Compensatory flux changes within an endocytic trafficking network maintain thermal robustness of Notch signaling

Abstract

Developmental signaling is remarkably robust to environmental variation, including temperature. For example, in ectothermic animals such as Drosophila, Notch signaling is maintained within functional limits across a wide temperature range. We combine experimental and computational approaches to show that temperature compensation of Notch signaling is achieved by an unexpected variety of endocytic-dependent routes to Notch activation which, when superimposed on ligand-induced activation, act as a robustness module. Thermal compensation arises through an altered balance of fluxes within competing trafficking routes, coupled with temperature-dependent ubiquitination of Notch. This flexible ensemble of trafficking routes supports Notch signaling at low temperature but can be switched to restrain Notch signaling at high temperature and thus compensates for the inherent temperature sensitivity of ligand-induced activation. The outcome is to extend the physiological range over which normal development can occur. Similar mechanisms may provide thermal robustness for other developmental signals.

Graphical abstract

Figure 1

Distinct Mechanisms of Notch Activation Have Opposing Temperature Dependencies (A) Ligand-independent signaling induced by Dx is reduced by Su(dx) coexpression. (B) N^D505A prevents Dl-induced signaling. (C and D) Coimmunoprecipitates: Dx binds N^D505A (C) but not N^R2027 (D). (E) Opposite temperature dependence of basal and Dx-induced N signaling. Su(dx) expression increases basal N signal at low temperatures but decreases signal at high temperatures. (F) Dl-induced signaling (fold change) increases with temperature. Su(dx) reduces signal more effectively at high temperatures. (G) Signaling from extracellular truncated N constructs NEXT and NICD is unaffected by temperature. (H) In situ for wingless expression in wing imaginal discs marks N signaling (discs shown dorsal up, ventral below). Dx expressed along anterior-posterior compartment boundary using ptc-Gal4. In WT discs, Dx-activated N signaling (arrow) becomes weaker at higher temperatures. In Su(dx) mutant discs, the temperature dependency of Dx activity is reversed. Data in (A), (B), (E), (F) and (G) are displayed as means ± SEM, p < 0.05 (minimum of n = 3) for all comparisons stated in legend (Student t test). See also Figure S1.

Figure 2

Dx and Su(dx) Induce Notch Endosomal Trafficking with Distinct Temperature Sensitivities (A and B) Dx coexpression promotes N endocytosis. (C–E) N endocytosis resulting from coexpression with Su(dx) increases with temperature. (F) Quantification of temperature dependence of N endocytosis. Localization was scored after indicated chase times as mostly plasma membrane (PM), plasma membrane and vesicular (PM = V), and mostly vesicular (V). Basal N endocytosis and Su(dx)-induced endocytosis increase with temperature. Dx-induced N endocytosis is markedly less sensitive to temperature, as is N endocytosis when Dx and Su(dx) are coexpressed. (G and H) Endocytosed N (purple) colocalization with Rab5-GFP (green) indicated by arrowheads after coexpression with Dx (G) and Su(dx) (H). (I and J) Time course of N progression through Su(dx) and Dx-induced N endocytic pathways. (K) N localizes to the edge of Rab7-GFP-marked vesicle when coexpressed with Dx. (L and M) N in Su(dx)-expressing cells (L) or Su(dx) + Dx-expressing cells (M) is localized within Rab-7-GFP-marked vesicles. (N) Su(dx)V5 induces N endocytosis, but N is localized to Rab7-GFP-marked vesicle limiting membrane. (O) Distance between Rab7-GFP and peak N localization in late endosomes. Increased value represents increased internalization within Rab7-GFP-marked vesicle. Su(dx) expression or increased temperature overcomes ability of Dx to retain N to the edge of the Rab7-marked limiting membrane. Su(dx)V5 prevents N transfer from the late endosomal limiting membrane. At 18°C, Su(dx) is less effective than at 25°C at transferring N into late endosome lumen. (P) Su(dx) promotes temperature-dependent ubiquitination of Notch, but Su(dx)V5 has no ubiquitination activity. (Q) Su(dx)V5 increases N signaling independently of temperature. Data in (F), (I), (J), (O), and (Q) are displayed as means ± SEM (n = 3, minimum 60 cells or vesicles scored per repeat), p < 0.05 for all differences stated in legend (Student t test). See also Figure S2.

Figure 3

Su(dx) and Dx Regulate Notch Trafficking by Distinct Endocytic Routes (A) When coexpressed with Su(dx), endocytic N (purple) extensively colocalizes with GPI-GFP (green, arrowheads). (B) Enlarged region boxed in (A) showing additional colocalization of Su(dx) (blue, arrowhead). (C) When coexpressed with Dx, endocytosed N does not colocalize with GPI-GFP. (D) Combined expression of Su(dx) and Dx partially relocalizes N into GPI-GFP-marked vesicles (arrowheads). (E) N-positive vesicles were scored as percent GPI-GFP positive or negative. (F) Dx-induced N endocytosis is unaffected in S2 cells treated with mβCD to deplete cholesterol. (G) Su(dx)-induced N endocytosis is suppressed in cholesterol-depleted cells. (H) N localization scored as mostly plasma membrane (PM), plasma membrane and vesicular (PM = V), and mostly vesicular (V) at different cholesterol levels. Cholesterol depletion suppresses Su(dx)-induced N endocytosis, and this is rescued by reloading cells with cholesterol. Dx-induced N endocytosis is unaffected by cholesterol depletion. Overloading of cells with cholesterol induces N endocytosis. (I) Cholesterol overload promotes N endocytosis into GPI-positive vesicles even when Dx is expressed. (J–M) In (J), N (purple) does not colocalize in wing imaginal disc epithelial cells with GPI-GFP (green) when Dx is expressed but does colocalize with GPI-GFP (arrowheads) following expression of Su(dx) (K) or Su(dx)-ΔHECT (L) or after coexpression of Su(dx) with Dx (M). (N and O) Time course of endocytosed N and GPI-GFP localization. (N) In S2 cells, Dx-induced N endocytosis is predominantly through GPI-GFP negative vesicles, while Su(dx) drives N endocytosis through GFP-GFP-positive compartments. (O) Temperature increases the localization of Notch to GPI-GFP-positive compartments when Su(dx) and Dx are coexpressed. An increased proportion of N and GPI-GFP colocalization is observed after longer chase periods. (P) N (blue), endocytosed after Dx expression is localized (arrow) in an endocytic vesicle, marked with Dextran (red) immediately adjacent to a GFP-GPI-marked compartment. Data in (E), (H), (I), (N), and (O) are displayed as means ± SEM (n = 3, minimum of 60 vesicles or cells scored per repeat), p < 0.05 for all differences stated in legend. See also Figure S3.

Figure 4

Distinct Requirements for Notch Signaling Initiated by Different Mechanisms (A) Basal N signaling in S2 cells is reduced by cholesterol depletion, but Dx-induced N signaling is unaffected. (B) Basal and Dx-induced N signals are reduced by RNAi of Mam, Dyn, and Rab5, but only Dx-induced signaling is reduced by RNAi of Rab7 and Dor. Basal, but not Dx, signal depends on Kuz. (C) N signaling in cells exposed to Dl is reduced by cholesterol depletion, but the fold change after ligand-induction is increased. Signaling by the N^D505A construct is removed by cholesterol depletion. (D) RNAi knockdown of components of the GSL synthesis pathway preferentially reduces the basal N signal compared to Dx and ligand-induced signal. chol., cholesterol. (E) In cholesterol-depleted cells, ligand-induced N signaling is reduced by RNAi of Kuz, Dyn, and Rab5 but insensitive to Rab7 or Dor RNAi. (F) Rab5DN reduces basal, ligand, and Dx-induced signaling, but only Dx signaling is reduced by Rab7DN. Data are displayed as means ± SEM (minimum n = 3), p < 0.05 for differences stated in the legend (Student t test). See also Figure S4.

Figure 5

Modeling Consequences of Su(dx) and dx Mutations on Notch Signaling in the Drosophila Wing (A) N endocytic trafficking routes. Asterisks mark experimentally observed temperature-dependent steps. Key fluxes are designated r. (B) Model of the effects of mutations of Su(dx) and dx on N signaling ([NICD] arbitrary units) in the Drosophila wing at 29°C. Arrowhead marks mutual suppression observed in double mutants. Black dot represents WT concentrations of both Dx and Su(dx). Asterisk marks stronger loss of Notch signaling expected from increased Su(dx) in the absence of Dx. Yellow shading indicates expected WT conditions, orange to red shading indicates increased expectation of a Notch gain of function phenotype (wing vein gaps) and green to blue shading indicates increased likelihood of a Notch loss of function phenotype (vein thickening). (C–F) Mutual suppression resulting from combined Su(dx) and dx mutations restores temperature-sensitive phenotypes of each back to WT at 29°C. (G) Increasing WT Su(dx) copy number enhances the dx mutant wing phenotype. (H) Su(dx) mutant wing phenotype is suppressed by car. (I) Simulation of N signaling (NICD) versus temperature in WT and mutant backgrounds. Arrowheads mark expected upper and lower signaling thresholds corresponding to yellow shaded area in (B). (J) dx mutant phenotype weakens as temperatures decrease from 25°C down to 16°C but worsen again at 14°C. At 25°C, Su(dx) mutants suppress dx phenotype, but as temperature is lowered, the loss of Su(dx) has less effect on the strength of the dx phenotype. At 18°C and below, Su(dx) mutation enhances dx phenotype. Arrowheads indicate distal thickening on L3 and L4 veins. (K) Percent wings with L3 and L4 vein thickening. The dx phenotype was increased at 14°C compared to 16°C (p < 0.05) and reduced at 18°C compared to 25°C (p < 0.001). Enhancement or suppression of dx phenotype in dx;Su(dx) double mutant flies was observed at different temperatures (p < 0.0001, Fisher’s exact test, n > 40 for each genotype tested). See also Data File S1.

Figure 6

Tuning of Network Parameters Predicts Tissue-Specific Cooperation between Su(dx) and Dx to Downregulate Notch (A and B) Modeling of the combined loss of function of Su(dx) and dx on N signaling ([NICD] arbitrary units) when lysosomal activation component is reduced (A) or increased (B). These models differ from that shown in Figure 5B only by a 5-fold reduction (A) or 5-fold increase (B) in k₉, which determines lysosomal activation component. Black dot represents WT [Dx] and [Su(dx)], color shading is as described in the legend for Figure 5B. (C–E) Su(dx), dx mutant combination results in N gain of function. (C) WT leg, tarsal joints between segments T2/T3, and T3/T4 indicated. (D) Extra joint tissue (arrowhead) observed in dx;Su(dx). (E) Percent legs with ectopic joint increases with temperature in double mutants (p < 0.01, Fisher’s exact test, n > 60 legs per genotype). (F) Additional copy of WT Su(dx) in dx mutant results in loss of joints at 25°C (86.4% legs, n = 66) not seen with additional WT Su(dx) copy in WT background (n = 80). (G) Dx expression in WT results in both partial joint loss (asterisk) and ectopic joint material (arrowhead). (H) When active Rab7 (Rab7^QL) is coexpressed with Dx, joint tissue is lost (arrow). (I) TRPML coexpression with Dx results in loss of joints. (J and K) Coexpression of TRPML with Dx increases wingless expression in wing discs compared to Dx alone (arrow). TRPML expression alone has no effect (data not shown). See also Data File S1.

Figure 7

Modularity of Network Masks Critical Roles for Su(dx) and Dx during Drosophila Embryogenesis (A) dx;Su(dx) maternal/zygotic mutant embryos show more frequent and temperature-dependent gaps in the central nervous system (CNS) compared to Su(dx) or dx (p < 0.005, n > 30). (B) Anti-Hrp-stained CNS of WT stage 15-16 embryo. (C) CNS loss in dx;Su(dx) embryo. (D) WT stage 15 embryo CNS, anti-ELAV (red), anti-HRP (green), DAPI (blue). (E) Neurogenic phenotype after Dx expression using mat-tubGal4. (F) WT embryo, anti-Eve (purple), anti-Hrp (green). Pairs of RP2 neurons are indicated by arrows. (G) Extra RP2 neurons in dx (arrowhead). (H) Loss of RP2 in dx;Su(dx) embryo (arrow). (I) RP2 loss in dx;Su(dx) at 29°C is more frequent than for either mutant on its own or for dx;Su(dx) at 25°C (p < 0.01). A gain of RP2s was observed in dx compared to WT at 29°C, p < 0.01 (>230 segments per genotype scored at stage 15/16). (J) Reduced sim expression in stage 7-8 dx embryos, (p < 0.001, n > 60), with increased penetrance at higher temperature. The dx phenotype was not strongly reduced by Su(dx). (K–L′) In situ staining of sim in WT (K) and dx;Su(dx) (L) embryos. (K′) and (L′) show enlarged images of similar areas of (K) and (L) where arrowheads indicate gap in sim expression in (L) and (L′). Statistics by Fisher’s exact test.

Figure S1

Notch Signaling Is Robust to Temperature Variation, Related to Figure 1 (A) N signaling marked by a Notch response element (NRE)-GFP reporter (green) at 18, 25 and 29^oC along presumptive vein borders in staged pupal wings (equivalent to 32 hours AP at 25^oC). (B) Late third instar imaginal discs (DAPI stained, blue) showing NRE-GFP expression along Dorsal-Ventral boundary at 18, 25 and 29^oC. (C) Late third instar imaginal discs from larvae raised at 18, 25 and 29^oC and in situ stained for wingless, a N responsive gene at the Dorsal-Ventral boundary (arrow head).

Figure S2

Dx and Su(dx) Induce Notch Endocytosis by Different Routes, Related to Figure 2 (A) Using N antibody uptake assay, Dx induced internalisation of Notch is reduced compared to control by Dynamin, Clathrin Heavy Chain and Synaptojanin RNAi (upper panels in B). In contrast Su(dx) induced Notch endocytosis is prevented by Dynamin but not Clathrin Heavy Chain or Synaptojanin RNAi (lower panels). (B) Scoring of Notch localisation as % cells with mainly plasma membrane localisation (PM), vesicular localisation (V) or a mixture of plasma membrane and vesicular (PM=V). Data displayed as mean ±SEM from minimum of 3 repeats, each scoring minimum of 60 cells, ^∗ indicates p<0.05 compared to GFP RNAi control (Student t-test).

Figure S3

Endocytosis through GPI-Protein Positive and Negative Routes in S2 Cells, Related to Figure 3 (A–D) Immunostained cells after Notch antibody uptake endocytosis assay (Notch, purple; GPI-GFP green). (A) Basal Notch endocytosis of N^D505A is into GPI-GFP positive vesicles. (B) Dx expression induces endocytosis of N^D505A through GPI-GFP negative vesicles. (C and D) Basal Notch endocytosis of N^R2027A is into GPI-GFP positive vesicles (C) and this is not altered by Dx expression (D). (E) Quantification of N localisation as % cells with mainly plasma membrane localisation (PM), vesicular localisation (V) or a mixture of plasma membrane and vesicular (PM=V). (F) % Notch containing vesicles which are GPI-GFP positive or negative. Data in E,F displayed as mean ± SEM from minimum of 3 repeats, each scoring minimum of 60 cells or vesicles per repeat, ^∗ indicates p<0.05 compared to respective controls (Student t-test). (G–I) GFP antibody uptake assay of GPI-GFP endocytosis (GFP-total green; GFP antibody, purple). (G) In control cells at 25^oC, most pulse labelled GFP is endocytosed (arrowheads). (H and I) GPI-GFP endocytosis is reduced by cholesterol depletion (H) or by lowering the temperature to 18^oC (I). (J) Quantification of anti-GFP localisation in endocytic uptake assay shown as % cells with mainly plasma membrane GFP localisation (PM), vesicular localisation (V) or a mixture of plasma membrane and vesicular (PM=V). Data displayed as mean ± SEM from minimum of 3 repeats, each scoring minimum of 60 cells per repeat, ^∗ indicates p<0.05 compared to respective controls (Student t-test). (K and L) Endocytosed GPI-GFP traffics through Rab5 (K) and Rab7 (L) positive vesicles. (Total-GPI-GFP, green; GFP antibody marking endocytosed GFP (blue); Rab5-RFP (K) or Rab7-RFP (L), red). GFP colocalisation with Rab5 or Rab7 indicated with arrowheads.

Figure S4

RNAi Knockdown Efficiency and Effect of ADAM 10 Inhibitor on Notch Signaling in S2 Cells, Related to Figure 4 (A) Expression of mRNA of target genes in S2 cells was detected by real time Q-PCR and effect of RNAi knockdown is shown relative to respective normalised control exposed only to GFP RNAi. Expression of Notch is from pMT-Notch transfected cells, the expression of other genes reflects endogenous expression. (B) RNAi Knockdown of Ser expression in S2 cells detected by Q-PCR. (C) Notch signaling luciferase assay. Basal signal through WT Notch is insensitive to Ser RNAi, but signal from N-VP16 shows small but reproducible reduction compared to control (GFP RNAi). (D) Western blots showing relative reduction in protein levels, Rab5 and Dor. Anti-Peanut was used as a loading control. (E) Quantification of western blots. (F–H) Notch signaling luciferase assay. (F) Su(dx)V5 induces signaling similarly through N and N^D505A but (G) Ser does not induce signaling through N^D505A. (H) Kuz inhibitor BB94 blocks basal and Dl-induced signaling but not Dx activation of N. As a control the Presenilin inhibitor DAPT blocks, basal, Dx and Dl-induced signal. Extracellular truncated N controls behave as expected. NEXT is inhibited by DAPT but not BB94, NICD is not affected by either inhibitor. Data in A-C, E-H displayed as means ± SEM for 3 repeats.