In this report we demonstrate a role of awd in regulating Notch signaling via its endocytic function including surface internalization and vesicle trafficking. This conclusion is based on our results that show: (1) multiple Notch target genes are mis-expressed in follicle cells and wing discs; (2) Notch accumulates in enlarged early endosomes; and (3) awd function is required for the Rab5 activity in early endosome maturation. Our results also indicate that during vesicles trafficking, the Awd action is downstream of the S2 cleavage, since over-expressed of NEXT accumulated intracellularly and could not rescue the awd defect. The same NEXT over-expression strategy could rescue the shi/dynamin defect
[57, 65], strongly supporting the notion that the Awd action on Notch signaling is post-membrane invagination. Since over-expression of NICD could rescue the awd defect, the Awd action is likely upstream or in parallel to the S3 cleavage event (γ-secretase activity). Although a role of awd in promoting the activity of γ-secretase cannot be completely ruled out, we considered this possibility unlikely. First, awd is a known endocytic factor demonstrated in multiple tissues including neurons, trachea, and follicle cells
[22–25]. Second, neither the expression level nor the expression pattern of Presenilin, the catalytic subunit of γ-secretase, is altered in awd mutant cells. Third, if the defect is in γ-secretase function, it would be expected that Notch should accumulate in Hrs-positive MVBs
. On the contrary, we did not observe such ectopic accumulation of Notch in Hrs-positive vesicles. Therefore, our results, in aggregate, suggest that the main action of Awd on Notch signaling is via its endocytic activity promoting the transition from early endosomes to late endosomes. However, potential defects downstream of γ-secretase cleavage, such as trafficking to nucleus, in awd mutant cannot be formally ruled out.
One curious exception for the awd function in relation to Notch signaling is found in the border cells. As we have recently reported
, during the migration of these cells, Awd expression is down-regulated. Re-expression of Awd can lead to reduction of surface receptors, such as PVR that is critical for directional movement, resulting in defective migration. Interestingly, Notch signaling is also important for border cell migration
. It, therefore, appears that Notch signaling in these specialized cells does not require Awd activity or is insensitive to Awd protein levels. To test this, we compared Notch expression in border cells with or without Awd re-expression. In wild-type border cells (no Awd), Notch is located on the cell surface as well as in the cell body, consistent with active signaling (data not shown). Forced re-expression of Awd in the border cells does not alter this pattern. This may be because Notch is already actively internalized; increasing the Awd level cannot further enhance such activity. Indeed, endocytosis is intrinsically highly active in border cells
[24, 67]. Alternatively, the differential dependence of Notch on Awd activity may be a function of how Notch is activated, not how Awd functions differently in different cell types. For example, Dobens et al.
 have shown that the Notch ligand Delta may be co-expressed with Notch in the same border cells. Recent reports have hinted that the requirement of endocytosis for Notch signaling may depend on the ligand-receptor relationship (for example, ligand-dependent or -independent, trans- or cis-activation, and so on)
. We, therefore, consider that the apparent Awd-independent Notch signaling in border cells has more to do with the intrinsic Notch signaling mechanism in these cells, and less to do with the function of Awd.
Our results indicate that the Notch signaling defect in awd mutant cells is the failure to deliver Notch past the Rab5-dependent early endosomal stage. On the other hand, the ESCRT complex mutants, which are defective in late endosome formation, promote Notch signaling
[34, 55]. Taken together, it appears that Notch activation occurs in the intermediate stage between early endosome formation and late endosome entry. Transition from early endosomes to late endosomes is accompanied by cargo sorting, intravesicular invagination and acidification of the luminal contents. Curiously, the matured early endosome and MVB marker hrs mutant has no effect on Notch signaling
, which indicates that endosomal cargo sorting per se is not required for Notch signaling. We have also shown that awd mutant cells do not exhibit altered levels of Lysotracker staining and that endosomal Notch remains on the surface of enlarged endosomes in awd mutants. The exact nature of this transition state that favors Notch processing, therefore, requires further analysis. The endocytic function of awd has traditionally been described as a ‘GTP supplier’ for Dynamin, based on genetic interaction data and logical extrapolation because of the GTP producing activity of Awd
. In this report, we demonstrate that, in relation to Notch signaling, awd functions downstream of, but not directly on, dynamin. It is instead critical for Rab5 activity. This is supported by the following evidence: 1) Notch in awd mutant accumulates in Avl-containing vesicles. Therefore, the awd defect is post Dynamin-mediated cleavage of membrane invagination. 2) Rab5CA can push Notch into enlarged early endosomes but failed to rescue the awd phenotype, thereby strengthening the notion that awd defect is post Shi/Dynamin function. 3) The Notch accumulation pattern in shi mutant is different from that in awd mutant. 4) Over-expression of NEXT could not rescue awd defect. The same NEXT over-expression strategy could rescue the shi defect, strongly supporting the notion that the Awd action concerning Notch signaling is post-membrane invagination
[57, 65]. It should be noted that we did observe surface accumulation of NECD antibody-detected Notch molecules, likely representing the full-length Notch not engaged in ligand binding and signaling. This indicates that Awd can affect constitutive internalization of full-length Notch.
The requirement of endocytosis in the signal-receiving cells for Notch activation has been amply demonstrated
. It has been shown that Notch signaling in follicle cells after stage 6 requires Delta
. Since in this report we show that Notch signaling cannot occur in the follicle cell without awd function, we conclude that, at least in follicle cells, endocytosis is a requisite process for ligand-dependent Notch signaling.
The involvement of endocytosis in Notch signaling is significant since many of the endocytic components shown to regulate Notch signaling have also been implicated in carcinogenesis. For example, V-ATPase is required for Notch signaling while mutations in ESCRT components, such as Tsg101, result in increased Notch signaling. V-ATPase has generally been considered an oncogene
 because it is associated with acidification of tumor cells. ESCRT components, on the other hand, have been shown to suppress tumor formation because they down-regulate surface growth factor receptor signaling
. As such, attempts to design therapeutics based on these prevalent functions should take into account the effects on Notch signaling, since the relationship between Notch signaling and carcinogenesis is context-dependent
[35, 72, 73].