The tumour suppressor Lethal (2) giant discs (Lgd) is a regulator of endosomal trafficking of the Notch signalling receptor as well as other transmembrane proteins in Drosophila. The loss of its function results in an uncontrolled ligand-independent activation of the Notch signalling receptor. Here, we investigated the consequences of loss of lgd function and the requirements for the activation of Notch. We show that the activation of Notch in lgd cells is independent of Kuz and dependent on γ-secretase. We found that the lgd cells have a defect that delays degradation of transmembrane proteins, which are residents of the plasma membrane. Furthermore, our results show that the activation of Notch in lgd cells occurs in the lysosome. By contrast, the pathway is activated at an earlier phase in mutants of the gene that encodes the ESCRT-III component Shrub, which is an interaction partner of Lgd. We further show that activation of Notch appears to be a general consequence of loss of lgd function. In addition, electron microscopy of lgd cells revealed that they contain enlarged multi-vesicular bodies. The presented results further elucidate the mechanism of uncontrolled Notch activation upon derailed endocytosis.
Work in recent years has established a fundamental role for the endosomal pathway in the regulation of cell signalling and receptor degradation (Dobrowolski and De Robertis, 2012). Trafficking of signalling receptors is initiated through the endocytosis and transport to the early endosome (EE) in early endosomal vesicles, which eventually fuse with the EE. After reaching the EE, the receptors are either separated from their ligands and transferred back to the plasma membrane, or remain in the maturing endosome (ME) to become degraded in the lysosome. During maturation, the receptors are concentrated in domains of the limiting membrane and translocated in the lumen of the maturing endosome through pinching off the corresponding region as intraluminal vesicles (ILVs). The formation of the ILVs achieves the separation of the intracellular domain of the receptors from the cytosol. This is important for their complete degradation and the cancellation of signalling. The events in the endosomal pathway are controlled by small GTPases, chief among them are Rab5 and Rab7 (Huotari and Helenius, 2011). Rab5 controls fusion of early endosomal vesicles with themselves as well as the EE. It further initiates the formation of ILVs, which is mediated by the four in sequence acting ESCRT complexes (ESCRT 0–III). The central component of the ESCRT-III complex in Drosophila is encoded by shrub. It encodes the Drosophila orthologue of mammalian CHMP4 and yeast Snf7 (Sweeney et al., 2006). As a result of the activity of the ESCRT complexes the maturing endosome (ME) contains an increasing number of ILVs and is called a multi-vesicular body (MVB). Rab7 orchestrates the fusion of the ME with the lysosome by recruiting effector proteins, such as the components of the HOPS tethering complex, which mediate recognition of the target membrane and the activation of SNARE controlled fusion. The fusion ends the life of ME. During maturation, the lumen of the ME acidifies through the activity of the vacuolar ATPase (vATPase) (Huotari and Helenius, 2011). Recent work has shown that the activity of vATPase is essential for the ligand-dependent activation of the Notch signalling pathway, suggesting that acidification is required for activation (Vaccari et al., 2010).
The Notch pathway is present in all metazoans and plays a major role in many developmental and homeostatic processes as well as in disease (Weinmaster and Fischer, 2011). In Drosophila the Notch receptor is activated by two ligands, Delta (Dl) and Serrate (Ser). The binding of the ligand initiates the cleavage and removal of the extracellular domain (NECD) in a process named ecto-domain shedding. This S2 cleavage is performed by the ADAM 10 metalloprotease encoded by kuzbanian (kuz) in Drosophila and creates an intermediate called NEXT (Notch extracellular truncation), which is cleaved a second time in its transmembrane in a process called RIP (regulated intramembraneous proteolysis). RIP is performed by the γ-secretase complex and causes the release of the intracellular domain (NICD) into the cytosol from which it migrates into the nucleus. There, it associates with the CSL transcription factor Suppressor of Hairless, Su(H), to activate the target genes. Recent work suggests that the S3 cleavage can occur in the EE (Vaccari et al., 2008).
The endosomal pathway regulates the degradation of Notch in the lysosome in a constitutive and ligand-independent manner (Jékely and Rorth, 2003). Recent work indicates that the pathway can be activated in a ligand-independent manner if the endosomal pathway is defective at certain stages. Activation can be observed if the function of lethal (2) giant discs (lgd) or components of the ESCRT complexes is lost (Thompson et al., 2005; Vaccari and Bilder, 2005; Childress et al., 2006; Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). lgd encodes a member of an evolutionary conserved family whose members contain four tandem repeats of the novel DM14 domain in addition to one C2 domain (Childress et al., 2006; Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). Loss of lgd function results in the accumulation of Notch in enlarged endosomes. The cause for this accumulation is not clear. However, the analysis so far suggests that Lgd is required for trafficking of Notch and other transmembrane proteins through the endosomal pathway. Moreover, enlargement of the endosomes in lgd cells has only been documented with the fluorescent microscope and it is possible that this is due to a mis-distribution of the endosomal markers on the endosomal membrane rather than due to a morphological change.
We have recently shown that the DM14 domains are required for direct interaction with Shrub (Troost et al., 2012). This interaction is important for the full activity of Shrub in vivo. Genetic experiments revealed that shrub and lgd act antagonistically with respect to Notch activation (Troost et al., 2012). The activation of Notch in lgd cells can be suppressed by lowering the activity of shrub (genotype: lgd shrub/lgd +), although loss of function of both genes alone results in activation of the Notch pathway. The reason why Notch is not activated in the double mutant cells is not clear. In contrast both genes act in concert with respect of endosome morphology indicated by the observation that lgd shrub/lgd + cells contain dramatically enlarged MEs.
The details of Notch activation in lgd cells are not clear. Previous work suggests that the ligand-independent activation of Notch in lgd cells requires the activity of the γ-secretase complex (Childress et al., 2006; Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). This conclusion is based on the observation that the activity of Psn is required for activation of the Notch pathway in lgd cells. However, there is evidence that Psn has additional functions that are independent of γ-secretase (see Ho and Shen, 2011). Moreover, it is not known whether the Kuz-mediated S2 cleavage is required for the activation of Notch in lgd cells. So far the function of lgd has been only investigated in imaginal disc cells. It is therefore also not known whether the activation of Notch is a consequence of loss of lgd function in other tissues.
Two orthologues of Lgd exist in genomes of mammals (Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). A mutation in one of the orthologues, Lgd2 (also named Aki, Freud-1, TAPE and Ccd2d1a), has been shown to cause autosomal recessive mental retardation in humans (Basel-Vanagaite et al., 2006). Recent experiments in mammalian cell culture and in vitro have confirmed the physical interaction between Lgd and the Shrub orthologues, CHMP4a, b and c and suggest a role during budding of HIV (Usami et al., 2012).
Here we further investigated the consequences of loss of lgd function and the requirements for the activation of Notch. We show that activation of Notch appears to be a general consequence of loss of lgd function. We report the electron microscopy (EM) analysis of lgd cells, which revealed that lgd cells contain enlarged MVBs. We found that the lgd cells have a defect that delays degradation of transmembrane proteins, which are residents of the plasma membrane. We show that the activation of Notch in lgd cells is independent of Kuz and dependent on γ-secretase. Furthermore, our results show that the activation occurs in the lysosome. The presented results further elucidate the mechanism of Notch activation in lgd cells.
We previously found that maturing endosomes appear to be enlarged in lgd cells (Childress et al., 2006; Gallagher and Knoblich, 2006; Jaekel and Klein, 2006; Troost et al., 2012). We here have extended our analysis using Rab5–CFP and Rab7–YFP driven by the tubulin promoter (figure 1, Marois et al., 2006). We found that most of the enlarged endosomes are positive for Rab7–YFP (Fig. 1A–E). In contrast only few of the enlarged endosomes were positive for Rab5–CFP (Fig. 1A,B,F,G). This is in contrast to wild-type cells where Rab5 and Rab7 overlap extensively and thus, label most of the Notch positive endosomes (Fig. 1H–M). The analysis confirms that the enlarged endosomes in lgd cells are MEs. In contrast to UAS versions of Rab5 and Rab7, the tubulin promoter driven constructs do not lead to a rescue of the lgd mutant phenotype (Fig. 1N,O, Jaekel and Klein, 2006). Loss of lgd appears not to affect autophagy and does not prevent differentiation of the cells of the eye imaginal disc (for further information see supplementary material Fig. S1).
EM analysis of lgd
Work so far is based on analysis with the fluorescence microscope and suggests that loss of lgd function results in a moderate enlargement of endosomes. However, the markers used for the analysis are restricted to distinct domains of the LM of the endosome and do not outline its whole circumference. It is therefore possible that marker domains are expanded or the markers are re-distributed throughout the limiting membrane in lgd cells (Fig. 2A). In this case, one would get the impression that the endosome is enlarged if monitored with the fluorescent microscope, although its morphology has not changed (see Fig. 2A). In order to unambiguously determine whether endosomes are enlarged in lgd cells, we analysed lgd mutant cells with the transmission electron microscope (TEM). To be able to compare mutant with wild-type cells in one disc, we generated genetic mosaic discs in which the cells of the posterior compartment are mutant for lgd. We depleted the posterior compartment of lgd function by expressing a UAS lgd-RNAi construct in the lgdd7 heterozygous background with hhGal4. We co-expressed UAS Dcr2, which has been shown to enhance the efficiency of the knock down (Dietzl et al., 2007). In this way, we produced a mosaic disc whose cells in the posterior compartment displayed a phenotype comparable to that caused by the complete loss of lgd function (Fig. 2B–E). The design allowed the unambiguous identification of mutant (posterior) and wild-type (anterior) cells in a single disc by their location in cross-sections (Fig. 2F). In the semi-sections we observed that the cells of the peripodial membrane in the mutant part have adopted a columnar shape and formed a continuation of the lateral part of the disc eptithelium (Fig. 2F, arrowheads). Thus, loss of lgd appears to cause a transformation of the peripodial into disc cells.
The TEM analysis indicated that mutant lgd cells contained enlarged MEs (Fig. 2G,H). For a more detailed analysis, we allocated them to several size classes (Fig. 2I,J). This analysis confirmed that lgd cells contained a class of endosomes that are larger than in normal cells (23% of all MVBs; Fig. 2J). Thus, loss of lgd function results in a morphological defect of MEs. Importantly, the MEs contained ILVs, indicating that the ESCRT complexes are active despite the loss of lgd function. However, we cannot rule out that the number of ILVs in MEs is slightly reduced. The presence of normal sized MVBs in lgd cells (Fig. 2J) suggests that the initial maturation of endosomes is normal in lgd cells, but at a fraction continues to enlarge beyond the normal size. This phenotype is expected if the turnover of ME through fusion with lysosomes is disturbed and suggests that it is also the case in lgd cells. We observed normal sized lysosomes in lgd cells (Fig. 2G,H, arrows), indicating that lysosomal biogenesis is probably not disturbed.
Degradation of transmembrane proteins from the plasma membrane is defective in lgd cells
Next we tested whether loss of lgd function results in a degradation defect of Notch and other transmembrane proteins in imaginal disc cells (Fig. 3A–N). In our assay, we expressed GFP tagged receptors in the posterior half with hhGal4 and induced lgd mutant clones in these discs. In an initial control experiment, we expressed GPI-linked GFP (GPI–GFP) and found that: (1) The expression of GPI–GFP has no effect on Notch trafficking in wild-type cells, judged by the size and number of the Notch-positive endosomes in posterior GPI–GFP expressing compared to the anterior non-expressing cells; (2) The concentration of GPI-GFP in lgd mutant and wild-type cells is similar (Fig. 3A–E). This indicates that loss of lgd function does not affect the activity of hhGal4. In contrast, the product of a construct encoding the intracellular domain of Notch tagged with GFP and inserted into the membrane by a CD8 fragment [UAS CEN (Sanders et al., 2009)] is present in much higher concentration in mutant cells (Fig. 3F–H). We obtained similar results with two other receptors, Tkv–GFP or T48–GFP (Kölsch et al., 2007, Fig. 3I–M). These results indicate that lgd mutant cells have a defect in degradation of these trans-membrane proteins. We did not find any defect in degradation of LAMP1–GFP, a transmembrane protein that traffics from the Golgi apparatus to maturing endosomes, nor for aPKC, which is a peripheral protein of the inner surface of the plasma membrane (not shown). Thus, it appears that the degradation defect is specific for transmembrane proteins of the plasma membrane.
During our experiments with CEN we noticed that endogenous Notch accumulates in the lgd mutant cells where the receptors are expressed (Fig. 3H,K,N, arrows). This observation suggests that CEN and endogenous Notch compete in a process, which has limited capacities in lgd cells. In order to further investigate the consequences of the competition, we assed the effect of expression of UAS tkv-GFP, UAS CEN and UAS T48-GFP on ectopic Notch activation in lgdd7 wing imaginal discs. Expression of all these constructs resulted in a suppression of the ectopic activation of Notch (Fig. 3O–Q). In contrast, expression of UAS Notch enhanced the activation as expected (Fig. 3R), while expression of UAS LAMP1–GFP had no effect (Fig. 3S). These findings suggest that transmembrane proteins destined for degradation are in competition for a process that is required for Notch activation in lgd cells. This process has a limited capacity. In contrast to full-length Notch, CEN cannot be cleaved, since it contains the extracellular and transmembrane domains of CD8 and therefore lacks the Furin cleavage site (Sanders et al., 2009). Thus, it does not activate the Notch pathway. This is true also for the other trans-membrane proteins tested. Thus, if these proteins are expressed in high amounts they out-compete endogenous Notch and thereby repress the ectopic activation of the Notch pathway in lgd cells.
Activation of the Notch pathway in lgd mutant cells is independent of kuzbanian
We among others have previously shown that the activation of Notch in lgd cells is independent of its ligands Dl and Ser, but dependent on the activity of the γ-secretase and ESCRT-0 complexes (Childress et al., 2006; Gallagher and Knoblich, 2006; Jaekel and Klein, 2006) (Fig. 4A–G). In order to further investigate the requirements for activation of Notch in lgd mutant cells, we tested whether the activation is independent of Kuz, which mediates the ligand-induced S2 cleavage. For this purpose we monitored the activity of the Notch pathway in kuz lgd double-mutant cell clones. We found ectopic activation of Notch in these clones, indicating that the activation of Notch in lgd cells is independent of the activity of kuz (Fig. 4H–J). Thus, it appears that the S2 cleavage is not required for activation of Notch in lgd cells.
It has been shown that strong overexpression of lgd with the Gal4 system also results in activation of Notch (Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). We here found that this activation is also independent of kuz activity, while it is dependent on activity of hrs (supplementary material Fig. S2). These results suggest that strong overexpression of lgd disturbs the Lgd dependent process in a similar manner than loss of its function.
Previous work has shown that the activation of Notch in lgd cells is dependent on the activity of Psn (Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). This result is the basis for the conclusion that γ-secretase activity is required for Notch activation. However, it is possible that the loss of one of the described γ-secretase independent functions of Psn is responsible for the suppression of the activation of Notch (reviewed in Parks and Curtis, 2007). To exclude this possibility, we monitored the effect of loss of an additional γ-secretase component, aph-1, on Notch activation in lgd cells. We found that loss of aph-1 function prevents the activation of Notch in lgd cells, just as it has been observed previously with loss of Psn function (Fig. 4K–M). Thus, it is the function of the γ-secretase that is required for the activation of Notch in lgd cells. Interestingly, the loss of one already copy of aph-1 (lgd aph-1/lgd) reduces the ectopic expression of Wg in lgd mutant discs (Fig. 4N,O). This indicates that one limiting factor for the activation of Notch in lgd cells is the amount of γ-secretase.
Activity of the proton pump vATPase is required for Notch activation in lgd cells
In order to test whether vATPase is also required for the ectopic activation of Notch in lgd cells, we expressed an RNAi-construct of the VhaA subunit of the vATPase in lgd mutant discs and clones. Indeed, we find that depletion of vATPase function prevents the activation of Notch in lgd cells (Fig. 5; supplementary material Fig. S3). Note, that the activation of Notch at the dorsal–ventral (D/V) boundary, which is induced by the ligands, is not affected in our experiments (arrow in Fig. 5A; supplementary material Fig. S3E,F). This observation suggests that expression of the RNAi construct results in a weak depletion of the vATPase function that is not sufficient to prevent ligand-induced activation of Notch. The result suggests that acidification is also required for the atypical activation of Notch in lgd cells and that this activation is more sensitive towards loss of acidification than the one induced by the ligands.
Furin cleavage of Notch appears to be required for activation in lgd cells
The loss of lgd function results in the activation of Notch in all imaginal disc cells in a ligand- and kuz-independent manner. We therefore wondered which domains of Notch are required for its activation in this situation. We expressed various Notch deletion constructs in lgd mutant wing imaginal discs and monitored its effects on Notch signalling (Fig. 5). Expression of full-length Notch in lgd cells results in a strong enhancement of the activation of the pathway (Fig. 3R). Likewise, NΔEGF, a variant that lacks the EGF repeats and therefore most of the extra-cellular domain, but contain the Lin12 repeats (Loewer et al., 2004) also enhanced the activation of Notch in lgd cells, indicating that it is activated (Fig. 5B). Note, that expression of this variant in wild-type discs results in a slight suppression of Wg expression along the D/V boundary (Fig. 5C, arrow), indicating that it suppresses ligand-dependent activation of Notch. Thus, NΔEGF is a suppressor of ligand-dependent activation of Notch and is not activated in wild-type cells, but appears to become activated in a ligand-independent manner in lgd cells.
Expression of a variant that lacks the furine cleavage site, NBC (Lieber et al., 2002), results in a suppression of activation of Notch in wild-type and lgd cells, indicating that it cannot be activated in both cell types and suppresses the ligand-dependent activation of endogenous Notch (Fig. 5D). N-BC has been shown to reach the plasma membrane and to be active during embryogenesis (Kidd and Lieber, 2002). However, we here find that it weakly suppresses the activation of Notch at the D/V boundary of the wing imaginal disc if expressed during wing development (Fig. 5E, arrow). This is more in agreement with a recent paper, which used a similar construct than N-BC and concluded that the Furine cleavage site in Notch is required for its function (Lake et al., 2009).
In a complementary experiment we tested a variant of Drosophila Notch where the identified Furine cleavage sites of mammalian Notch1 were introduced (Lieber et al., 2002). This variant strongly activated the Notch pathway in lgd cells in a comparable manner than Notch, even in the absence of the function of kuz (supplementary material Fig. S4). Altogether, the results suggest that S1 cleavage of Notch is a prerequisite for its ligand-independent activation in lgd cells. This conclusion is further supported by our finding that CEN, which consists of the NICD fused to the trans-membrane and extra-cellular domains of CD8 and therefore lack a Furin cleavage site, cannot activate the Notch pathway (see above).
Activation of Notch in lgd cells requires fusion with the lysosome
We have recently found that reducing the activity of shrub by 50% (genotype: lgdd7 shrub/lgdd7 +), suppresses the activation of Notch in lgd cells (Troost et al., 2012). This antagonism is surprising because loss of function of each gene alone results in activation of Notch (Klein, 2003; Sweeney et al., 2006; Hori et al., 2011). We therefore wondered what might be the reason for this suppression. We have recently reported that endosomes in lgdd7 shrub/lgdd7 + cells are strongly enlarged and contain the marker Spinster and Hrs, indicating that they are MEs. Nevertheless, we here found that the association of the mutant endosomes with Rab7 is dramatically reduced in comparison to lgd cells (compare Fig. 6A–E with Fig. 1A–E). Thus, reduction of shrub activity in lgd cells results in a loss of association of Rab7 with MEs.
We wondered whether it is the loss of Rab7 association that causes the suppression of Notch activation in lgdd7 shrub/lgdd7 + cells. Therefore, we depleted lgd cells of Rab7 function by expression of a UAS rab7-RNAi construct, which efficiently depletes Rab7 in cells (Fig. 6F–H). Indeed, we found that depletion of Rab7 specifically suppressed the ectopic activation in lgd mutant wing imaginal discs (Fig. 6I,J), while the ligand-dependent activation along the D/V boundary was unaffected (Fig. 6I,J, arrow). Altogether, these results indicate that the loss of rab7 function prevents the ectopic activation of Notch in lgd cells. It further suggests that it is probably the absence of Rab7 on the late endosomes that prevents activation of Notch in lgdd7 shrub/lgdd7 + cells.
The main function of Rab7 is to organise the fusion of MEs with the lysosome. In order to gain further evidence that fusion with the lysosome is required for activation of Notch in lgd cells, we made use of the car1 mutant, which is a hypomorphic allele of carnation (car). It encodes the Drosophila orthologue of Vps33A, a member of the HOPS tethering complex (Akbar et al., 2009). We found that homozygousity of car1 caused a strong suppression of ectopic Notch activation in lgd mutant wing imaginal discs (Fig. 6K,L), while the ligand-dependent activation was not affected (Fig. 6L, arrow). To further corroborate the notion that fusion with the lysosome is required, we depleted lgd cells of the function of vps39, which encodes another component of the HOPS complex (Epp et al., 2011). As expected, expression of UAS vps39-RNAi by hhGal4 specifically suppressed the ectopic ligand-independent activation of Notch in lgd cells (Fig. 6M, arrow). The results strongly suggest that Rab7 controlled fusion of the ME with the lysosome is a prerequisite for activation of Notch in lgd cells. Thus, it appears that the activation of Notch in lgd cells occurs in the lysosome.
Loss of lgd function in the follicle epithelium results in activation of the Notch pathway
So far all experiments that addressed the function of lgd were performed in imaginal disc cells. In order to investigate whether loss of lgd function has a similar effect in cells of other tissues, we investigated its function in the follicle epithelium of the oocyte. During oogenesis, Notch signalling plays a role in several developmental decisions that are required for the development of the oocyte (Horne-Badovinac and Bilder, 2005). One of the best-characterised events is the switch from the mitotic cycle to the endocycle in follicle cells that occurs between stage 6 and 7 (M/E switch) (Horne-Badovinac and Bilder, 2005). This switch is mediated by a Dl signal from the germline cells, which initiates the differentiation of the follicle epithelium through activation of expression of Hindsight (Hnt) and suppression of Cut. A consequence of this switch is that the cells increase in size and become polyploid. We found that follicle cells mutant for lgd start to express Hnt precociously at stage 5 (Fig. 7A–D). Moreover, expression of Cut was suppressed in lgd cells (Fig. 7I–L). This indicates a precocious activation of Notch signalling in these cells. Precocious expression of Hnt in lgd follicle cells of egg chambers was not observed before stage 5. Using the Notch sensor construct Gbe+Su(H), we found that mutant cells at earlier stages readily initiate expression of Gbe+Su(H) (Fig. 7E–H). The expression of Gbe+Su(H) was seen even in the earliest stages of follicle cell development (Fig. 7F–H, arrowheads), indicating that activation of Notch is an immediate consequence of loss of function of lgd function. The activation of Notch in early stages of oogenesis resulted in morphological defects typical for Notch activation. First, we observed that the nuclei of several lgd cells are increased indicating that they have precociously gone through the M/E switch (e.g. see Fig. 7L, arrow). Second, we observed an elongation of the stalk that connects the egg chambers (Fig. 7E,M,N, arrow). This phenotype is also observed upon ectopic activation of Notch, e.g. through expression of NICD (Larkin et al., 1996). Together, these results show that the activation of Notch is a consequence of loss of lgd function in several tissues and is therefore a general property of lgd cells.
The loss of lgd in follicle cells also resulted in an endosomal defect, which manifested itself in the enlargement of endosomes, which contain Notch and are positive for Rab5 and Rab7 (Fig. 8). We used a gene trap that generates a YFP–Notch fusion protein that is functional and contain the YFP in the extra-cellular domain (Rees et al., 2011). The combination with an antibody directed against NICD revealed that the enlarged endosomes are positive for NECD and NICD and therefore probably contain the full-length receptor and also Dl (Fig. 8A–G). Note, however, that while all mutant cells activated the Notch pathway, not all of them had enlarged Notch-positive endosomes. This suggests activation of Notch is independent of the morphological changes and the accumulation of the Notch receptor in the endosomes. It appears that enlargement of endosomes and Notch activation are two independent consequences of the loss of lgd function. The activation of Notch in lgd mutant follicle cells depends on aph-1 function, but is independent of the function of kuz (supplementary material Fig. S5A–G). Like in imaginal disc cells, the activation of Notch is suppressed upon removal of hrs function (supplementary material Fig. S5H–J). Moreover, the activation is dependent on the function of rab7 (Fig. 7O–R). Thus, fusion of the ME with the lysosome is probably required. These results suggest that the activation of Notch in lgd mutant follicle cells occurs in a similar manner than in imaginal disc cells and is probably ligand-independent.
Ectopic activation of Notch upon loss of shrub function occurs differently to that upon loss of lgd function
Next we asked whether the activation of Notch upon loss of ESCRT components is also dependent on Rab7. shrub cells in the wing imaginal disc survive poorly. In contrast, follicle cells are more resistant to loss of shrub function and can be analysed (Vaccari et al., 2009). We first confirm that loss of function of shrub results in the activation of Notch (Fig. 8K–M). We found that this activation occurs in much younger stages of egg chamber development than previously reported and results in the activation of Gbe+Su(H) and Hnt (Fig. 8K–M). Importantly, in contrast to loss of lgd, the activation of Notch is not affected by depletion of Rab7 (Fig. 8N–P). This finding suggests that the activation of Notch upon loss of function of shrub and probably also other components of the ESCRT I–III complexes occurs through another mechanism than upon loss of lgd function, since it does not require the fusion with the lysosome.
Notch signalling is involved in many homeostatic and developmental processes in all metazoans and uncontrolled activation is a cause of disease in humans (Radtke et al., 2005). Hence, it is important to unravel the mechanisms of its normal as well as its uncontrolled activation. Previous work has shown that loss of lgd function results in the ligand-independent activation of the Notch signalling pathway in imaginal disc cells. This activation was still dependent on the activity of the function of Psn, which encodes a component of the γ-secretase complex. In this work we have extended our characterisation of lgd and define the condition under which Notch is activated in lgd cells more precisely. Moreover, we showed that activation of Notch is a consequence of loss of lgd function also in another tissue, the follicle epithelium. This suggests that it is a general consequence of loss of lgd function in cells in Drosophila. We confirmed that the γ-secretase complex is necessary for the activation of Notch in lgd cells. In addition, we presented the first EM analysis of lgd mutant cells, which indicates that a fraction of the endosomes of lgd cells is indeed enlarged and contain ILVs.
Prerequisites of Notch activation in lgd cells
We have previously suggested a model of how Notch is activated in lgd cells (Troost et al., 2012). Several of the uncertainties of this model are resolved by the results of this work. We here showed that activation of Notch requires the fusion of the endosome with the lysosome. Thus, activation probably occurs in the lysosome and not the ME. This finding has several implications. First, it indicates that the defect occurs during endosome maturation in lgd cells, since fusion of the ME with the lysosome does not result in activation of Notch in wild-type cells. Secondly, it indicates that although lgd cells are defective in degradation of trans-membrane proteins, the MEs eventually fuse with lysosomes. Thus, degradation is delayed rather than prevented. This delay in fusion can explain the observation that lgd cells contain a fraction of moderately enlarged MVBs, because it allows the MEs to undergo more homotypic fusions and, thus, to grow to a larger size than normal over time. Thirdly, it indicates that the observed accumulation of Notch in MEs to high levels is not per se the cause of activation of the pathway. This notion is also supported by our observation that although all lgd cells of the follicular epithelium activate the Notch-pathway, not all show strong accumulation of Notch in MEs.
A prerequisite for all Notch activation is that the NICD must face the cytosol so that it can access the nucleus after its release. During normal degradation, Notch is incorporated into ILVs and the NICD is separated from the cytosol. Thus, it cannot access the nucleus even if cleavage would occur. This consideration implies that in lgd cells the formation of ILVs must either fail or Notch is inefficiently incorporated in them. Our EM analysis revealed that the enlarged MEs contained ILVs. Thus, it appears that ILV formation is at least not strongly affected by loss of lgd. We have recently shown that Lgd physically interacts with Shrub and is required for its full function (figure 9A, Troost et al., 2012). A similar interaction has been reported between the mammalian orthologues of Lgd interact and the orthologues of Shrub, CHMP4a, b, c (Martinelli et al., 2012; Usami et al., 2012). Moreover, in vitro experiments suggest that Lgd1 and Lgd2 can influence the polymerisation of CHMP proteins (Martinelli et al., 2012). We therefore favour the possibility that upon loss of lgd function, Notch is inefficiently incorporated into the ILVs due to a reduction, but not abolishment in the activity of shrub. Consequently, a fraction or all of Notch is not incorporated into ILVs and remains in the limiting membrane of the ME (NLM fraction). This NLM fraction is then activated in a ligand-independent manner. However, generating a NLM fraction is a prerequisite for activation of Notch in lgd cells, but it is not sufficient. Loss of hrs function prevents activation of Notch in lgd cells (Gallagher and Knoblich, 2006; Jaekel and Klein, 2006). In hrs mutants, the formation of ILVs is suppressed (Lloyd et al., 2002) and, as a result, cargo (Notch) remains in the LM.
Our results provide more information about the further requirements of Notch activation in lgd cells. We show that activation of Notch is not only independent of the ligands, but also of kuz. Kuz is required for ecto-domain shedding, which is a prerequisite for RIP by the γ-secretase complex. Since the activity of the γ-secretase complex is necessary for activation of Notch in lgd cells (Jaekel and Klein, 2006), ecto-domain shedding of Notch must occur by an alternative manner in lgd cells. Alternative ecto-domain shedding of the NLM fraction in the lysosome can be easily explained by the degradation of the NECD, which extends into the lumen by the activated acidic hydrolases. In this way a NEXT-like fragment could be generated that serves as a substrate for the γ-secretase complex. It has been shown that the γ-secretase complex is active in the lysosome and it is likely that it can perform RIP on the NEXT-like fragment to release NICD into the cytosol (Parks and Curtis, 2007). In agreement with this scenario is our finding that NΔEGF is activated in lgd, but not in wild-type cells. This suggests that the remaining lin12 repeats of NΔEGF that are protruding into the lumen change their conformation or become degraded in lgd cells. We have found that activation of Notch in lgd cells requires the function of the activity of vATPase. This proton pump is required for the acidification of the lumen of MEs, which in turn is a prerequisite for the activation of the acidic hydrolases in the lysosome. The function of vATPase during activation of the Notch pathway in lgd cells could be twofold: Its activity is the prerequisite for the activation of hydrolases. Additionally, the resulting acidification of the lumen might also denature NECD or even resolve the Ca2+ salt bridges between NICD and NECD. It is worth pointing out that the function of vATPase during Notch activation in lgd cells is different from that during ligand-dependent activation. This activation appears to occur in the EE, whose luminal pH is significantly higher (Huotari and Helenius, 2011). However, our data suggest that the S1 cleavage is required in addition. Thus, it is probably an interplay of several factors that causes the activation of Notch.
Our results refine our previously suggested model of Notch activation in lgd cells (see Fig. 9A,B). It suggests that the loss of lgd function results in a reduction in the activity of Shrub. As a consequence a fraction or all of Notch remains at the limiting membrane of the ME. Upon fusion with the lysosome the activated hydrolases, possibly in combination of the acidic environment causes alternative ecto-domain shedding. This creates a NEXT-like substrate for the γ-secretase complex, which releases NICD into the cytosol.
The model is very similar to that suggested for the regulation of activation of Notch by the E3-ligase Deltex (Wilkin et al., 2008). Recent work reports an intimate functional relationship between the arrestin-like protein Kurtz (Krz), Deltex (Dx) and Shrub upon regulation of Notch (Hori et al., 2011). The authors suggested a model in which Shrub downregulates the activity of Notch by incorporating the poly-ubiquitinated receptor into ILVs, while Dx antagonises this incorporation and promotes activation. Dx exerts its function by regulating the ubiquitination status of NICD. Its overexpression shifts the equilibrium from poly- to mono-ubiquitylation. It will be interesting to determine how Lgd fits into this scheme. Qualitatively, the phenotype of loss of lgd function is similar to that of Dx overexpression, although the level of Notch activation is stronger. This would suggest that Lgd antagonises the function of Dx. The model proposed by Hori et al. (Hori et al., 2011) suggests that Dx prevents the incorporation of Notch into ILVs by Shrub. Lgd appears to influence the activity of Shrub by direct physical interaction (Troost et al., 2012). Thus, we favour the possibility that the relationship between Dx and Lgd is indirect and mediated through Shrub.
Our experiments indicate that a process is required for activation of Notch in lgd cells that has limited capacity. A possibility is that the affected process is the cleavage of Notch the γ-secretase complex, as we have observed that the ectopic activation of Notch in lgd cells is suppressed through reduction of the activity of the complex. If this is true it has to be explained how other transmembrane proteins that can compete with Notch are transformed into substrates for the γ-secretase complex. A possibility is that they also undergo alternative ecto-domain shedding.
Uncontrolled activation of the Notch pathway can occur in different endosomal compartments
We here confirm that loss of shrub function results in the activation of the Notch pathway also during oogenesis (Vaccari et al., 2009). We here also provide further support for our previously drawn conclusion that the activation of Notch in shrub cells occurs through another mechanism than in the case of lgd, because we found that the activation of Notch is suppressed if Rab7 is depleted in lgd but not shrub cells (Troost et al., 2012). This finding indicates that the activation in shrub cells occurs in the ME rather than in the lysosome. Thus, activation of Notch can occur in several endosomal compartments.
It is possible that different levels of inactivation of shrub trigger different modes of Notch activation. The loss of shrub function results in a strong activation of Notch and loss of epithelial integrity (Hori et al., 2011; Troost et al., 2012). However, slight reduction of shrub function only weakly activates Notch and appears to have little effect on the epithelial integrity (Hori et al., 2011). This raises the possibility that activation of Notch through reduction of shrub activity might occur through a different mechanism than upon abolishment of shrub function. It would be interesting to investigate whether the slight activation of Notch occurs in the lysosome or endosome.
Materials and Methods
The fly stocks used were as follows: UAS T48-GFP (gift of M. Leptin) (Kölsch et al., 2007), UAS tkv-GFP (gift of K. Basler), UAS CEN-GFP (Sanders et al., 2009), UAS GFP-GPi (Greco et al., 2001), UAS NBC (Lieber et al., 2002), UAS N-LV, UAS NΔEGF (Loewer et al., 2004), UAS atg8-GFP (Rusten et al., 2007), UAS vhaA-RNAi (VDRC #17102), tubGAL80ts (Bloomington stock #7018), UAS GFP-Lamp1 (gift of H. Krämer), UAS lgd-RNAi (NIG-Fly 4713-1 + 4713-3), UAS vps39-RNAi (VDRC #40427), UAS rab7-RNAi (VDRC #40337), car1 (BL #19), Notch-YFP (CPTI 000433) (Rees et al., 2011). Mutant stocks were as follows: shrub4-1 FRTG13; lgdd7 aph-1D35 (this study), lgdd7 hrsD28 (Jaekel and Klein, 2006), lgdd7 kuzES24 (this study), kuzES24 FRT40A (Klein, 2002), hrsD28 FRT40A (Lloyd et al., 2002) lgdd7 FRT40A (Jaekel and Klein, 2006), lgdd7 shrub4-1 and lgdP-lgdRFP (68E) FRT2A (Troost et al., 2012). Other lines used were: Gbe+Su(H)-lacZ (Furriols and Bray, 2001), tub. rab5-CFP tub. rab7-YFP (Marois et al., 2006), Gbe+Su(H)-nlsGFP (de Navascués et al., 2012). Gal4 lines were: hhGal4 (Tanimoto et al., 2000).
Clones were generated with the FLP/FRT system (Xu and Rubin, 1993) and induced at the first larval instar (24–48 hours after egg laying) by applying a 70-minute heat shock (37°C). Mutant eyes/headcapsules were generated with w; GMR-Hid l(2) *[*] FRT40A; eyGAL4 UASFlp (Bl #5250) at 25°C. Mutant follicle cell clones were generated as follows: Females were heat shocked for 2 hours at 37°C on two consecutive days. Flies were kept at 25°C and ovaries were dissected 2 days after heat shock.
Immunostainings and microscopy
Antibodies used: anti-Wg (4D4) antibody, mouse Notch antibodies against the extra- (C458.2H) and intra-cellular (C17.9C6) domains, anti Cut (2B10), anti-Hnt (1G9). These antibodies were obtained from the Developmental Studies Hybridoma Bank. Further antibodies used in this work: anti Rab7 (Tanaka and Nakamura, 2008), anti-β-Gal (rabbit, Cappel). Fluoro-chrome conjugated secondary antibodies were purchased from Molecular Probes/Invitrogen. Images were obtained with a Zeiss Apotome Microscope.
Wing discs were fixed in 2.5% glutaraldehyde in 100 mM phosphate buffer washed in 100 mM phosphate buffer and postfixed in 2% osminum tetroxide in phosphate buffer for 1 hour on ice. After contrasting en bloc in 2% uranyl acetate, the specimens were dehydrated in EtOH and embedded in araldite using acetone as an intermediate solvent. Thin sections were stained with 2% uranyl acetate and lead citrate. Sections were observed under an EM 902 (Zeiss) microscope at 80 KV.
We thank Sylvia Tannebaum and Gisela Helbig for excellent technical support. We thank M. Leptin, K. Basler, G. Merdes, S. Bray, J. Knoblich, H. Krämer, S. Eaton, G. Davies, H. Bellen, F. B. Gao, M. Gonzalez-Gaitan, T. Schüpbach and A. Nakamura for supplying antibodies and fly stocks used in this work. We are especially grateful to A. Bachmann for critical reading and comments on the manuscript. We thank B. Schnute, J. Steinbring, Kim Morawa and C. Benscheid for help during some of the experiments. The Bloomington Stock Center, VDRC and the Developmental Studies Hybridoma Bank supplied fly stocks and reagents.
The work was funded by the Deutsche Forschungsgemeinschaft (DFG) through SFB 590.