A targeted in vivo RNAi screen reveals deubiquitinases as new regulators of Notch signaling

Abstract

Notch signaling is highly conserved in all metazoan animals and plays critical roles in cell fate specification, cell proliferation, apoptosis, and stem cell maintenance. Although core components of the Notch signaling cascade have been identified, many gaps in the understanding of the Notch signaling pathway remain to be filled. One form of posttranslational regulation, which is controlled by the ubiquitin-proteasome system, is known to modulate Notch signaling. The ubiquitination pathway is a highly coordinated process in which the ubiquitin moiety is either conjugated to or removed from target proteins by opposing E3 ubiquitin ligases and deubiquitinases (DUBs). Several E3 ubiquitin ligases have been implicated in ubiquitin conjugation to the receptors and the ligands of the Notch signaling cascade. In contrast, little is known about a direct role of DUBs in Notch signaling in vivo. Here, we report an in vivo RNA interference screen in Drosophila melanogaster targeting all 45 DUBs that we annotated in the fly genome. We show that at least four DUBs function specifically in the formation of the fly wing margin and/or the specification of the scutellar sensory organ precursors, two processes that are strictly dependent on the balanced Notch signaling activity. Furthermore, we provide genetic evidence suggesting that these DUBs are necessary to positively modulate Notch signaling activity. Our study reveals a conserved molecular mechanism by which protein deubiquitination process contributes to the complex posttranslational regulation of Notch signaling in vivo.

Keywords: Drosophila melanogaster; Notch signaling; deubiquitinase; ubiquitination.

Figure 1

Inventory and domain architectures of annotated DUBs in Drosophila. Drosophila DUBs are characterized into five subfamilies on the basis of their signature DUB catalytic domains. The UCH, USP, MJD, OUT, and JAMM domain-containing proteases are shown. Apart from signature DUB domains, we retrieved domain architectures for each DUB by using the Pfam and the NCBI Conserved Domain database. The abbreviations for additional domains are listed as follows: Cap-Gly, cytoskeleton-associated proteins, glycine-rich domain; DUSP, domain in ubiquitin-specific proteases; EFh, EF-hand, calcium binding motif; EXOIII, exonuclease, RNase T/DNA polymerase III; MATH, meprin and TRAF-homology; MIT, microtubule interacting and trafficking molecule domain; PRP8, pre-mRNA processing splicing factor 8; RBP, zinc finger, RanBP2-type; RhoD, rhodanese homology domain; RPT, internal repeats; RRM_4, RNA recognition motif of the spliceosomal PrP8; Tudor, Tudor domain; U5 and U6, U5 and U6 snRNA binding domains; UBA, ubiquitin-associated; UBL, ubiquitin-like; UIM, ubiquitin interaction motif; WD40, WD40-repeat-containing domain; and ZnF-UBP, zinc finger ubiquitin binding domain. Proteins and domains are plotted on an approximate scale.

Figure 2

The adult wing margin and scutellar bristle phenotypes are simple but effective readouts for altered Notch signaling. The larval wing imaginal disc is the primordial tissue of the adult wing blade and notum. The expression patterns of the dpp-Gal4 (A) and the C96-Gal4 drivers (B) in wing discs were marked by the UAS-gfp transgene (green). The dpp-Gal4 drives transgene expression abutting the anterior-posterior (a-p) boundary (A), whereas the C96-Gal4 confers gfp expression along the dorsal-ventral (d-v) boundary, i.e., presumptive wing margin in the wing pouch (B). Note that the dpp-Gal4 is also expressed in regions where the adult scutellar structures are derived (box; A). DAPI staining was used to mark nuclei in wing discs (magenta; A and B). As expected, ectopic expression of gfp RNAi by the C96-Gal4 did not produce any effect in the adult wing margin (C). In contrast, reduced expression of Notch receptor gene by RNAi resulted in serrations along the wing margin (D). However, knockdown of the expression of faf, which encodes a DUB that specifically regulates Notch signaling in the developing Drosophila eye, had no effect on patterning of the wing margin (E). Note that wings dissected from female flies are shown in all figures. Two pairs of scutellar bristles, anterior SC and posterior SC (aSC and pSC), are present in the scutellum (F). When the expression of Notch was knocked down by RNAi using the dpp-Gal4, an increased number of scutellar bristles was observed (G). In contrast, overexpressing RNAi targeting either gfp (F) or faf (H) had no effect on the specification of scutellar bristles. Phenotypes shown in panels D (n > 20) and G (n = 10) are fully penetrant.

Figure 3

Reduced expression of several DUBs leads to defects on the formation of the margin vein in the adult fly wing. Reduced expression of CG3416 (A), CG18174 (B), CG8445 (C), CG9124 (D), CG9769 (E), or CG32479 (F) in the wing imaginal disc by respective RNAi transgenes driven by the c96-Gal4 driver led to defective wing margin formation, with different degrees of severity. Note that the wing margin defects in panels A and F are stronger than those caused by Notch RNAi (cf. Figure 2D), suggesting that CG3416, which encodes an essential component of the 19S proteasome, and probably CG32479, play additional roles other than those in Notch signaling. Phenotypes shown in panels A (n = 19), B (n = 9), C (n = 15), D (n = 11), E (n = 6), and F (n = 23) are fully penetrant.

Figure 4

Two candidate DUBs regulate the specification of scutellar bristles in the adult fly notum. Reduced expression of CG8445 (A) or CG32479 (B) in the wing imaginal disc by respective RNAi driven by the dpp-Gal4 driver resulted in an increased number of scutellar bristles, with a penetrance of 18% (n = 11) for CG8445 RNAi and 100% (n = 20) for CG32479 RNAi.

Figure 5

The Effect of CG8445 RNAi on the expression of Notch signaling targets in the wing imaginal disc. The expression of Wg (magenta; A) and Cut (magenta; C) along the dorsal-ventral boundary (i.e., presumptive wing margin) is activated by Notch signaling in a wild-type (WT) wing disc. Knocking down the expression of Notch receptor by the ptc-Gal4 driver along the anterior-posterior boundary (marked by GFP) led to a complete loss of downstream targets, Cut (E, F) and Wg (G, H), on the presumptive wing margin cells intersecting the dorsal-ventral boundary. In contrast, the expression of both Cut (J) and Wg (L) was unaffected in wing discs ectopically expressing faf RNAi, consistent with previous studies indicating that faf mutant flies did not exhibit any defect in the wing. Knockdown of the expression of CG8445 by RNAi in the wing disc resulted in modulate reduction of Cut (arrow; N) and Wg (arrow; P), suggesting that CG8445 is necessary to positively regulate Notch signaling. All phenotypes are fully penetrant (n > 20 discs).

Figure 6

CG9769 positively regulates Notch signaling in the wing disc. When the expression of CG9769 was knocked down by RNAi at the anteroposterior boundary of the wing disc by the dpp-Gal4 driver (marked by GFP; A and C), Notch signaling activity was largely compromised as indicated by decreased expression of Cut (B) and Wg (D). The abundance of Notch protein, as examined by specific antibodies raised against either the extracellular domain (NECD; E and F) or the intracellular domain (NICD; G and H) of the Notch receptor, also was reduced. The effect of CG9769 on the production of Notch protein was cell-autonomous (arrows; J), as demonstrated in FLIPout clones (Ito et al. 1997) overexpressing CG9769 RNAi (positively marked by GFP; I). All phenotypes are fully penetrant (n > 20 discs).

Figure 7

CG32479 positively regulates Notch signaling in the wing imaginal disc. Random clones overexpressing CG32479 RNAi were generated by the FLIPout technique. Knocking down the expression of CG32479 by RNAi in these clones (positively marked by GFP, green) cell-autonomously reduced the expression of Cut (arrows; B) and Wg (arrow; D). Consistently, the production of Notch protein was cell-autonomously downregulated in random clones overexpressing CG32479 RNAi (positively marked by GFP), as demonstrated by immunostaining with antibodies specific for the NECD (arrows; F) and NICD (arrows; H), respectively. All phenotypes are fully penetrant (n > 20 discs).

Figure 8

CG32479 is both necessary and sufficient to control the formation of scutellar bristles. The adult scutellar bristles (A) are differentiated from macrochaete SOP cells in early third-instar wing imaginal discs, which can be marked by the neur-lacZ reporter (magenta; B-F′). Knocking down the expression of CG32479 by the ptc-Gal4 driver (green) led to increased numbers of scutellar bristles in the adult notum (stars; C): 64% of flies with 5−6 scutellar bristles and 36% of flies with more than 6 scutellar bristles (n = 83). This fully penetrant scutellar bristle phenotype was a consequence of expanded SOP fate observed in wing discs along the ptc-Gal4 expression pattern (arrows; D). In contrast, overexpressing CG32479 resulted in loss of anterior scutellar bristles (100% penetrant, n = 25; stars; E) and reduction of SOP cells (arrows; F). Enlarged boxed areas shown in B′, D′, and F′ highlight distinct neur-lacZ staining in the scutellar (SC) and dorsal-central (DC) SOPs resulted from altered expression of CG32479. Note that the expression pattern of the ptc-Gal4 driver (marked by GFP; B-F′) overlaps the SC but not the DC SOPs.