Notch ligand activity is modulated by glycosphingolipid membrane composition in Drosophila melanogaster

Abstract

Endocytosis of the transmembrane ligands Delta (Dl) and Serrate (Ser) is required for the proper activation of Notch receptors. The E3 ubiquitin ligases Mindbomb1 (Mib1) and Neuralized (Neur) regulate the ubiquitination of Dl and Ser and thereby promote both ligand endocytosis and Notch receptor activation. In this study, we identify the alpha1,4-N-acetylgalactosaminyltransferase-1 (alpha4GT1) gene as a gain of function suppressor of Mib1 inhibition. Expression of alpha4GT1 suppressed the signaling and endocytosis defects of Dl and Ser resulting from the inhibition of mib1 and/or neur activity. Genetic and biochemical evidence indicate that alpha4GT1 plays a regulatory but nonessential function in Notch signaling via the synthesis of a specific glycosphingolipid (GSL), N5, produced by alpha4GT1. Furthermore, we show that the extracellular domain of Ser interacts with GSLs in vitro via a conserved GSL-binding motif, raising the possibility that direct GSL-protein interactions modulate the endocytosis of Notch ligands. Together, our data indicate that specific GSLs modulate the signaling activity of Notch ligands.

Figure 1.

Suppression of mib1 by GS2078. (A–F’) Genetic interactions between mib1 and GS2078 were studied in third instar wing imaginal discs (A, B, C, D, E, and F) and in adult wings (A’, B’, C’, D’, E’, and F’). (A) The pattern of Ser-Gal4 expression was visualized using nlsGFP (green), and wing margin cells were identified using Cut (red). (A and A’) Wild-type control UAS-nlsGFP/+; Ser-Gal4 tub-Gal80^ts/+ flies (Ser > nlsGFP) expressed nuclear GFP under the control of Ser driver in dorsal (d) cells as well as in some ventral (v) cells. Notch activation along the dorso–ventral boundary results in Cut expression at the wing margin. (B and B’) Trans-heterozygous mib1²/mib1³ mutant disc and wing. Defective wing margin formation and wing pouch growth (B) result in a strong wing loss phenotype (B’). (C and C’) Expression of Mib1^C1205S in UAS-nlsGFP/UAS-mib1^C1205S; Ser-Gal4 tub-Gal80^ts/+ flies (Ser > mib1^C1205S + nlsGFP) led to defective wing margin specification and reduced growth of the pouch. (D and D’) GS2078 suppressed the Mib1^C1205S-induced phenotype in UAS-mib1^C1205S/+; Ser-Gal4 tub-Gal80^ts/GS2078 flies (Ser > mib1^C1205S + GS2078). Cut expression at the wing margin and wing pouch growth were significantly rescued. (E and E’) Down-regulation of mib1 expression in UAS-nlsGFP/+; Ser-Gal4 tub-Gal80^ts UAS-mib1^RNAi/+ flies (Ser > mib1^RNAi + nlsGFP) gave a mib1 partial loss of function phenotype (compare with B and B’). (F and F’) GS2078 suppressed the hypomorphic mib1 phenotype in Ser-Gal4 tub-Gal80^tsUAS-mib1^RNAi GS2078/GS2078 flies (Ser > mib1^RNAi + GS2078). Cut expression at the wing margin and tissue growth were largely restored. Bar, 10 µm.

Figure 2.

α4GT1 is a gain of function suppressor of mib1. (A) Molecular map showing that the GS2078 P-element (blue) is inserted between and upstream of the α4GT1 and CG3542 genes. The breakpoints of the Df(2L)7819 and of the small α4GT1¹ deletions are indicated. Df(2L)7819 deletes 24 kb of genomic DNA located between the 5HA-2924 and CB-5583-3 P-elements (red). It deletes the α4GT1, CG17264, CG17224, and CG17265 genes and also partially deletes the CG3542 gene. The α4GT1¹ allele is a 1,201-nucleotide-long deletion that removes the sequence encoding the first 286 amino acids of the α4GT1 protein. Bar, 1 kb. (B) Molecular map of the α4GT2 locus. The position of the roo 1,422 element disrupting the α4GT2 open reading frame (red; transcript is in blue) in the α4GT2¹ mutant allele is indicated. (C and D) Expression of α4GT1 in UAS-mib1^C1205S/+; Ser-Gal4 tub-Gal80^ts/ UAS-α4GT1 flies (Ser > mib1^RNAi + α4GT1; C) and α4GT2 in UAS-mib1^C1205S/+; Ser-Gal4 tub-Gal80^ts/UAS-α4GT2 flies (Ser > mib1^RNAi + α4GT2; D) suppressed the Mib1^C1205S-induced wing phenotypes. (E and F) RNAi-mediated down-regulation of α4GT1 blocked suppression by GS2078 of the Mib1^C1205S-induced wing phenotype in UAS-mib1^C1205S; Ser-Gal4 tub-Gal80^ts/GS2078; UAS-α4GT1^RNAi/+ (Ser > mib1^C1205S + GS2078 + α4GT1^RNAi; F). Note that the Ser > mib1^C1205S wing phenotype (E) is stronger than the one shown in Fig. 1 C’. This is because the x-linked mib1^C1205S transgene is expressed at higher levels in males (E and F) than in females (Fig. 1 C’). (G) α4GT1¹/Df(2L)7819; α4GT2¹ double-mutant flies have no detectable phenotype (compare with Fig. 1 A’). (H–L) Expression of α4GT1 suppressed the hypomorphic mib1 wing nick phenotype. (H and I) GS2078 lowered the penetrance of the hypomorphic mib1³/mib1⁴ wing nick phenotype: 11% (n = 194) of the mib1³/mib1⁴ wings exhibited nicks (H), whereas only 1% (n = 202) of Ser-Gal4 tub-Gal80^ts/GS2078; mib1³/mib1⁴ wings had nicks (I). In contrast, expression of α4GT1 did not suppress the mib1-null phenotype. (J–L) The wing phenotype of mib1²/mib1³ flies (J) was rescued by the expression of mib1 (L) but was not modified by the expression of α4GT1 in Ser-Gal4 tub-Gal80^ts/GS2078; mib1³/mib1⁴ flies (K). (M and N) Loss of α4GT1 function enhances the severity and penetrance of the haploinsufficient wing Notch phenotype: 19% (n = 16; 25°C) and 15% (n = 20; 29°C) of the N^55e11 heterozygous flies show a small wing nick, whereas 66% (n = 132; 25°C) and 100% (n = 26; 29°C) exhibit nicks of increased size in the complete absence of α4GT1 activity. Double-heterozygous Notch α4GT1 flies were similar to N^55e11 heterozygous flies in severity and penetrance. N^55e11/+; α4GT1¹/+: 32% (n = 125) and N^55e11/+; Df(2L)7819/+: 28% (n = 96) at 25°C.

Figure 3.

α4GT1 rescued mib1^C1205S defects in Dl and Ser distribution. Ser (green), Dl (red), YFP/mib1^C1205S (green), and Dlg (blue) distribution were analyzed in wing imaginal discs. (A–I’’’) The following genotypes were studied: wild-type (WT; A–B’’ and G–G’’’), UAS-nlsGFP/UAS-mib1^C1205S; Ser-Gal4 tub-Gal80^ts/+ (Ser > mib1^C1205S + nlsGFP; C–D’’), UAS-nlsGFP/UAS-YFPmib1^C1205S; Ser-Gal4 tub-Gal80^ts/+ (Ser > YFPmib1^C1205S + nlsGFP; H–H’’’), UAS-mib1^C1205S/+; Ser-Gal4 tub-Gal80^ts/GS2078 (Ser > mib1^C1205S + α4GT1; E–F’’), and UAS-YFPmib1^C1205S/+; Ser-Gal4 tub-Gal80^ts/GS2078 (Ser > YFPmib1^C1205S + α4GT1; I–I’’’). Because Dl and Ser have distinct expression patterns (A), we focused our analysis to a dorsal region located near the margin where cells coexpress Dl and Ser. (B–F’’) High magnification views of the areas boxed in A–E are shown. (G–I’’’) Z-section views are shown. (A–B’’ and G–G’’’) In wild-type cells, Dl (B’ and G’) and Ser (B’’) colocalized at the apical cortex, apical to Dlg (G’’’). (C–D’’ and H–H’’’) Expression of Mib1^C1205S (C–D’’) or YFPmib1^C1205S (H–H’’’) led to the accumulation of Dl (D’, H’) and Ser (D’’) into dots at the apical cortex, apical to Dlg (H’’’). (E–F’’ and I–I’’’) Coexpression of α4GT1 with Mib1^C1205S (E–F’’) or YFPmib1^C1205S (I–I’’’) did not significantly change the distribution of Dl (F’ and I’) and Ser (F’’) compared with wild-type controls. Bars, 10 µm.

Figure 4.

α4GT1 rescued loss of mib1 defects in Ser accumulation. (A–F) The accumulation of Ser at the apical cortex of the cells and wing margin formation (using Cut as a marker) were examined in wild-type (WT; A–B), UAS-nlsGFP/+; Ser-Gal4 tub-Gal80^tsUAS-mib1^RNAi/+ (Ser > mib1^RNAi + nlsGFP; C–D), and Ser-Gal4 tub-Gal80^tsUAS-mib1^RNAi/GS2078 (Ser > mib1^RNAi + α4GT1; E–F) wing imaginal cells. High magnification views of the areas boxed in A’, C’, and E’ are shown in B, D, and F, respectively. (A’, B, C’, D, E’, and F) Single apical sections of stacks acquired using parameters adjusted to the high intensity signals measured in D are shown. These settings account for the low Ser signal in B. (A–B) Wild-type controls are shown. (A) Wing margin cells are specified as revealed by Cut expression along the dorsal–ventral boundary. (B) Low levels of Ser were detected at apical sections (fluorescence signal intensity = 1 ± 0.2 arbitrary units; n = 3). (C–D) RNAi-mediated inactivation of mib1 resulted in increased Ser levels (D; fluorescence signal intensity = 16 ± 2.5; n = 3) and loss of wing margin (C). (E–F) Overexpression of α4GT1 rescued the mib1^RNAi defects in Ser accumulation (F; fluorescence signal intensity = 3.4 ± 0.5; n = 3) and restored wing margin formation and pouch growth (E). Our quantification clearly indicates that the level of apical Ser was increased upon reduction of Mib1 activity in mib1 RNAi cells and that expression of α4GT1 counteracts this effect. Bars, 10 µm.

Figure 5.

α4GT1 rescued the endocytosis of Dl blocked by Mib1^C1205S. The endocytosis of Dl was monitored in wing imaginal discs using an antibody uptake assay in wild-type (A–B’), UAS-nlsGFP/UAS-mib1^C1205S; Ser-Gal4 tub-Gal80^ts/+ (Ser > mib1^C1205S + nlsGFP; C–D’), and UAS-mib1^C1205S/+; Ser-Gal4 tub-Gal80^ts/GS2078 (Ser > mib1^C1205S + α4GT1; E–F’). The approximate position of the dorsal (d)–ventral (v) boundary is indicated in red. (A–B’) In wild-type discs, iDl was detected in subapical sections (B’) in both dorsal and ventral cells that express Dl (Fig. 3 A). (C–D’) Expression of Mib1^C1205S in dorsal cells using Ser-GAL4 inhibited the endocytosis of Dl (D’). In these cells, dots of Dl were only detected in apical sections (D). This staining is very similar to the one seen in cell surface staining experiments (Fig. S3), indicating that Mib1^C1205S inhibits the internalization of Dl. iDl was only seen in the ventral cells (D’) that express Mib1^C1205S later and at a lower level (Fig. 1 A). (E–F’) Expression of α4GT1 restored the endocytosis of Dl. (F’) iDl was detected in both ventral and dorsal cells. (F) Only weak cell surface staining was observed. High magnification views of boxed areas shown in A, C, and E are shown in B and B’, D and D’, and F and F’, respectively. Bar, 10 µm.

Figure 6.

α4GT1 restored the Neur-dependent endocytosis of Dl in SOPs. (A–L’) The endocytosis of Dl (iDl in red) was monitored in SOPs (marked by Senseless [Sens] in green) using an antibody uptake assay in pupae of the following genotypes: wild-type (WT; A–B’); GS2078/+; pnr-GAL4 tub-Gal80^ts/+ (pnr>α4GT1; C–D’); UAS-Tom/+; pnr-GAL4 tub-Gal80^ts/+ (pnr > Tom; E–F’); UAS-Tom/GS2078; pnr-GAL4 tub-Gal80^ts/+ (pnr > Tom + α4GT1; G–H’); UAS-neur^RNAi/+; pnr-GAL4 tub-Gal80^ts (pnr > neur^RNAi; I–J’); and UAS-neur^RNAi/GS2078; pnr-GAL4 tub-Gal80^ts (pnr > neur^RNAi + α4GT; K–L’). (A–D’) Expression of α4GT1 did not detectably affect the endocytosis of Dl in SOPs (D and D’) and did not significantly change bristle density (C). (E–H’) Inhibition of Neur by Tom blocked Dl endocyosis (F and F’) and resulted in a very strong neurogenic phenotype (E) with too many SOPs being specified (F). Expression of α4GT1 restored both Dl endocytosis (H and H’) and proper SOP specification (G and H). (I–J’) RNAi-mediated down-regulation of neur strongly inhibited Dl endocyosis (J and J’) and resulted in a strong neurogenic phenotype (I) with an excess of SOPs (J’). Expression of α4GT1 restored both Dl endocytosis (H and H’) and suppressed the neur RNAi bristle phenotype (G and H). Bar, 10 µm.

Figure 7.

Rescue of mib1 inhibition by α4GT1 depends on egh and brn activities. (A) Molecular structure of GalNAc-α1-4-GalNAc-β1-4-GlcNAcβ-1-3Manβ1-4Glcβ1-1Cer or N5. The enzymes acting sequentially in the N5 biosynthetic pathway are indicated below (see Results). The two dominant suppressors identified in our screen are highlighted in yellow. (B–C’) Analysis of α-GalNAc distribution in pupal notum epithelial cells using HPA (TRITC-HPA in red). HPA staining was strongly reduced in clones of α4GT1 mutant cells (B and B’; mutant cells are marked by nuclear GFP in green), indicating that α4GT1 is required for α-GalNAc localization at the cell surface. Overexpression (o/e) of α4GT1 in clones (CD8-GFP in green) resulted in increased HPA staining, indicating that α4GT1 is a limiting enzyme for addition of α-GalNAc. (D) HPTLC analysis of GSLs purified from wild-type (WT) and mutant larvae. (lane 1) Standard GSLs: CMH, Cer monohexoside (GlcCer); CDH, Cer dihexoside (LacCer); CTH, Cer trihexoside (Gb3); and CPH, Cer pentahexoside (Forsmann glycolipid). The GSL species detected in larvae extracts (N1, N2, N4, and N5) were identified on the basis of their chromatographic mobility as compared with standard GSLs. (lane 2) Wild type. (lane 3) egh^62D18/Y. (lane 4) brn^I.6P6/Y. (lane 5) α4GT1¹/Df(2L)7819; α4GT2¹. (lane 6) Overexpression of α4GT1 in wild-type larvae: tub-GAL4/GS2078. (lane 7) Overexpression of α4GT1 in egh mutant larvae: egh^62D18/Y; tub-GAL4/GS2078. (lane 8) Overexpression of α4GT1 in brn mutant larvae: brn^I.6P6/Y; tub-GAL4/GS2078. (E–M) Rescue of mib1 inhibition by α4GT1 depends on egh and brn activities. Wing margin specification (marked by Cut) was examined in wild-type (E), Ser > mib1^RNAi + nlsGFP (F), Ser > mib1^RNAi + α4GT1 (G), egh^62D18/Y (H), egh^62D18/Y; Ser > mib1^RNAi + nlsGFP (I), egh^62D18/Y; Ser > mib1^RNAi + α4GT1 (J), brn^I.6P6/Y (K), brn^I.6P6/Y; Ser > mib1^RNAi + nlsGFP (L), and brn^I.6P6/Y; Ser > mib1^RNAi + α4GT1 (M) wing discs. Zygotic loss of egh and/or brn did not significantly alter the expression of Cut at the wing margin (E, H, and K) and did not enhance the Ser > mib1^RNAi phenotype (F, I, and L). In the absence of egh activity, α4GT1 expression failed to restore wing margin specification (J) as it did in wild-type discs (G). (M) Suppression by α4GT1 was strongly reduced in brn mutant discs. The partial suppression seen in brn mutants may result from maternally provided brn gene products. Alternatively, egh may have a brn-independent function in this tissue. Bars, 10 µm.

Figure 8.

Identification of a conserved GBM. (A) Schematic representation of the structure of Dl and Ser (adapted from Parks et al., 2006). The domain structure of Ser[1–288] is also indicated. The potential GBM detected in the N2 domain of Dl and Ser appears in yellow. This GBM is conserved in mammals: the sequences of human Dl–like 1 and Jagged1 are aligned with the GBM of Drosophila Dl and Ser. The sequences of the synthetic peptides used in B and C are boxed in yellow. The Trp (W) residue shown in B required for interaction with GSLs is indicated with an asterisk. DSL, Dl/Ser/LAG-2; DOS, Dl- and OSM-11–like proteins; CRD, cysteine-rich domain; TM, transmembrane domain. (B) Analysis of GBM–GSL interactions. Synthetic peptides corresponding to the GBM of Ser and Dl interacted with total GSLs extracted from wild-type larvae. Interactions were quantitatively measured using the Langmuir film balance technique. The conserved Trp residue is essential for these interactions. The following peptides were studied: Ser GBM (open squares; VLPFTFRWTK), Ser GBM^WA (closed squares; VLPFTFRATK), Dl GBM (closed triangles; SFSWPGTFS), and Dl GBM^WA (open triangles; SFSAPGTFS). (C and D) Analysis of Ser–GSL interactions. Interactions of the Ser[1–288] (closed squares in C; blue bars in D) and Ser[1–288]W180A (open triangles in C; green bars in D) proteins secreted from S2 cells with phosphatidylethanolamine (PE) and total GSLs extracted from wild-type, egh, α4GT1 α4GT2 double-mutant, and tub-GAL4 UAS-α4GT1 (overexpressed α4GT1) larvae were quantitatively measured using the Langmuir film balance. The binding kinetics are shown in C, and the values of the maximal surface pressure increases are given in D. (D) Ser[1–288] and Ser[1–288]^WA similarly interacted with PE and neutral lipids prepared from egh mutant larvae. This indicates that the N-terminal part of Ser interacts in a GBM-independent manner with lipid monolayers in this assay. In contrast, Ser[1–288] interacted more strongly than Ser[1–288]^WA with total GSLs extracted from wild-type, α4GT1 α4GT2 double-mutant, and tub-GAL4 UAS-α4GT1 larvae, indicating that the N-terminal part of Ser interacts in a GBM-dependent manner with GLSs. Moreover, a stronger and GBM-dependent interaction correlated with high levels of N5 (and low levels of N4) in overexpressed α4GT1 larvae. Error bars indicate SD (n = 3 experiments).