 4F, and 6D). Similar phenotypes have not been observed for gap1, argos and sprouty [12][13][14][15][16][17][18] (data not shown). It is unclear how these phenotypes are caused, but they may be independent of EGFR. Further studies are needed to clarify these observations.

Implications for mammalian Cbl and oncogenesis
This work may also have some important implications for our understanding of the oncogenic nature of mammalian cbl [55]. Increased proliferation and reduced apoptosis are hallmarks of cancer [56]. v-cbl is a retroviral transforming oncogene causing pre-B lymphoma and myeloid leukemia [23]. v-cbl contains only the TKB domain [24] and behaves genetically as a dominant negative mutant [35]. Furthermore, inappropriate activation of mammalian EGFR can lead to various forms of human cancers [57][58][59][60]. Thus, genetic studies in model organisms may contribute to our understanding of oncogenic processes in mammals.

MATERIALS AND METHODS Identification of D-cbl mutant alleles
Eye-specific expression of hid under GMR enhancer control (GMRhid) results in an eye ablation phenotype (Fig. 1B). Using the GMR-hid ey-FLP (GheF) method [44], we conducted an EMSmutagenesis screen for chromosome arm 3L to identify recessive suppressors of the GMR-hid eye ablation phenotype. This method induces homozygous mutant clones in the eye by ey-FLP/FRTmediated mitotic recombination in otherwise heterozygous background [61]. For GheF screening, ey-FLP; FRT80 males were incubated with 25 mM EMS in 5% sucrose solution for 24 hours. After recovery for 3 hours, they were mated to GheF; FRT80 P[w + ] females and incubated at 25uC. 45,000 F1 progeny were screened for suppression of the GMR-hid small eye phenotype. In the screen for 3L, 4 dronc [44] and 5 D-cbl alleles (this study) were recovered.