Structural features of the Notch ankyrin domain-Deltex WWE₂ domain heterodimer determined by NMR spectroscopy and functional implications

Summary

The Notch signaling pathway, an important cell fate determination pathway, is modulated by the ubiquitin ligase Deltex. Here we investigate the structural basis for Deltex-Notch interaction. We used NMR spectroscopy to assign the backbone of the Drosophila Deltex WWE₂ domain and mapped the binding site of the Notch ANK domain to the N-terminal WWE_A motif. Using cultured Drosophila S2R+ cells, we find that point substitutions within the ANK-binding surface of Deltex disrupt Deltex-mediated enhancement of Notch transcriptional activation and disrupt ANK binding in cells and in vitro. Likewise, ANK substitutions that disrupt Notch-Deltex heterodimer formation in vitro block disrupt Deltex-mediated stimulation of Notch transcription activation and diminish interaction with full-length Deltex in cells. Surprisingly, Deltex-NICD interaction is not disrupted by deletion of the Deltex WWE₂ domain, suggesting a secondary Notch-Deltex interaction. These results show the importance of the WWE_A:ANK interaction in enhancing Notch signaling.

eTOC blurb

Notch signaling determines cell fate in animals. Notch receptor activity is modulated by several effectors, including Deltex, an E3 ubiquitin ligase. Here, Carter et al. use NMR to map the Notch binding site on the Deltex WWE₂ domain, and show this interaction to enhance Notch-mediated transcriptional activation.

Graphical Abstract

Introduction

The Notch signaling pathway is a ubiquitous transmembrane juxtracrine pathway in animals ranging from invertebrates to humans ^1,2. Adjacent cells interact asymmetrically through the Notch receptor and its ligands to specify cell fate. Notch signaling is important for development in a variety of tissues including the nervous system, the skin, and various organ systems. Notch signaling is also important for stem cell maintenance and differentiation ³ and plays a role in hematopoiesis ⁴. In humans, defects in Notch signaling can lead to diseases including cancers and leukemias ^5–7. Thus, managing aberrant Notch signaling has generated significant therapeutic interest ^8,9.

Mutations that disrupt the Notch pathway were first identified in Drosophila nearly a century ago (see Takebe et al., 2015). In recent decades, a detailed molecular understanding of the Notch pathway has emerged. The central player is the Notch receptor, a ~2,700 residue single-pass transmembrane receptor. Upon activation by ligands attached to adjacent cells, two proteolytic cleavages (termed S2 and S3) cooperate to release the Notch intracellular domain (NICD) into the cytosol (Figure 1, left) through what is referred to as the “canonical pathway” for Notch signaling. Released from the plasma membrane, NICD can enter the nucleus, interact with the DNA-binding protein CSL, recruit transcriptional activators, and initiate transcription of Notch-responsive genes ¹¹.

Figure 1. Mechanisms of Notch signaling and regulation by Deltex.

The left side of the figure shows canonical Notch signaling, in which a DSL-type ligand on an adjacent cell binds to the extracellular domain of the Notch receptor, stimulating S2 cleavage by an ADAM protease. This cleavage liberates the Notch extracellular domain (NECD) and leaves a transmembrane-bound fragment that contains the intracellular domain (NEXT). NEXT is a substrate for S3 cleavage by γ-secretase either at the plasma membrane or in an endocytic vesicle, which liberates the Notch intracellular domain (NICD). Cytoplasmic NICD then enters the nucleus, where it activates transcription through the transcription factor CSL. The right-side of the figure shows non-canonical Notch signaling, which is mediated by Deltex. In the absence of extracellular ligands, Deltex can stimulate full-length Notch receptor to enter endocytic vesicles in a Rab5-dependent process. These vesicles can either be recycled or can fuse with early endosomes (EE) and then late endosomes (LE). At this stage, the intact Notch receptor can either be down-regulated by internalization to a multivesicular body that is destined for degradation in the lysosome (Lys), or can be activated through a second Deltex-mediated process, again forming NICD. Down-regulation through internalization is promoted by Kurtz and Suppressor of Deltex (SuDx). Depending on the balance between the SuDx and Deltex activity at the late endosomal stage, Deltex can either promote Notch signaling by generating NICD or can inhibit Notch signaling by promoting internalization followed by degradation.

In addition to the ligand-dependent canonical signaling pathway, the Notch receptor can be activated (and also down-regulated) through ligand-independent mechanisms ^12,13. These mechanisms involve endocytosis of intact Notch receptors from the plasma membrane into early endosomes ¹⁴. These internalized Notch receptors then advance to late endosomes ^15–17, where they can either be incorporated into internal vesicles of multivesicular bodies (MVBs) and subsequently degraded, or can be processed in a proteolytic mechanism that appears to resemble that of ligand-dependent Notch signaling (Figure 1, right), generating active cytosolic NICD.

The interactions that regulate ligand-independent signaling are not completely understood. Several proteins have been identified that control various aspects of ligand-independent signaling. The proteins Suppressor of Deltex, Kurtz, and Shrub have all been shown to promote entry into internal vesicles of multivesicular bodies ^18–20. Another protein involved in ligand-independent Notch signaling is Deltex. The effects of Deltex on Notch signaling are context dependent, activating Notch signaling in some settings, and inhibiting it in others ¹⁵. Deltex appears to modulate the fates of vesicular Notch at two different stages. First, Deltex promotes endocytosis of full-length Notch receptor at the plasma membrane. This Deltex-mediated endocytosis can either activate or repress Notch, depending on whether degradation of internal vesicular Notch in MVBs or ligand-independent processing of late endosomal Notch is dominant. Second, Deltex promotes the processing of the full-length Notch receptor to NICD in the late endosome described above, activating Notch signaling provided the receptor has late endosomal access.

Deltex is composed of three different regions (Figure 2; Busseau et al., 1994). The most C-terminal region encodes a RING E3-ubiqutin ligase domain ²² and a unique domain that is conserved among Deltex genes ²³. The most N-terminal region is composed of a tandem pair of WWE motifs (WWE₂). Each WWE motif is about 85 residues in length; Aravind, 2001), and is essential for modulation of Notch signaling ²⁵. WWE motifs are typically found in genes encoding either E3 ubiquitin ligase domains (either RING or HECT domains) or poly(ADP-ribose) polymerases (PARPs; Aravind, 2001). Deltex is unique among these genes in encoding a tandem pair of WWE motifs.

Figure 2. Primary structure of the Drosophila Deltex and Notch proteins, and constructs used in this study.

The top schematic shows the Notch intracellular domain, which extends from the γ-secretase cleavage site to the C-terminus. The seven ankyrin repeats are shown in blue, and the TWFP sequence that binds tightly to CSL is shown in green. The second schematic gives the boundaries of the Notch ankyrin domain used in this study. The third schematic shows the full-length Deltex protein. The N-terminal WWE₂ tandem spans residues 35 to 201. Two polyglutamine segments span from residues 261 to 302 and from residues 488 to 513. The C-terminal region contains a RING and a DTC domain. The bottom schematic gives the boundaries of the Deltex WWE₂ construct used in this study.

The Deltex WWE₂ domain has been shown to be responsible for binding to the intracellular ankyrin (ANK) domain of the Notch receptor ^26,27. A crystal structure of the Drosophila WWE₂ domain shows that the two domains form a rigid unit with roughly two-fold rotational symmetry ²⁸. A large cleft is present between the two motifs, although it may be partly occluded by a loop in the WWE_A motif that is not visible in the crystal structure.

The structural basis of the ANK:WWE₂ interaction is unclear, as is the importance of this interaction in the two Deltex-mediated steps in ligand-independent Notch processing. Mutational studies of the recognition loops of ANK identify a hot-spot for binding involving the R1985 and R2027 residues of the Notch receptor ²⁶. However, the interaction surface on Deltex is unknown. It is possible that the ANK domain interacts with the cleft formed by the two WWE motifs; alternatively, it is possible that only one WWE motif is involved in binding, with the other providing an alternative function or interaction.

Here we use solution NMR to map the binding surface for ANK domain on the Drosophila Deltex WWE₂ domain. Using standard triple-resonance experiments, we have assigned NH resonances of 89 percent of the non-proline backbone residues that are ordered in the crystal structure. Titrating the ¹⁵N-labelled WWE₂ domain with an unlabeled ANK domain produces a variety of changes to the ¹⁵N, ¹H-TROSY spectrum in which some resonances shift, others broaden, but some remain unchanged. Mapping these changes onto the structure of the Deltex WWE₂ domain shows that the ANK domain binds to the N-terminal WWE_A motif, away from the central cleft and the C-terminal WWE_B motif. These chemical shift perturbations (CSPs) are disrupted by the ANK R2027A point-substitution, which has previously been shown to destabilize the Deltex WWE₂ interaction in sedimentation velocity analytical ultracentrifugation (sv-AUC) studies ²⁶. We find that alanine substitutions of Deltex WWE₂ residues K74 and Y91 in the ANK binding site identified by chemical shift perturbation also disrupt binding in vitro. Using a cell culture assay for non-canonical Notch signaling by Deltex, we show that these mutants disrupt Deltex-mediated enhancement of Notch transcriptional activation. Consistent with these observations, the Deltex WWE_A K74A and Y91A substitutions decrease co-precipitation of full-length Deltex with NICD in Drosophila S2R+ cells. Surprisingly, although the Notch ANK R2027A and R1985A substitutions in the full-length Notch receptor disrupt Deltex-mediated Notch transcriptional activation as well as Deltex WWE₂-ANK dimerization in vitro, they do not disrupt co-precipitation of full-length Deltex with full-length NICD in S2R+ cells. This disparity between co-precipitation in cells, in vitro binding, and Deltex-mediated enhancement of Notch transcriptional activation suggests that there are additional modes of interaction between regions of Deltex outside the WWE₂ domain and NICD. Indeed, a Deltex construct lacking the ANK-interacting tandem WWE₂ domain, though unable to enhance Notch transcriptional activation, still co-precipitates with the Notch intracellular domain in cells, suggesting the ANK:WWE₂ interaction is essential for Deltex-mediated transcription enhancement, whereas a secondary NICD:Deltex interaction facilitates a separate aspect of the Notch-Deltex signaling axis.

Results

NMR assignments of the WWE₂ domain of Drosophila Deltex

To assign the backbone resonances of the WWE₂ domain of Deltex (234 residues), we prepared a ¹³C-, ¹⁵N-labelled Dx1A* sample in which >90% of the nonlabile (carbon-bound) hydrogens are replaced with deuterium. This triply-labelled protein displays well-resolved cross peaks (Figure 3A). Using TROSY-based triple-resonance experiments on ²H-, ¹³C-, ¹⁵N-labelled WWE₂ samples with ¹H amide hydrogens, we were able to assign backbone resonances for 89 percent of non-proline residues at 25 °C over the region of Dx1A* that is ordered in the crystal structure (residues 43–211), providing good coverage throughout the tandem WWE₂ domain (Table S1). In addition, 31 residues were assigned within the disordered N- and C-terminal segments (residues 26–42 and 212–261).

Figure 3. NMR spectra of the Drosophila Deltex WWE₂ domain and its interaction with the Notch Ankyrin domain.

(A) ¹H-¹⁵N-TROSY spectrum of a ²H-, ¹⁵N-labelled fragment of Drosophila Deltex spanning residues 26–261 (Dx1A*), showing high chemical shift dispersion. Using triple-resonance techniques, assignments were made for the 169 of the 217 backbone amides (excluding prolines). (B) Titration of 100 μM ²H-, ¹⁵N-labelled Dx1A* from (A) with the R2007A variant of unlabeled Notch ANK. Colors correspond to molar ratios of 1:0 (black), 1:1 (blue), 1:2 (green), and 1:4 (red) Deltex to ANK R2007A. (C) Spectrum of a 1:4 molar ratio of the Dx1A* (100 μM) and an unlabeled Notch ANK R2007A/R2027A (red; black spectrum shows Dx1A* on its own). Conditions: 300 mM NaCl, 25 mM NaPO₄, pH 7.0, 25 °C (A) and 35°C (B, C).

Chemical shift perturbations in the WWE₂ domain from binding of the Notch Ankyrin domain

To determine whether the binding of the Notch ANK domain to the Deltex WWE₂ domain could be detected by NMR, we added excess unlabeled ANK R2007A to ²H-, ¹⁵N-labelled Deltex WWE₂ at 25 °C. The R2007A point-substitution disrupts ANK dimerization ²⁹ but has been shown not to interfere with ANK-WWE₂ interaction (Allgood and Barrick, 2011; note that this variant was identified as R107A in that study). Addition of unlabeled ANK produced significant changes to the WWE₂ TROSY spectrum, confirming binding. However, one of the clearest changes in the spectrum was a fairly uniform reduction in cross-peak intensities, which is likely to result from the increased size of the WWE₂:ANK complex (57.2 kDa), as well as potential exchange broadening. This level of broadening precluded chemical shift mapping, which relies on resolving resonances that shift upon binding from resonances do not. We found that we could significantly improve the spectral quality of the WWE₂-ANK complex by increasing the temperature to 35 °C, with little effect on the TROSY spectrum of the WWE₂ domain alone (Figure S1). Consequently, assignments made at 25 °C could easily be transferred to 35 °C.

Taking advantage of the improved spectral resolution at 35 °C, we performed a titration of the ¹⁵N, ²H-labelled WWE₂ domain that spanned from sub-stoichiometric to saturating concentrations of unlabeled ANK R2007A. Samples were prepared at molar ratios of 1:0, 4:1, 2:1, 1:1, 1:2, and 1:4 WWE₂ to ANK, each at a WWE₂ concentration of 100 μM, and ¹H-, ¹⁵N-TROSY spectra were collected at each concentration. With a dissociation constant of about 13 μM under these conditions ²⁶, almost all (96%) of the labelled WWE₂ should be heterodimeric in the 1:4 sample. Cross peaks in this titration exhibited a variety of responses to the addition of ANK (Figure 3B). Some peaks disappear above a stoichiometric excess of ANK (i.e., at 1:2, Figure 4A) whereas others disappear only at the highest ANK concentration (1:4, Figure 4B, C). Other peaks undergo modest shifts in a traceable manner (Figure 4D). Some cross peaks broaden but do not shift (Figure 4H). Others are unchanged throughout the titration aside from an overall broadening effect that is expected from the increased molecular weight (Figure 4I). Based on these different responses, we classified cross peaks into five categories (Table S2). There are 58 cross peaks that undergo large scale changes in peak intensity and/or chemical shift (categories 1, 2, and 3; Table S2), and 76 cross peaks that show little or no changes in chemical shift (categories 4 and 5, Table S2).

Figure 4. Representative behavior of Deltex WWE₂ cross peaks titrated with unlabeled Notch ANK domain.

Molar ratios of Deltex WWE₂ domain to Notch ANK are 1:0, black; 1:1, blue; 1:2, green; 1:4, red. (A) T77 disappears at a molar ratio of 1:2 (category 1). (B) W133 disappears at a molar ratio of 1:4 (category 2). (C) S47 shifts slightly and disappears at a molar ratio of 1:4 (category 2). (D) G115 shifts completely from one location to another in a traceable manner at a molar ratio of 1:4 (category 3). (E) Y60 shifts slightly (category 4). (F) V153 loses intensity, but it does not disappear completely (category 4). (G) I156 both shifts slightly and loses intensity (category 4). (H) K56 broadens (category 4). (I) S53 exhibits no change (category 5). Conditions: 300 mM NaCl, 25 mM NaPO₄, pH 7.0, 35 °C, 800 MHz.

Most cross peaks with high sensitivity (categories 1–3) come from residues in the N-terminal WWE_A motif (44 out of 58 cross peaks; Table S2) and are localized to the N-terminus of β-strand A1, α-helix A1 and β-strands A3 and A4, as well as the loops connecting βA1 to βA3, and βA3 to βA4 (Table S1). Of the 23 cross peaks in the N-terminal WWE_A motif that show little or no change in intensity or chemical shift in response to ANK binding, the majority are in the C-terminus of β-strands A1 and in β-strands A2, A5, and A6 (Table S1). Most of the residues in the large disordered loop within WWE_A (spanning from S102 to L111) show little to no chemical shift change; significant changes are limited to the residues at the loop termini (S102, G110, and L111), suggesting that this loop remains disordered in the ANK-bound state.

In contrast, most of the cross peaks that have low sensitivity to ANK binding (categories 4 and 5, Table S2) come from residues in the C-terminal WWE_B motif (51 out of 76 cross peaks, purple-shaded entries in column 3 of Table S2). Of the 11 cross peaks from the C-terminal WWE_B motif that undergo large changes in intensity, chemical shift, or line width, three are the most N-terminal residues in WWE_B (M119, F120, Y121), and are part of a continuous β-strand B6 that spans both motifs, and three (S123, K132, and W133) are in the interface with WWE_A

Four residues that undergo large chemical shift changes are in the C-terminal sequence beyond the WWE_B motif (T202, Q204, V210, K211). This region makes extensive interactions with the WWE_A motif, including a β-strand (β’, Table S1) that hydrogen bonds directly with the N-terminal β-strand A1 of WWE_A.

Structurally, the residues that show high chemical shift sensitivity to binding of the Notch ANK domain are localized on the outside face of the WWE_A motif (Figure 5), including the αA1 helix and the βA1, βA3, and βA4 strands (Figure 5C, D). This surface is away from the large cleft between the two WWE motifs, and forms a continuous patch on the N-terminal end of the WWE₂ domain (Figure 5D). Although there are a few resonances on the analogous surface of the WWE_B motif that disappear at high molar ratios of ANK, most of the chemical shifts from this surface remain unchanged. Importantly, this is not a result of missing chemical shift information, as most of the residues on the outside surface of the WWE_B domain are assigned (Figure 5E).

Figure 5. Locations on the Deltex WWE₂ domain of chemical shift perturbations resulting from binding of the Notch ANK domain.

(A) Ribbon diagram of the crystal structure of Dx1A* (2A90.pdb, Zweifel et al., 2005) showing the WWE_A and WWE_B motifs (orange and purple) and the C-terminal linker (dark grey) in different orientations. (B) Ribbon diagram showing each secondary structure element in the WWE_A motif. Analogous secondary structure elements in the WWE_B motifs have the same color (see Figure S1). (C) Ribbon diagram showing chemical shift perturbations from binding of Notch ANK. Resonances that disappear above 1:1 molar ratio of ANK to WWE₂ (class 1) are colored black, residues that disappear above a molar ratio of 1:2 (class 2) are colored red, resonances that shift upon binding (class 3) are colored brown. Resonances that show modest shifts or increases in linewidth upon binding (class 4) are colored beige. Resonances that are unaffected by binding are colored light grey. Based on the intersection of conservation (see panel F) and chemical shift perturbations, we identified a set of five residues (sticks) to substitute to attempt to disrupt binding of the Notch ANK domain. Two of these (K74A and Y91A) destabilized the Dx1A*: Notch R2007A ANK complex (Table S3). (D) Surface representation with the same color scheme as (C). (E) Surface representation showing unassigned non-proline (yellow) and proline (green) residues. (F) Conservation based on CONSURF analysis. Residues are colored by conservation score, in the following order: gray (least conserved), light pink, light orange, pale yellow, pale green, cyan, blue, purple, black (most conserved). In each panel, the center-right image is an end-on view of WWE_A, and the right image is an end-on view of WWE_B.

The R2027A mutation in the Notch ANK domain disrupts binding to the WWE₂ domain

To test whether the Notch ANK:Deltex WWE₂ interaction probed by NMR is specific, we titrated an ¹⁵N-, ²H-labelled WWE₂ sample with a variant of the ANK domain bearing the R2027A substitution (in addition to the R2007A variant that disrupts ANK dimerization). The R2027A substitution has been shown to disrupt ANK:WWE₂ binding in sv-AUC studies (Allgood and Barrick, 2011; this variant was identified as R127A in that study). Unlike the titration with the binding-competent ANK R2007A variant, the chemical shifts of the Deltex WWE₂ domain are largely unperturbed by addition of a four-fold molar excess of the R2027A/R2007A variant (Figure 3C). All of the cross peaks in the ¹H-, ¹⁵N-TROSY spectrum remain sharp, and nearly all retain their chemical shifts. Three assigned NH cross-peaks disappear from the 1:4 spectrum (from residues 82–84). These three residues are in the WWE_A motif, in the loop between beta strands βA3 and βA4. This loop displays high sensitivity to binding of the binding-competent R2027A ANK variant (Tables S1, S2). These minor changes to the WWE₂ TROSY spectrum indicate that interaction with the ANK is significantly weakened by the R2027A substitution, consistent with the results of sv-AUC. Given the sharp lines for the vast majority of WWE₂ cross peaks, it seems likely that the disappearance of cross peaks for residues 82–84 results from local exchange-broadening due to a transient interaction.

Effects of point substitutions on the Deltex WWE₂:Notch ANK interaction in vitro

To examine whether substitutions on the binding interface of the Deltex WWE₂ domain identified from CSPs disrupt binding to the Notch ANK domain, we made alanine substitutions of several conserved residues that show large CSPs and quantified interactions with the R2007A variant of ANK using sv-AUC. Sedimentation of a 45 μM 1:1 mixture the wild-type Deltex WWE₂ and R2007A ANK domains at 50,000 rpm yields a weight-averaged sedimentation coefficient (s_w) of 3.51×10⁻¹³ sec (Table S3), compared to s_w values of 2.20×10⁻¹³ and 2.25×10⁻¹³ sec for each domain in isolation. Sedimentation of the Y91A Deltex WWE₂ variant with R2007 ANK showed a significantly reduced sedimentation coefficient (2.36×10⁻¹³ sec, Table S3). The K74A WWE₂ variant showed a modest decrease, whereas 1:1 mixtures of the other variants had s₂₀ values that were indistinguishable from wild-type Deltex. These results suggest Deltex residues Y91 and, to a lesser extent, K74, make direct stabilizing interactions with the Notch ANK domain. These two residues span the central β-sheet in the WWE_A motif and are not particularly close to one another (the distance between respective C_α atoms is 19.6 Å), suggesting that the ANK binding surface on the WWE_A surface is rather extended.

Cell-based assay to detect Deltex activation of Notch signaling

To evaluate the contribution of the Deltex WWE₂ interaction with the Notch ANK domain to the functional and biochemical aspects of Notch signaling, we developed a dual luciferase assay using Drosophila S2R+ cells to probe the effects of point-substitutions that disrupt the WWE₂:ANK interaction on enhancement of Notch-mediated transcription ^30,31. In this assay, we transfected vectors containing Notch and Deltex genes with a vector containing the firefly luciferase gene downstream of a Notch responsive promotor along with a control vector containing the Renilla luciferase gene (see Materials and Methods). Because Deltex is primarily thought to modulate ligand-independent aspects of Notch signaling (Figure 1), we examined the effect of Deltex on signaling from a full-length (FL) Notch receptor, rather than the constitutively active Notch intracellular domain (NICD; consistent with signaling through a ligand-independent mechanism, transcriptional activation by NICD is unaffected by co-transfection with Deltex (not shown). Transfection of Notch alone results in a modest increase in transcriptional activation (Figure 6A), as expected in the absence of ligands. However, co-transfection with Deltex results in a dose-dependent increase in luciferase activity (Figure 6B); this increase is dependent on Notch, since transfection of Deltex alone fails to activate luciferase expression (Figure 6A). Moreover, this enhancement of Notch signaling depends on the Deltex WWE₂ domain, since co-transfection of a construct in which the WWE₂ domain is deleted (ΔWWE₂) fails to activate luciferase activity (Figure 6A). This requirement for the WWE₂ domain is consistent with previous results that show this domain to be essential for various Deltex-mediated developmental transformations in Drosophila ³².

Figure 6. Co-transfection with Deltex stimulates Notch transcriptional activation.

(A) Transfection of S2R+ cells with full-length Deltex alone or Deltex lacking the WWE₂ domain alone does not activate transcription from a luciferase reporter gene under the control of a Notch-responsive promoter. Transfection with full-length Notch receptor (10 ng DNA) results in an increase in transcriptional activation. Co-transfection of Notch with full-length Deltex (40 ng DNA) enhances transcriptional activation by full-length Notch; this enhancement is eliminated when the Deltex WWE₂ domain is deleted. The statistical significance (p) of differences in transcriptional activation was quantified using a two-sided t-test; *** corresponds to p < 0.001, n.s. indicates not significant (p>0.05). (B) Dose-dependence of Notch-mediated transcription by full-length Deltex.

To directly probe the importance of binding of the Notch ANK domain to the Deltex WWE₂ domain in Deltex-mediated transcriptional enhancement, we performed our dual-luciferase assay using Deltex constructs with point substitutions that disrupt heterodimer formation. As shown above in our NMR studies, K74 and Y91 in the Deltex WWE_A motif undergo large chemical shift perturbations upon ANK binding, and substitutions of these residues with alanines destabilize the heterodimer in sv-AUC. Both substitutions decreased the Deltex-mediated transcriptional enhancement of WT Notch and when combined completely abolished Deltex-mediated transcriptional enhancement (Figure 7, left), consistent with the picture that Deltex-mediated transcriptional enhancement is facilitated by ANK-WWE₂ binding.

Figure 7. Variants that disrupt ANK:WWE₂ binding decrease Deltex-mediated transcriptional activation by full-length Notch.

Co-transfection of S2R+ cells with wild-type Notch (left), Notch R1985A (middle), or Notch R2027A variants (right) and either wild-type Deltex (grey bars) or Deltex K74A, Y91A, or K74A/Y91A variants (stippled bars). Luciferase activities were normalized to that of wild-type Notch in the absence of Deltex (white bars). Both Deltex single-residue variants enhance transcription activation by wild-type Notch, though to a lower level than wild-type Deltex. However, when combined into a double mutant construct, Deltex-mediated enhancement is no longer detected. In contrast, the Notch R1985A and R2027A variants are refractory to activation by Deltex, though both Notch variants do activate transcription over background. Error bars represent +/− the standard deviations of three replicates each from two independent experiments. Statistical significance from a two-sided t-test is indicated at p < 0.001 (***); n.s. indicates not statistically significant. Cells were transfected with 10ng Notch and 30ng Deltex as indicated.

Point substitutions to the Notch ANK domain were chosen based on a previous study ²⁶, which introduced substitutions to the recognition loops and proximal helices of repeats two through seven of the Drosophila Notch ANK domain, and measured the effects on binding to the Drosophila Deltex WWE₂ domain using sv-AUC. This study identified a hot-spot for binding centered on residue R2027 (ANK repeat 4); as described above the R2027A substitution rendered binding of Notch ANK domain to the Deltex WWE₂ domain undetectable, both by sv-AUC ²⁶ and by NMR (Figure 3C). Substitution of an adjacent arginine 1985 (ANK repeat 3) with alanine resulted in a 4-fold decrease in the binding constant (Allgood and Barrick, 2011; this variant was identified as R85A in that study). Co-transfection of Notch receptors bearing the strongly binding disruptive R2027A substitution completely eliminates the Deltex-mediated enhancement of Notch transcriptional activation (Figure 7, right), as does the modestly disruptive Notch R1985A substitution (Figure 7, middle). As with the K74A and Y91A substitutions to the WWE_A domain, these results support the involvement of the interface between the Notch ANK and Deltex WWE_A domains in Deltex-mediated transcriptional enhancement.

An alternative explanation for the lack of Deltex-mediated stimulation of Notch transcriptional activation in the R2027A and R1985A variants is that these point-substitutions disrupt interaction of Notch with CSL. Indeed, R2027 and R1985 are conserved across species and make direct contacts with CSL in human Notch-1 and worm crystal structures (Nam et al., 2006; Wilson and Kovall, 2006). However, comparison of transcriptional activation of wild-type to R2027, and R1985A FL Notch shows that the variants retain significant (albeit decreased) transcriptional activation (Figure 7, grey versus white bars) Thus, the insensitivity of these two Notch variants to Deltex cotransfection is not a result of the inability of these variants to activate Notch-responsive promotors through CSL. This observation is consistent with an observation from Del Bianco et al. that substitution of R2005 with alanine and glutamate in human Notch 1, the residue analogous to R2027 in Drosophila Notch, has no effect on the extent of CSL-ANK-MAML1 ternary complex formation as monitored by gel-filtration chromatography ³⁵.

To determine whether the point-substitutions in the Deltex WWE₂ and Notch ankyrin domain also disrupt binding between full-length Deltex and NICD, we performed co-immunoprecipitations of wild-type and variant Deltex and NICD in S2R+ cells. The Deltex WWE_A K74A point substitution largely blocks co-precipitation with wild-type NICD (Figure 8), consistent with results from sv-AUC (Table S3) and transcriptional activation studies (Figure 7). Although the Y91A substitution does not appear to decrease the amount of co-precipitation with wild-type NICD (though there is significantly more Y91A Deltex in the lysate than for the wild-type Deltex; Figure 8A), the Y91A substitution does disrupt co-precipitation with the NICD R2027 variant (Figure 8B). The NICD R2027A and R1985A substitutions variably disrupt co-precipitation of Deltex K74A and Y91A variants (Figure 8B), with the R2027A substitution being more strongly disruptive than R1985A, consistent with sv-AUC ²⁶. Somewhat surprisingly, the R1985A and R2027A substitutions do not appear to disrupt NICD co-precipitation with wild-type Deltex. These weakened sensitivities may reflect a difference in dynamic range for the co-IP and sv-AUC experiments.

Figure 8. Co-immunoprecipitation from S2R+ cells shows decreased interaction of NICD with Deltex variants.

(A) Lysates from S2R+ cells co-transfected with C-terminally FLAG-tagged Notch variants and C-terminally V5-tagged Deltex variants. Transfected cells were probed for NICD-FLAG (top), Deltex-V5 (middle) and control protein α-tubulin (bottom). (B) Co-immunoprecipitation of wild-type and variant full-length Deltex proteins with wild-type and variant NICD proteins. Co-IP elutions from anti-FLAG (Notch) beads were probed for NICD-FLAG (top) and Deltex-V5 (bottom). Red lines indicate the position of a 75kDa marker protein, illustrating the size difference between the WT (82kDa) ΔWWE (64kDa) Deltex constructs.

To further test the role of the Deltex WWE₂ domain in binding to NICD in cells, we performed co-IP experiments using a Deltex construct with the tandem WWE₂ domain deleted (ΔWWE₂). Surprisingly, NICD robustly precipitates ΔWWE₂ Deltex (Figure 8B). This observation suggests a second mode of interaction between Deltex and NICD involving a region of Deltex outside the WWE₂ domain. This interaction is not disrupted by the R1985A and R2027A point substitutions in ANK repeats 3 and 4 of NICD; thus, it may involve a region outside the ANK domain (or possibly another surface on the ANK domain, away from the canonical ANK recognition loops). Intriguingly, this secondary mode of interaction does not appear to engage when the interaction between ANK and the WWE₂ is disrupted by the WWE_A K74A substitution, suggesting that the WWE₂ domain may inhibit this secondary interaction, perhaps by binding and sequestering the secondary binding site on Deltex.

Discussion

The interaction between the Notch ANK domain and the WWE₂ domain of Deltex has been recognized as an important (yet multifaceted; see Baron, 2012) control point for Notch signaling for many years ^25,27, although the structural details of this interaction have been elusive. Previously, using sv-AUC we determined the binding affinity of the Drosophila Deltex WWE₂ domain to the Notch ANK domain, and used alanine-scanning mutagenesis to map the binding surface on the ANK domain ²⁶. There we found that the two domains form a heterodimer with a dissociation constant of 13 μM, and that there is a hot-spot for binding on the ANK domain that is centered on arginine 2027.

Here we have used solution NMR to map the ANK binding surface on the Deltex WWE₂ domain and created point substitutions on the WWE₂ domain that weaken the ANK:WWE₂ interaction. Disruption of the Notch ANK:Deltex WWE₂ interaction by the strongly destabilizing R2027A and R1985A ANK substitution impairs Deltex-mediated Notch transcriptional activation, substitutions to the WWE₂ domain retain at least some activation, suggesting an asymmetry in how these two domains contribute to heterodimer stability in the context of transcriptional activation.

The Deltex WWE₂ motif interacts asymmetrically with the Notch ANK domain

Among genes that encode WWE motifs, Deltex is unique in that it encodes a tandem pair of WWE₂ motifs (WWE_A and WWE_B). The crystal structure of the Deltex WWE₂ tandem shows extensive interactions between the two motifs involving side-chain packing, main-chain hydrogen bonding, and what appears to be a minor domain swap involving the C-terminal β-strand of WWE_B ²⁸. Chemical denaturation demonstrates that these two motifs are coupled, folding as a single unit rather than as two independent domains ²⁸. Chemical shift perturbation studies presented here clearly indicate that the binding site for the Notch ANK domain is localized to the N-terminal WWE_A motif (Figure 5), rather than to both motifs. This binding site is composed primarily of α-helix A1 and β-strands A3 and A4, along with the N-terminus of β-strand A1. Conservation in this region is higher than at the analogous region on WWE_B (Figure 5F), especially in helices αA1 and αB1. Substitution of conserved residues within this surface with alanine (Y91A and, to a lesser extent, K74A) reduces heterodimer formation (Table S3), consistent with the map generated from CSPs, and disrupts Deltex-stimulated enhancement of Notch transcriptional activation.

Given that only one WWE motif appears to be involved in binding to the Notch ANK domain, the presence of a second C-terminal WWE_B motif that is conserved in all Deltex genes and is tightly coupled to the N-terminal ANK-binding WWE_A motif is somewhat of a mystery. The WWE_B motif may be used to bind to another protein or protein domain, forming a tertiary complex that bridges the ANK domain to another modulator of Notch signaling. This secondary interaction may involve another region of the intracellular region of the Notch receptor, another region of Deltex, or a third protein. One candidate for binding to WWE_B is the RING finger of Deltex; an analogous interaction has been seen in the RNF146 PARylation-dependent E3 ubiquitin ligase, in which the RING and WWE motifs form an extensive interface ³⁶. Structural alignment of the RNF146 WWE:RING complex to the Deltex WWE_B motif positions the RING domain near the large cleft between the two WWE motifs, largely avoiding steric clash (Figure 9A). Moreover, the aligned RING domain is positioned adjacent to the ANK-binding site, which may facilitate interaction of the Deltex RING and Notch ANK domain and ubiquitination of the latter. We note, however, that the linker connecting the RING and the WWE domain in RNF146 (about 15 residues) is much shorter than in Deltex (350 residues), which may decrease the likelihood that the Deltex WWE_B and RING domains are in close proximity. Moreover, the RING domain of human Deltex2 interacts directly with the adjacent C-terminal DTC domain ³⁷, which may compete with the WWE_B motif for RING interactions in cis. Indeed, in the AlphaFold structure of Drosophila Deltex (AF-Q23985-F1-model_v2.pdb; ³⁸ the RING and DTC domains are interacting through a large interface, but make few contacts with the WWE_B domain.

Figure 9. Potential roles of the Deltex RING domain in WWE₂-mediated Notch activation and inhibition.

(A) Structural comparison of the Deltex WWE₂ domain with the WWE and RING domains of RNF146. The two structures were superposed using the WWE_B motif of Deltex (purple; 2A90.pdb) and the WWE motif of RNF146 (yellow; 4QPL.pdb). The RING finger of RNF146 (blue), which makes extensive interactions with the RNF146 WWE motif, localizes to a position near to but not overlapping with the Deltex WWE_A motif (orange). If the Deltex RING finger were to interact similarly with the Deltex WWE_B motif, it would be positioned adjacent to the binding site for the Notch ANK domain. (B) The Deltex WWE_B motif may bind the RING domain as in panel (A); interaction of the Notch ANK domain with Deltex WWE_A motif leads to Deltex-mediated stimulation of transcription from the late endosome (Figure 1). Disruption of the WWE_A-ANK interaction by point-substitution at the ANK or WWEA binding site prevents this activation (left). Dissociation of RING domain from WWE_B, which can be mimicked by deletion of the WWE₂ domain, facilitates formation of the secondary interaction between the Deltex RING domain and an alternative site on NICD. This interaction fails to stimulate Notch-mediated transcription, and may lead to inhibition of Notch signaling through MBV/lysosomal degradation (Figure 1, lower right).

Another possible role for WWE_B, which may be linked to RING binding, is interaction with poly(ADP-ribose) (PAR). Several WWE motifs have been shown to interact with either PAR or iso-ADP-ribose (the repeating unit in PAR) including Deltex ^39,40. A crystal structure of the WWE domain of RNF146 identified a set of conserved WWE residues that make direct contacts with iso-ADP-ribose; site-directed mutagenesis showed five of these residues to be essential for high-affinity iso-ADP-ribose binding ⁴⁰. These residues are strongly conserved in the WWE_B motif of Drosophila Deltex (three identities, two similarities; Figure S2A), suggesting that the WWE_B motif may bind PAR (these residues are less conserved in the WWE_A motif). Interestingly, the bound iso-ADP-ribose makes contacts to both the WWE motif and the RING domain of RNF146 ³⁶. It is possible that a Deltex RING-WWE_B complex is similarly stabilized, although to our knowledge, Notch signaling through Deltex has not been shown to be modulated by PARP activity. However, the PARP polymerase tankyrase has recently been shown to modulate Notch receptor levels; for Notch2, direct binding and PARylation by tankyrase signaling was detected ⁴¹. Additionally, the C-terminal DTC domain of Deltex2 has been shown to bind directly to ADP-ribose, which activates ubiqutination by the RING domain, especially on targets involved in DNA damage repair ³⁷.

A secondary mode of binding between NICD and Deltex

In addition to the Deltex WWE_A-NICD ANK interaction characterized here, our co-IP data support a secondary mode of interaction between Deltex and NICD based on the observation of co-precipitation of ΔWWE₂ Deltex with WT, R1985A, and R2027A NICD (Figure 8B). The observation that this secondary binding mode is not adopted when the ANK-WWE_A interface is disrupted by mutation suggest the WWE₂ domain inhibits this secondary interaction, perhaps by sequestering the secondary binding site on Deltex in cis. It is possible that the RING-WWE_B interaction described above (Figure 9A) may serve such an inhibitory function. Whereas disruption of NICD ANK-Deltex WWE_A binding by point substitution would leave the WWE_B-RING interaction intact, thereby preventing the secondary binding mode between Deltex and NICD, deletion of the WWE₂ domain would free up the RING domain to engage its secondary site on NICD (Figure 9B).

It is noteworthy that although the secondary NICD-Deltex interaction suggested by deleting the WWE₂ domain is quite robust in the co-IP experiments, this mode of interaction may not be involved in activating Notch signaling since co-transfection of ΔWWE₂ Deltex with WT Notch does not activate transcription (Figure 7). It is possible that this secondary interaction is associated with the down-regulation of Notch signaling by promoting Notch internalization into multivesicular bodies for lysosomal degradation (Figure 1, bottom right).

STAR Methods

Lead contact

Further information and requests for resources and reagents can be directed to and will be addressed by the lead contact Doug Barrick (barrick@jhu.edu)

Materials availability

This study generated novel plasmids containing Drosophila Deltex and NICD constructs. These are available upon request to the lead contact.

Data and code availability

• Chemical shift assignments for Dx1A*, including backbone amide ¹H, amide ¹⁵N, ¹³CA, ¹³C(O), and side-chain ¹³CB atoms, have been deposited into the Biological Magnetic Resonance Data Bank (BMRB) and are publicly available as of the date of publication. Entry number: 51810.

• This paper does not report original code or other original data.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Non-isotopically labeled protein expression

Drosophila Notch ANK (residues 1901–2048; “Nank1–7*”) and Deltex WWE₂ domain (residues 26–261; “Dx1A*”) were expressed in E. coli BL21(DE3) cells in 1 L cultures of either LB or TB with 0.1 g/L ampicillin. LB cultures were grown at 37 °C to an OD₆₀₀ of ~0.7 and induced with 1 mM isopropyl β-D-1 thiogalactopyranoside (IPTG) for four hours at 37 °C. TB cultures were grown at 37 °C to an OD₆₀₀ ~1.3 and induced with 1 mM IPTG overnight at 20 °C.

Isotopically labeled protein expression

Dx1A* was expressed in E. coli BL21(DE3) cells in 1 L cultures of M9 minimal medium with 0.1 g/L ampicillin. For ¹⁵N-labeled samples, media contained 1 g/L ¹⁵NH₄Cl (Cambridge Isotope Laboratories, Inc.). For deuterated samples, media was prepared using sterile D₂O (Cambridge Isotope Laboratories, Inc.) and contained 10 mL of 20% (w/v) ²H-labeled glucose (Cambridge Isotope Laboratories, Inc.). For ¹³C-labeled samples, media contained 10 mL of 20% (w/v) ²H-¹³C-labeled glucose (Cambridge Isotope Laboratories, Inc.). Cultures were grown at 37 °C to an OD₆₀₀ of 0.6–0.7 and induced with 1 mM IPTG for 4 hours at 37 °C.

Drosophila cell culture

Drosophila S2R+ cells (DGRC; RRID:CVCL_Z831) were maintained at 25 °C in Schneider’s Drosophila medium containing L-glutamine (ThermoFisher) supplemented with 10% heat-inactivated fetal bovine serum (HI-FBS, Corning), 100 IU/mL penicillin, and 100 μg/mL streptomycin (Gibco). Cells were passaged at a 1:5 dilution when confluency reached 90% and passage number did not exceed 20.

METHOD DETAILS

Bacterial cell lysis

Bacteria were collected by centrifugation, and pellets were stored at −80 °C. Cell pellets were resuspended in lysis buffer (300 mM NaCl, 25 mM sodium phosphate, pH 7.0) and lysed using a French press. After lysis, 0.2 mg/mL DNase I and 10 mM MgCl₂ were added, and the lysate was stirred at 4 °C for 30 minutes. Samples were clarified by centrifugation at an RCF of 20,070 xg for 20 minutes.

Nank1–7* purification

Nank1–7* was purified from the lysate supernatant, which was applied to a 5 mL benchtop Ni-NTA column (Qiagen) equilibrated in lysis buffer containing 20 mM imidazole and eluted in lysis buffer with 500 mM imidazole. Elution fractions were then dialyzed to 300 mM NaCl, 25 mM Tris, pH 8.0, and were further purified on a HiPrep 26/60 Sephacryl S-100 size exclusion column (Cytiva). Protein purity was assessed by SDS-PAGE. Proteins were concentrated using Vivaspin 6 10 kDa MWCO centrifugal concentrators (Cytiva).

Dx1A* purification

Dx1A* partitioned primarily into lysis pellets, which were resuspended in lysis buffer containing 6 M urea (urea lysis buffer) and further clarified by centrifugation as described above. The resulting supernatant was applied to a benchtop Ni-NTA column (Qiagen) equilibrated with urea lysis buffer, washed with urea lysis buffer containing 20 mM imidazole, and eluted in urea lysis buffer containing 500 mM imidazole. Urea was removed by two-step dialysis to 300 mM NaCl, 25 mM Tris, pH 8.0 and then to 50 mM NaCl, 25 mM Tris, pH 8.0. Dialyzed Dx1A* was applied to a 5 mL HiTrap SP HP cation exchange column (Cytiva) equilibrated in 50 mM NaCl, 25 mM Tris, pH 8.0, washed with 69 mM NaCl, 25 mM Tris, pH 8.0 and eluted with a linear gradient to 145 mM NaCl, 25 mM Tris, pH 8.0. Protein purity was assessed by SDS-PAGE. Proteins were concentrated using Vivaspin 6 10 kDa MWCO centrifugal concentrators (Cytiva).

NMR spectroscopy

NMR spectra were obtained on a Varian Inova 800 MHz (¹H) spectrometer and a Bruker Avance II 600 MHz (¹H) spectrometer, each equipped with cryogenic probes and z-axis pulsed field gradient coils. TROSY spectra were collected at both 600 and 800 MHz. All TROSY spectra were collected on ²H-, ¹⁵N-labeled Dx1A* at a concentration of 500 μM. The TROSY titration of 100 μM Dx1A* with Nank1–7* was recorded at 35 °C in 300 mM NaCl, 25 mM sodium phosphate, pH 7.0 at 800 MHz. All spectra were processed in nmrPipe (Delaglio et al., 1995).

Triple-resonance experiments were recorded at 25 °C on ²H-, ¹³C-, ¹⁵N-labeled Dx1A* at a concentration of 500 μM in 300 mM NaCl, 25 mM sodium phosphate, pH 7.0 at 800 MHz. Triple-resonance experiments included HNCO (Kay et al., 1990), HNCACO (Clubb et al., 1992), HNCACB (Grzesiek et al., 1992), CBCACONH (Grzesiek et al., 1992) modified for ²H labeled proteins, and HN(CA)N (Panchal et al., 2001; Frueh, et al., 2006). An HMQC-NOESY-HSQC spectrum (modified from Ikura et al., 1990) was acquired to aid in sequential H^N-H^N assignments, especially in helical regions.

All spectra were processed using nmrPipe (Delaglio et al., 1995). Backbone H^N, N, C_α and side chain C_β resonance assignments were made in CARA (Keller et al., 2004) using a combination of triple resonance HNCO, HNCACO, HNCACB, HN(CA)N, and ¹³C-excite CBCACONH spectra. In addition, the HMQC-NOESY-HSQC spectrum was used to corroborate triple resonance assignments and correlate N peaks between residues in spatial proximity. In some cases, as many as five connecting N resonances could be identified for a given N_i.

Analytical ultracentrifugation

Sedimentation velocity analytical ultracentrifugation runs of equimolar mixtures of the ANK R2007A domain and point-substitutions to the Dx1A* (Table S3) were performed as previously described ²⁶. Briefly, proteins were dialyzed extensively against a buffer containing 300 mM NaCl, 25 mM Tris, pH 8.0, and were mixed to 45 μM each. The mixture was diluted 1:3, 1:9, and 1:27 using dialysis buffer. Samples (undiluted and diluted) were loaded into two-sector SedVel60k 1.2 cm meniscus matching ultracentrifuge cells, were matched at 6000 rpm, shaken, and then thermally equilibrated at 20 °C for 2 hours. Cells were centrifuged at 50,000 rpm, and interference scans were collected for each cell every 30 seconds.

Weight-averaged sedimentation coefficients (s_W) were determined from g(s*) distributions using SEDANAL ⁵⁰. The degree of heterodimer formation was estimated by comparing the s_W values from ANK 2007A and variant Dx1A* mixtures at the highest concentrations (45 μM each protein) with values for wild-type Dx1A* and for separate components (ANK R2007A, Dx1A* variants) at the same concentration (Table S3).

Dual luciferase assays

For experiments to monitor Deltex-mediated activation of Notch transcription in S2R+ cells, we expressed the full-length Drosophila Notch receptor (residues 1 to 2703) using the pMt-N-cDNA vector kindly provided by Spyros Artavanis-Tsakonas. Full-length Drosophila Deltex (residues 1 to 738) was cloned into pAc5.1/V5-HisB vector (Invitrogen). The WWE₂ domain (residues 34 to 201) was deleted from this construct using PCR to amplify the plasmid construct with primers that created an NheI site in place of the WWE₂ domain. Missense mutations in the Notch ANK and Deltex WWE₂ domain were introduced using QuikChange Lightening (Agilent).

To measure transcriptional activation, cells were seeded at 2 × 10⁵ cells/well in 24-well plates 24 hours pre-transfection and Notch (10ng) and/or Deltex (30ng) constructs were co-transfected with TP-1, a reporter plasmid containing firefly luciferase controlled by 6 tandem CSL-binding promoter elements (90ng; Kurooka et al., 1998) and a control plasmid containing Renilla luciferase (30ng) using Effectene transfection reagent (Qiagen) per manufacturer’s protocol. Transfected cells were incubated for 24 hours, at which point the media was replaced by media containing 0.35mM CuSO₄ to activate the metallothionein promoter of the Notch constructs. Cells were harvested 48 hours post-transfection, 4 hours prior to harvesting, proteasome-mediated degradation was inhibited by addition of 50μM MG132 (Selleckchem). Cells were lysed in 150 μL Passive Lysis Buffer provided with the Dual-Luciferase Reporter Assay system (Promega) and 35 μL of cell lysate was assayed for firefly luciferase activity with 60 μL of Luciferase Assay Substrate then activity was quenched and Renilla activity was measured by addition of 60 μL of Stop & Glo Substrate (both provided with Promega Dual Luciferase Assay Kit). Readings were performed for 10 seconds following a 10 second delay. Data reported are the average +/− standard deviation of three replicates each in two independent experiments.

Co-immunoprecipitations and Western blots

For co-immunoprecipitation experiments, a construct containing only the Notch intracellular domain (residues 1767–2703) (NICD) was cloned into the pAc5.1/V5-HisB vector as described for Deltex. PCR primers were designed to delete the V5-tag and insert a FLAG-tag at the C-terminus of NICD for immunoprecipitation.

S2R+ cells were seeded at 1.5 × 10⁶ cells/well in 6-well plates and grown as described above for 24 hours prior to transfection. Cells were transfected with 400ng NICD-FLAG and 1200ng Deltex-V5 per well. Transfected cells were incubated for 44 hours, at which point 50 μM MG-132 was added to each well. Cells were incubated a further 4 hours, and were dislodged from the well by pipetting up and down; empty wells were washed with ice-cold PBS. The PBS wash was combined with the dislodged cells, and cells were pelleted at 500 xg for 2 min in a microcentrifuge at 4 °C. Cell pellets were washed twice with ice-cold PBS, and were resuspended in 0.4 ml/well of ice-cold ES2 Lysis Buffer (20 mM HEPES, pH 7.5, 100 mM KCl, 0.05% Triton X-100, 2.5 mM EDTA, 5% glycerol) supplemented with HALT protease inhibitor cocktail (Thermo; ⁵¹). Lysed cells were centrifuged at 15,000 xg at 4 °C for 10 min. Supernatants were transferred to a new tube and a 20 μL (5%) aliquot was removed for Western blotting. Anti-FLAG M2 magnetic beads (Sigma) were washed 2 times with PBS and 25 μL were added to remaining lysates. The beads-lysate mixture was incubated overnight at 4 °C with gentle rotation.

The next morning, the beads were recovered magnetically and separated from the unbound fraction. Beads were then washed twice with ES2 Lysis Buffer, once with TBS, and once with 0.05 M Tris pH 6.8. Immunoprecipitated proteins were eluted by adding 25 μL of 1X Laemmli Buffer and incubating 5 min at 100 °C and were separated from magnetic beads.

Cell lysates and immunopreciptated proteins were separated on 7.5% acrylamide Mini-PROTEAN TGX gels (Bio-Rad) and transferred to PVDF membranes using a Trans-Blot Turbo semi-dry transfer system (Bio-Rad). Blots were blocked overnight with gentle agitation in 2.5% BSA in PBS containing 0.05% Tween-20 (PBS-T). Membranes were probed with primary antibodies for 2 hours at room temperature with gentle agitation at the following dilutions in blocking buffer: 1:3000 anti-FLAG (Sigma F7425), 1:5000 anti-V5 (Invitrogen 2F11F7), and 1:3000 anti-α-tubulin (Cell Signaling 2144). Following incubation with primary antibodies, membranes were washed at least six times with PBS-T. Blots probed with anti-FLAG and anti-α-tubulin primary antibodies were reacted with anti-rabbit IgG1-HRP secondary antibody (Cell Signaling 7074) at 1:10000 dilution in blocking buffer; blots probed with anti-V5 primary antibody were reacted with anti-mouse IgG1-HRP secondary antibody (Cell Signaling 7076) diluted 1:10000 in blocking buffer. Blots were incubated with their respective secondary antibodies for 1 hour at room temperature with gentle agitation and were then washed six times with PBS-T. Bands were visualized using ECL Prime Western Blotting Detection Reagents (Amersham) and imaged on autoradiography film.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data reported for transcriptional activation from dual luciferase assays are the average +/− the standard deviation of three replicates for each condition in two independent experiments. Statistical significance was determined using a two-tailed, unpaired Student’s t-test.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
ANTI-FLAG	MilliporeSigma	Cat #: F7425
V5-tag monoclonal antibody	Invitrogen	Cat #: 2F11F7
α-tubulin antibody	Cell Signaling Technology	Cat #: 2144
Anti-rabbit IgG HRP-linked antibody	Cell Signaling Technology	Cat #: 7074
Anti-mouse IgG HRP-linked antibody	Cell Signaling Technology	Cat #: 7076
Bacterial and virus strains
NEB 5-alpha competent E. coli	New England Biolabs	Cat #: C2987H
BL21(DE3) competent E. coli	MilliporeSigma	Cat #: 702354
Chemicals, peptides, and recombinant proteins
Schneider’s Drosophila medium	ThermoFisher/Gibco	Cat #: 21720024
Fetal bovine serum (heat-inactivated)	Corning	Cat #: 35-011-CV
Penicillin-streptomycin (5000 U/mL)	ThermoFisher/Gibco	Cat #: 15070063
Anti-FLAG M2 magnetic beads	MilliporeSig ma	Cat #: 8823
MG-132	Selleckchem	Cat #: S2619
15NH4Cl	Cambridge Isotope Laboratories	Cat #: NLM-467
D2O	Cambridge Isotope Laboratories	Cat #: DLM-4
13C-labeled glucose	Cambridge Isotope Laboratories	Cat #: CLM-1396
2H-13C-labeled glucose	Cambridge Isotope Laboratories	Cat #: CDLM-3813
HALT protease inhibitor cocktail	ThermoFisher	Cat #: 87786
Critical commercial assays
QuikChange Lightning Site-Directed Mutagenesis Kit	Agilent	Cat #: 210519
PureLink HiPure Plasmid Midiprep Kit	ThermoFisher	Cat #: K210004
Dual-Luciferase Reporter Assay System	Promega	Cat #: E1910
Effectene Transfection Reagent	Qiagen	Cat #: 301425
ECL Prime Western Blotting Detection Reagents	Cytiva/Amersham	Cat #: RPN2232
Deposited data
NMR chemical shift assignments for Deltex WWE2	This manuscript	BMRB: 51810
Experimental models: Cell lines
S2R+ cell line	DGRC	RRID:CVCL_Z831
Oligonucleotides
Dm Deltex insert generation with KpnI site FOR: ATTAGGTACCGTCTTGAGTACGCAAAGAAAGCGGGC	This manuscript	N/A
Dm Deltex insert generation with NotI site REV: ATTAGCGGCCGCAACTACAAGTTGGCGGCCTTTTTTGAGGCCG	This manuscript	N/A
Dm NICD insert generation with KpnI site FOR: ATTAGGTACCATGGTCTTGAGTACGCAAAGAAAGCG	This manuscript	N/A
Dm NICD insert generation with NotI site REV: ATTAGCGGCCGCCAAGTTGGCGGCCTTTTTTGAG	This manuscript	N/A
Dm NICD-FLAG generation FOR: CGGTTCGAAGACTACAAAGACGATGACGACAAGTAATAGGGTAAGCCTATCCCTAACCCTCTCCTCGG	This manuscript	N/A
Dm NICD-FLAG generation REV: GTAGTCTTCGAACCGCGGGCCCTCTAGACTCGAGCG	This manuscript	N/A
Dm Dx ΔWWE2 generation FOR: ATTAGCTAGCACCCAACAGGCGCCGTATC	This manuscript	N/A
Dm Dx ΔWWE2 generation REV: ATTAGCTAGCGGTGGACAGGGCCATGGT	This manuscript	N/A
Dm Dx K74A mutagenesis FOR: CGCACGTTGACGTAGGCCTGCTCCAGGCTGGG	This manuscript	N/A
Dm Dx K74A mutagenesis REV: CCCAGCCTGGAGCAGGCCTACGTCAACGTGCG	This manuscript	N/A
Dm Dx Y91A mutagenesis FOR: GCGCGTCAGTTTCGCGGCGTGGGCGCGT	This manuscript	N/A
Dm Dx Y91A mutagenesis REV: ACGCGCCCACGCCGCGAAACTGACGCGC	This manuscript	N/A
Recombinant DNA
pMT-N-cDNA vector	Dr. Spyros Artavanis-Tsakonas	Fehon et al. 44
pET15b-Dx1A* vector	Barrick Lab	Zweifel et al. 28
pET15b-Nank1–7* vector	Barrick Lab	Bradley & Barrick 45
pAc5.1/V5-His B vector	Invitrogen	Cat #: V411020
pAc5.1/V5-His B Deltex vector	This manuscript	N/A
pAc5.1/ Dm NICD-FLAG	This manuscript	N/A
6xCSL-firefly luciferase vector (TP-1/pGa981-6)	RIKEN BRC	Minoguchi et al., Kurooka et al. 46,47
pRL-CMV Renilla luciferase vector	Promega	Cat #: E2261
Software and algorithms
nmrPipe	NIST IBBR 48	https://www.ibbr.umd.edu/nmrpipe/
CARA	Swiss NMR 49	http://cara.nmr.ch/doku.php
SEDANAL	Dr. Walter Stafford 50	https://sedanal.org/
Pymol	Schrodinger, L.	RRID:SCR_000305 https://pymol.org/2/
Other
Ni-NTA agarose	Qiagen	Cat #: 30210
HiPrep 26/60 Sephacryl S-100	Cytiva	Cat #: 17-1194-01
HiTrap SP HP	Cytiva	Cat #: 17115201
ÄKTA Pure system	Cytiva	Cat #: AKTApure
Vivaspin 6 10 kDa MWCO spin concentrators	Cytiva	Cat #: 28932296
Avance II 600 MHz spectrometer	Bruker	N/A
Inova 800 MHz spectrometer	Varian	N/A
GloMax Multi Microplate Multimode Reader with dual injectors	Promega	Cat #: E7031 and E7081
7.5% acrylamide Mini-PROTEAN TGX gels	Bio-Rad	Cat #: 4561023
Trans-Blot Turbo Transfer System	Bio-Rad	Cat #: 17001917

Highlights.

• We have made backbone NMR assignments of the tandem WWE₂ domain of Drosophila Deltex.

• By measuring chemical shift perturbations, we mapped Notch ANK binding site to WWE_A.

• Mutating binding site residues disrupts Deltex-mediated transcriptional enhancement.