To further understand the role of dAux under physiological conditions, we have isolated additional dAux mutations from several rounds of F₂ non-complementation screens. We have nine mutations in dAux, and, for eight of these nine alleles, the molecular lesions are known (Figure 1A). Furthermore, three of these alleles, namely dAux^L78H, dAux^F956*, and dAux^G257E, were generated from an FRT-containing chromosome; thus permitting dAux phenotypic analysis in clones.

To assess the strength of these dAux alleles, the lethal phases of animals carrying these dAux mutations over Df(3R)ED5021, a deletion (82A1-A2) that removes the entire dAux locus, were determined. Five of these alleles (dAux^S214*, dAux^W254*, dAux^W328*, dAux^F956*, and dAux^W1150*), when trans-heterozygous with the deletion, died prior to the larval stage. Three of them, dAux^S214*, dAux^W254*, and dAux^W328*, all contain nonsense mutations within the N-terminal kinase domain, suggesting that they are strong or null alleles. The early lethal phase of dAux^F956* and dAux^W1150*, two nonsense mutations near the C-terminus, most likely reflects the importance of the J-domain in dAux function. This analysis indicated that dAux^L78H and dAux^G257E, two missense mutations disrupting highly conserved residues in the sub-domain II and IX of the N-terminal kinase domain 49 respectively, are weaker alleles. Thus, the allele series, in descending allelic strength, can be described as: S214*≈W254*≈W328* > F956*≈W1150* > L78H≈G257E > I670K (dAux^I670K is a viable hypomorph).

Previous studies have shown that mutations in the PTEN-related region cause defects in photoreceptor specification 2526. To test whether mutations in other dAux domains have similar effects, FLP-induced mutant clones of dAux^L78H and dAux^F956*, mutations disrupting the kinase and the J-domain respectively, were stained with α-Elav, which labels the nuclei of neuronal cells 50. In wild-type tissues, an organized set of eight Elav-positive cells is formed in each cluster. In contrast, as shown in Figure 1B–E, increased numbers of disorganized Elav-positive cells were seen in both dAux^L78H and dAux^F956* mutant clones (Figure 1B–E, indicated by the absence of the nuclear GFP). This suggests that the integrity of the kinase and the J-domain, in addition to the PTEN-related region, is important for the function of dAux during neuronal cell specification. In addition, this neural hypertrophy exhibited by dAux^F956* tissues seemed to be more severe than those of dAux^L78H (in the sense that additional dAux^L78H Elav-positive cells appeared to be grouped in clusters, whereas dAux^F956* cells did not), correlating well with the allelic strength defined by the lethal phase analysis.

The Drosophila compound eye consists of regular arrays of ~800 ommatidia, each containing eight stereotypically positioned photoreceptors (R1-R8) and other accessory cells. The formation of this elaborate pattern is initiated in the larval eye imaginal disc along the morphogenetic furrow (MF, an indentation in the eye disc epithelium), which sweeps from posterior to anterior across the eye disc 51. Behind the furrow, undifferentiated cells are sequentially recruited to adopt distinct developmental fates. To further understand the roles of dAux in photoreceptor cell differentiation, we investigated the effect of dAux mutations on the specification of R8, the founder cell within each cluster. The specification of R8 cells is known to require at least two Notch-dependent events: 1) proneural enhancement: an up-regulation of atonal (ato, a proneural gene) 52 expression in cells near the morphogenetic furrow to confer them the capability to adopt a neuronal fate, and 2) lateral inhibition: a subsequent restriction of the broad Ato expression to one single R8 cell per cluster behind the MF 5354.

To determine if dAux is required for these processes, mosaic eye discs containing dAux^F956* mutant clones were stained for Ato. As shown in Figure 2A, most dAux^F956* mutant cells located near the MF expressed less Ato than their wild-type counterpart, suggesting that dAux has a role in the proneural enhancement event. Noticeably, some dAux^F956* mutant cells at the clone border still expressed an elevated level of Ato (Figure 2C and 2D, solid arrows). One plausible explanation is that dAux functions non-cell-autonomously during Notch signaling (see below); therefore, the Notch pathway in these dAux^F956* mutant cells could still be activated because they were juxtaposed to the wild-type cells.

Behind the MF in normal eye discs, the broad Ato expression is gradually restricted, by Notch-mediated lateral inhibition, to a single cell in each cluster. Unlike wild type, dAux^F956* mutant tissues contained clumps of multiple Ato-positive cells that are not restricted into single Ato-expressing cells (Figure 2A, open arrows), suggesting that the process of restricting Ato expression was disrupted. To further confirm that lateral inhibition during R8 specification was affected by dAux, FLP-induced eye disc clones were stained for Senseless (Sens), a Zn finger-containing transcription factor expressed in the nuclei of R8 cells 55. Unlike Ato, whose expression disappears in more mature R8 cells, sens expression persists in all R8 cells in the eye disc 55. Similar to the Ato staining, clusters of extra Sens-positive nuclei were seen in dAux^F956* mutant tissues after the furrow (Figure 2H). Together, these observations suggest that both proneural enhancement and lateral inhibition, two known Notch-dependent processes, require dAux.

To ask if dAux regulates other Notch-dependent processes, we examined the effect of dAux mutations on the formation of the dorsal-ventral (DV) boundary in the developing wing discs. In normal wing discs, cells in both the dorsal and ventral compartments express Notch. However, because of the modification of Notch receptor by fringe in the dorsal compartment 5657, these dorsal cells respond preferentially to Dl signaling from cells in the ventral compartment. Conversely, the cells in the ventral half respond preferentially to the other Notch ligand, Ser, which is expressed in the dorsal half. As a result, margin-specific genes like cut are expressed in a stripe of cells along the DV border, in a Notch-dependent manner 58.

To test if this process requires dAux, wild-type and dAux^F956* mosaic wing discs were stained with 2B10 α-Cut antibody 59. In wild-type wing discs, a stripe of Cut-positive cells was seen along the DV border (Figure 3A). In contrast, Cut staining was absent in dAux^F956* mutant clones located at the DV border (Figure 3B–D), suggesting that dAux is required for the DV boundary formation in wing development. Again, at a higher magnification (Figure 3D, inset), some dAux^F956* mutant cells at the clone border still expressed Cut, supporting the possibility that dAux functions non-cell-autonomously during Notch signaling (see below).

To determine in which cell dAux is required during Notch signaling, eye discs containing FLP-induced dAux^F956* mutant clones were stained for Enhancer of split (E(spl)), a transcriptional target of Notch signaling 60. We reasoned that, if dAux functions in the receiving cells (cell autonomous), all mutant cells, regardless of their locations within the clones, will be unable to activate Notch and will, therefore, not express E(spl). On the other hand, if dAux functions in the signaling cells (non-cell autonomous), mutant cells at the clone border can still receive signals from neighboring wild-type cells, and will express E(spl). Consistent with this reasoning, none of the cells mutant for Notch receptor (N^264-39) within the clones expressed E(spl) (Figure 4A–C) 6162. Conversely, as Dl is one of the ligands of this signaling cascade, several Dl^RevF10 mutant cells at the clone border expressed E(spl) (Figure 4D–F) 19. Similarly, some dAux^F956* mutant cells at the clone border clearly expressed E(spl) (Figure 4G–I). This, along with the observations described in previous sections regarding Ato and Cut expression, suggests that, like Dl, dAux is required in the signaling cells in the Notch cascade.

As our genetic data suggest that dAux functions in the signaling cells during Notch signaling, a relevant cargo of dAux-dependent endocytosis is likely to be Notch ligand. To test this, FLP-induced mutant clones of dAux^L78H and dAux^F956* were first stained with C594.9B α-Dl antibody, which recognizes the extracellular domain of Dl 63. Compared to the wild-type cells, the intensity of Dl staining was increased in both dAux^F956* and dAux^L78H mutant cells (Figure 5C&F). This increase of Dl staining intensity was more pronounced when the clones were at or near the morphogenetic furrows, consistent with the phenotype shown by a mutation in the PTEN-related region 26.

Although this elevated level of Dl seemed to accumulate around dAux mutant cell periphery (Figure 5C&F), it was not clear whether Dl proteins were trapped at the plasma membrane or in vesicular structures inside the cells. To distinguish between these possibilities, eye discs containing dAux^F956* mutant clones were stained with C594.9B under a non-permeabilized condition to label Dl at the cell surface. In wild-type tissues, a high level of surface Dl staining was first seen in cells behind the morphogenetic furrow (Figure 5G&I). In more mature clusters located in the posterior region of the eye disc, less Dl was seen at the surface, suggesting that most of Dl was internalized 64. In dAux^F956* mutant clones, the surface Dl staining appeared excessive, indicating that the previously observed peripheral Dl most likely represents Dl accumulated at the cell surface.

Knockdown of GAK function in mammalian cells was shown to greatly inhibit the internalization of EGFR 65. To determine whether the endocytic function of dAux is specific to Notch ligand, FLP-induced dAux^F956* mutant eye disc clones were stained with a α-DER antibody 66 and a α-Notch antibody (C17.9C6) 67, respectively. In wild-type eye discs, DER is expressed in cells ahead of furrow, and its expression is reduced behind the furrow 68. Compared to wild-type cells, DER staining is elevated in the mutant cells (Figure 6A–C), suggesting that the internalization of DER was inhibited. Similarly, the intensity of Notch staining was increased in dAux^F956* mutant cells (Figure 6D–F). Thus, although our genetic data suggest that dAux is required in the signaling cells during Notch signaling, the endocytic function of dAux is not limited to the Notch ligand.

We have previously observed a genetic interaction between dAux and clathrin light chain (Clc) 26. To test directly whether clathrin distribution was disrupted by dAux, the localization of a Clc-GFP fusion 69 was examined in FLP-induced dAux mutant clones in eye discs. In wild-type cells (marked by the presence of a membrane-associated, myr-mRFP), Clc-GFP appeared as vesicular structures near the cell periphery (Figure 7). In both dAux^L78H and dAux^F956* mutant cells (marked by the absence of myr-mRFP), vesicular Clc-GFP still appeared near cell periphery but its intensity was clearly elevated (Figure 7). Furthermore, large and bright spots of Clc-GFP staining could be seen (arrows), suggesting that Clc-GFP or Clc-positive structures may form aggregates in dAux mutant cells. These data suggest that normal clathrin distribution depends on both the kinase and the J-domain of dAux. Interestingly, although the level of Dl proteins appeared elevated, staining of Clc-GFP- expressing dAux^F956* mutant cells with C594.9B α-Dl antibody showed that Dl was not enriched in these large Clc-positive structures (Figure 7G–N).

To investigate the domains critical for dAux function, we generated a series of mRFP-tagged dAux derivatives, each with a particular domain or a subset of domains removed (Figure 8A). These include UAS-dAux^FL-mRFP (full-length), UAS-dAux^ΔK-mRFP (kinase domain deleted), UAS-dAux^CJ-mRFP (kinase and PTEN-related domains deleted), UAS-dAux^ΔC-mRFP (CBM deleted) and UAS-dAux^ΔJ-mRFP (J-domain deleted). Using a α-DsRed antibody (Clontech) and the Act5C-GAL4 driver, bands of expected sizes were detected in blots of extracts prepared from these transgenic flies (Figure 8B), indicating that these truncated dAux proteins were expressed.

To investigate the functional relevance of various dAux domains during Notch signaling, the abilities of these deletions, expressed using the Act5C-GAL4 driver, to rescue the supernumerary photoreceptor phenotype exhibited by dAux mutant eye discs were determined. While Act5C-GAL4 or UAS-dAux-mRFP constructs (not shown) alone had no effect, expression of dAux^FL-mRFP completely suppressed the disorganization and the extra Elav-positive cell defects in FLP-induced dAux^F956* clones (Figure 8C). Similarly, expression of dAux^ΔK-mRFP or dAux^CJ-mRFP in dAux^F956*clones displayed arrays of normal complement of Elav-positive cells, suggesting that over-expression of dAux without its kinase domain and the PTEN-related region can restore normal Notch signaling (Figure 8D&E). In addition, it should be noted that expression of dAux^FL, dAux^ΔK or dAux^CJ could rescue dAux mutants to adult viability (data not shown). In contrast, expression of dAux^ΔC-mRFP and dAux^ΔJ-mRFP could not rescue the extra Elav-positive cell defects in dAux^F956* clones (Figure 8F&G). Over-expression of dAux^ΔJ-mRFP appeared to have some dominant negative effects. For instance, animals mutant for dAux^L78H/dAux^W328* died during the larval stage. However, these mutants died before the larval stage when the J-domain deletion was expressed under the control of Act5C-GAL4 (data not shown). The results of these rescue experiments were confirmed using two independent transgenic lines from each construct and held true for another dAux allele (dAux^L78H), indicating that it is not allele specific. Together, these results suggest that while the kinase and PTEN-related domains are less critical, the J-domain and the CBM region are essential for dAux function.

All fly crosses were carried out at 25°C in standard laboratory conditions unless otherwise specified. To facilitate the analysis of dAux phenotypes in clones, 5-5Z3515, an FRT-containing P-element insertion from the DrosoDel project (Cambridge, UK), was used to isolate additional dAux alleles. This strategy was chosen because 1) the conventional FRT site for the third chromosome right arm 82 is more distal from the centromere then the dAux locus, therefore not applicable for generating dAux mutant clones, and 2) crossovers between FRT^5-5Z3515 and dAux would have been astronomically rare, as the P element insertion in 5-5Z3515 is only 20 kb closer to the centromere than the dAux locus.

The screens for additional dAux alleles on FRT^5-5Z3515 chromosomes were performed as previously described 26. Briefly, w; FRT^5-5Z3515 males were mutagenized with 25 mM ethyl methane sulfonate (Sigma), and mass mated with w/w; TM3, Sb/TM6B, Hu virgins. Progeny were then individually mated with dAux^I670K, p [w⁺]/TM6B, Hu flies, and those that failed to complement dAux^I670K were recovered and maintained over TM6B or TM3 balancers. To determine the mutations in dAux alleles, coding regions were amplified from genomic DNA extracted from homozygous mutant embryos by PCR. Multiple independent PCR products were analyzed by direct sequencing.

The mRFP-tagged dAux deletions were constructed using PCR and standard cloning techniques. The dAux deletions, dAux^ΔK, dAux^CJ and dAux^ΔJ, correspond to amino acids 339–1165, 762–1165 and 1–998 respectively. dAux^ΔC lacks the amino acids 768–1049. In all dAux constructs, mRFP or EGFP was appended in-frame at the C-terminus. The sequence and construction detail will be provided upon request. All the constructs were verified by sequencing, and multiple transgenic flies carrying the constructs were generated by P-element-mediated transformation 83.

Immuno-staining of eye and wing imaginal discs was performed as previously described 84. Rat α-Elav 7E8A10 (DSHB, Iowa), mouse α-Atonal 52, guinea pig α-Senseless 55, mouse α-Cut 2B10 (DSHB, Iowa), α-E(spl) mAB323 60, mouse α-Dl C594.9B (DSHB, Iowa), mouse α-Notch C17.9C6 (DSHB, Iowa), and rat α-DER 66 were used at 1:100, 1:3000, 1:1000, 1:100, 1:2, 1:100, 1:100, and 1:50 dilutions, respectively. Fluorescently conjugated secondary antibodies (Molecular Probes) were used according to the manufacturer's instructions. Fluorescent microscopy was performed using the Olympus BX61 microscope equipped with the Olympus DSU confocal system and processed with Photoshop (Adobe) and Volocity (Improvision).

For surface Dl labeling, mosaic eye discs were dissected and fixed in 4% paraformaldehyde. The peripodial membranes were partially removed, and the discs were then stained with mouse α-Dl C594.9B and washed with PBS in the absence of any detergent. The stained eye discs were examined using a BioRad MRC1024 laser confocal microscope (Nikon OPTIPHOT-2) and the images were processed with Adobe Photoshop.

To prepare fly extracts, flies expressing mRFP-tagged dAux derivatives were homogenized in 2× SDS-loading buffer containing 0.2 mM DTT (10 flies/100μl), boiled immediately for 5 minutes and separated by SDS-PAGE using 10% acrylamide gel. The gel was transferred to nitrocellulose membrane and probed with α-DsRed rabbit polyclonal antibody (Clontech) at 1:1000 dilution, followed by HRP-conjugated goat anti-rabbit secondary antibody (Jackson Lab). The immunodetection was performed using ECL substrate (Amersham Biosciences).