System-Wide Modulation of HECT E3 Ligases with Selective Ubiquitin Variant Probes

Abstract

HECT-family E3 ligases ubiquitinate protein substrates to control virtually every eukaryotic process and are misregulated in numerous diseases. Nonetheless, understanding of HECT E3s is limited by a paucity of selective and potent modulators. To overcome this challenge, we systematically developed ubiquitin variants (UbVs) that inhibit or activate HECT E3s. Structural analysis of 6 HECT-UbV complexes revealed UbV inhibitors hijacking the E2-binding site and activators occupying a ubiquitin-binding exosite. Furthermore, UbVs unearthed distinct regulation mechanisms among NEDD4 subfamily HECTs and proved useful for modulating therapeutically relevant targets of HECT E3s in cells and intestinal organoids, and in a genetic screen that identified a role for NEDD4L in regulating cell migration. Our work demonstrates versatility of UbVs for modulating activity across an E3 family, defines mechanisms and provides a toolkit for probing functions of HECT E3s, and establishes a general strategy for systematic development of modulators targeting families of signaling proteins.

Figure 1. A panel of high affinity ubiquitin variants (UbVs) that bind selectively across the HECT E3 family

(A) Schematic diagrams of HECT E3 ligases, with variable N-terminal domains and a conserved C-terminal HECT domain comprised of N- and C-lobes. The variable region of the largest HECT family (NEDD4-family) contains an N-terminal C2 domain and 2–4 WW domains. Domain functions are listed. (B) Schematic drawing of HECT sub-domains and known binding interactions with E2~Ub or Ub. Different classes of HECT-UbV binding complexes are anticipated from an unbiased screen. (C) Positions subjected to diversification in the phage-displayed library used to select HECT-binding UbVs. Basic, acidic, polar, hydrophobic and Gly residues are colored blue, red, green, gray and yellow, respectively. (D) Phage display selection of UbVs binding to HECT E3 ligases, adapted with modification from (Zhang and Sidhu, 2014). See Experimental Procedures for details. (E). Representative sensorgrams and curve fits from binding measured by bio-layer interferometry (BLI), with soluble Ub/UbVs and immobilized GST-HECT domain fusions (NEDD4L and WWP1, 52% sequence identity). Error bars show SEM from two independent measurements. (F). Dissociation constants for Ub and UbV binding from titrations in (E). NL refers to UbVs selected for binding to NEDD4L, and P1 and P2 refer to UbVs selected for binding to WWP1 or WWP2, respectively. Weak affinities for second sites are in parentheses. (G) The binding specificities of phage-displayed UbVs (x-axis, detailed sequence information in Table S2) are shown across the HECT family (y-axis), as assessed by phage ELISA. Cognate HECT E3s are noted on top of individual graphs. Sub-saturating concentrations of phage were added to immobilized proteins as indicated (20 HECT domains and 4 control proteins, GST, BSA, and NA (neutravidin), and SA (streptavidin)). Bound phage were detected by the addition of anti-M13-HRP and colorimetric development of TMB peroxidase substrate. The mean value of absorbance at 450 nm is shaded in a purple gradient (white = 0, purple = 2.2 or greater signal). (H). Sequence identity matrix shows conservation amongst the 20 HECT domains, but not negative control proteins shown in (G) (white = 0 and purple = 100% identity). See also Figures S1 and S2.

Figure 2. Auto-ubiquitination assay for 20 HECT E3 ligases

(A) NEDD4 full-length (NEDD4^FL) protein (pre-mixed for 15 min with wt Ub or UbV as indicated) was incubated for 1 hour at room temperature with E1 (UBE1), E2 (UBE2L3), ATP, and Ub. Western blots were probed with an anti-Ub antibody (clone FK2) to detect mono- and polyubiquitinated NEDD4^FL. UbVs are not incorporated into chains because their C termini do not contain a di-glycine motif that is required for recognition by the E1 enzyme. (B–T) Analysis of in vitro reactions to detect auto-ubiquitination of other members of the HECT E3 family under conditions described in (A). The following HECT E3s were analyzed: (B) NEDD4^FL,(C) WWP1^FL, (D) WWP2^FL (E) ITCH^FL, (F) SMURF1^WW(all)-HECT, (G) SMURF2 HECT domain, (H) HECW1 HECT domain (UBE2J2 was used as E2), (I) HECW2 HECT domain, (J) Rsp5^FL, (K) HERC1 HECT domain (UBE2S was used as E2), (L) HERC2 HECT domain (UBE2N1 was used as E2), (M) HERC4 HECT domain (UBE2L3 was used as E2), (N) HERC6 HECT domain (UBE2L3 was used as E2), (O) HACE1^FL (UBE2L3 was used as E2), (P) HUWE1 HECT domain (UBE2D2 was used as E2), (Q) UBE3C HECT domain (UBE2L3 was used as E2), (R) EDD1 HECT domain, (UBE2D2 was used as E2), (S) HECTD1 HECT domain, (UBE2D3 was used as E2), and (T) KIAA0317 HECT domain. (UBE2D1 was used as E2). E2s were selected according to published work (Sheng et al., 2012).

Figure 3. UbV inhibitors block the E2-binding site

(A) Crystal structures of UbV P1.1 and IT.2 in complex with the HECT domains of WWP1 or ITCH, shown beside a complex of the WWP1 HECT domain and E2 enzyme UBCH7. Structures are shown aligned by the highlighted E2-binding subdomain. Details of interactions are in Figure S3. (B) UbV hydrophobic patch residues hijack the canonical binding site for F63 and P97 from the E2 UBCH7 (Huang et al., 1999). (C) Schematic view of HECT E3 reaction involving E2, binding of which would be blocked by UbVs. (D). Phosphorimager data from pulse-chase assay showing transfer of fluorescent Ub to indicated E3 HECT domain, showing effects of selected inhibitory UbVs. See also Figure S3.

Figure 4. UbV activators bind to the N-lobe exosite

(A) Close-up view of crystal structures of indicated HECT-Ub and HECT-UbV complexes, with HECT domains in magenta, Ub in olive and UbVs in yellow. Details of interactions are in Figure S4. (B) Scheme of pulse-chase reactions. A thioester-bonded E2~Ub intermediate was enzymatically generated using E1, E2 UBCH7 and fluorescently-labeled Ub. After quenching formation of the E2~Ub intermediate, various versions of HECT E3s were added either alone, or with the substrate WBP2 or free Ub. Reactions were monitored by following the fluorescent Ub, first in E2~Ub, then in E3~Ub, and where tested ultimately in substrate~Ub or Ub~Ub products. (C) Schematic diagrams of NEDD4L and WWP1 deletion mutants used in assays to define domains (C2, all WW domains, proximal WW domain, and/or catalytic HECT domain) required for UbV modulation of ubiquitination activities. (D) Pulse-chase reactions testing effects of UbVs (Top: UbV NL.1, Bottom: UbV NL.2) on NEDD4L-mediated Ub transfer from E2 to E3. Requirements of various E3 domains for UbV modulations were examined with four deletion constructs for each E3. For NEDD4L, the distal WW domains, present in NEDD4L^FL and NEDD4L^WW(all)-HECT but not NEDD4L^{WW(proximal)-HECT}, are required for UbV stimulation of catalysis. (E) Pulse-chase reactions testing effects of UbVs on free Ub chain formation by NEDD4L, from phosphorimager data monitoring effects of UbVs on fluorescent Ub transfer from an E2 (UBCH7), to the indicated WT or deletion mutant version of NEDD4L, to free Ub. (F) Pulse-chase reactions testing effects of UbVs on NEDD4L-mediated Ub transfer from E2 to E3 to substrate. These reactions require the WW domains for substrate recruitment. For NEDD4L, the distal WW domains, present in NEDD4L^FL and NEDD4L^WW(all)-HECT but not NEDD4L^{WW(proximal)-HECT} are required for UbV stimulation of catalysis. (G) Sequence alignment of Ub and UbV NL.1. The white letters on a black background indicate identical sequences and the black letters on a grey background indicate similar sequences. Due to sequence identity with UbV NL.1, K27 and K29 linkage of Ub could not be absolutely quantified. (H–J) UB-AQUA proteomics of total Ub-diGly (H) and individual Ub chain linkage types (I) for in vitro NEDD4L reaction mixtures (45 min) and the effect of UbV NL.1. Error bars represent experimental triplicate measurements (± SEM). (J) UB-AQUA proteomics of individual Ub chain linkage types measured from whole cell lysate HEK293 cells expressing UbV NL.1 for the time indicated. Error bars represent biological triplicate measurements (± SEM). *: Amount quantified can be from Ub and/or UbV NL.1. See also Figure S4.

Figure 5. UbVs binding to the N-lobe exosite differentially modulate related HECT E3 ligases

(A–C) Same reactions were performed as in Figure 3D–F, except for the HECT E3 WWP1. UbV P2.3 can activate all versions of WWP1 from E2 to E3 then to substrate of Ub~Ub synthesis. (D) Schematic of mechanisms by which UbVs activate (GO) or inhibit (STOP) Ub transfer from an E2 to a HECT E3. To prevent HECT E3 autoubiquitination, E3s were mutated with an Ala substitution at a conserved Asp that is dispensable for Ub transfer from E2 to NEDD4-family HECT E3s but that is required for Ub transfer from NEDD4-family HECT E3s to lysines (Kamadurai et al., 2013). (E) Roles of distal WW domains in UbV modulation of Ub transfer from E2 (UBCH5B) to NEDD4L, assayed by titrating UbVs into reactions with versions of NEDD4L harboring all WW domains (NEDD4L^WW(all)-HECT) or only the proximal WW domain (NEDD4L^{WW(proximal)-HECT}) in addition to the catalytic HECT domain. The distal WW domains are required for NL.1 and NL.2 to stimulate catalysis, whereas NL.3 and NL.4 inhibit Ub transfer from E2 to both versions of the E3. Note different reaction times used to highlight activation or inhibition. (F) Same as (E), but for the E3 WWP1 and an activating UbV. Notably, for WWP1, even the isolated catalytic HECT domain alone is stimulated by the UbV in these reactions. (G) Models for different steps in Ub chain formation affected by UbVs binding to various NEDD4-family HECT E3s. For NEDD4L and Rsp5, UbV stimulation requires distal WW domains, potentially by releasing their autoinhibition. For WWP1, UbV stimulation only requires the HECT domain, which may be conformationally stabilized by UbV binding. See also Figures S4, S5, and S6.

Figure 6. UbVs modulate NEDD4L functions in cells and intestinal organoids (mini-guts)

(A–B) Western blot analysis of protein levels of HA-tagged αENaC (A) and Myc-tagged βENaC (B) (in MDCK cells stably expressing α3xHA, βmyc,T7 and γFLAG-ENaC) with NEDD4L UbV NL.1 or NL.3 (or no UbV) in cells treated (or not) with the indicated concentrations of the lysosomal inhibitor chloroquine (ChQ). Actin blots are shown as loading control. Reduced levels of cleaved αENaC (the active form of αENaC) and βENaC were observed with the expression of NL.1 but not NL.3. (C) Immunofluorescence analysis of cell surface ENaC in MDCK cells (stably expressing tagged α, β, and γENaC, αβγENaC) co-expressing either UbV NL.1 or NL.3. Non-permeablized cells were stained for αENaC with antibodies directed to its ectodomain. Cells were then permeabilized and stained for DAPI (nucleus). (D) ENaC function (Isc) analyzed in Ussing chambers in the above MDCK cells stably expressing tagged αβγENaC alone or together with NL.1 or NL.3. The traces from one representative experiment (arrow: apical addition of the ENaC inhibitor amiloride, 10 μM) are shown. (E) Summary of 3 separate experiments (mean ± SEM) of resting Isc or amiloride-sensitive Isc as described in (D). (F–G) Quantification of surface area (in pixels) of control intestinal (distal colon) organoids (GFP-transduced) or organoids expression ubiquitin variant (NL.1 or NL.3), 7 days after seeding. Histogram bars represent mean ± SEM. N = 30–40 organoids per condition. Pixel count to surface area ratio is 1 pixel to 0.78 μm². In (F), Statistical analysis demonstrated a significant difference in surface area between the control and NL.1-expressing organoids (t-test, p < 0.05). In (G), NL.3-expressed organoids were incubated with or without amiloride (10 μM) for 30 min followed by analysis of surface area by microscopy. See also Figure S7.

Figure 7. UbVs reveal new functions of HECT E3s in cell migration

(A) Schematic representation of the HECT UbV lentiviral library screens for the identification of UbVs affecting cell migration (see Experimental Procedures for details). (B) Ranking by migration ratio of 83 UbVs from two independent pooled UbV lentivirus screens examining cell migration in HCT116 cells using a trans-well assay. UbVs discussed in the text are circled. (C) Quantitation of migrated HCT116 cells (%) stably expressing control Ub and indicated UbVs using the trans-well assay. The data were presented as the mean ± SEM (N = 3) normalized to non-Dox treatment control. (D–E) Wound healing assay was performed to examine the effect of indicated UbVs on cell migration efficiency. Representative photos of scratch wound closure with and without expression of NL.1 are shown in (D). (E) Quantitation of relative wound density closure after scratch in MDA-MB-231 cells stably expressing indicated UbVs (no UbV as control). The data are presented as the mean ± SEM (N = 3). (F) Expression of NL.1 destabilizes RhoB. Whole-cell extracts from HCT116 cells with transient expression of vector or FLAG-tagged NL.1 were subjected to western blotting using the indicated antibodies. (G–H) NEDD4L immunoprecipitated (G) and ubiquitinated RhoB in cells (H). NL.1 stimulated the activity of NEDD4L (H). HCT116 cells were transfected with constructs encoding HA-Ub, Myc-NEDD4L, FLAG-RhoB, and UbV NL.1. Whole cell lysates were subjected to immunoprecipitation (IP) with Myc (G) or FLAG (H) antibody and followed by western blotting using the indicated antibodies. Cell lysates were also immunoblotted with the indicated antibodies to monitor expression levels. (I) RhoB is required for cell migration of HCT116 cells. Quantitation of migrated HCT116 cells expressing control shRNA or two different shRNAs targeting Rac1 or RhoB. Scatter blots of mean migrate cell counts from 3 independent experiments were shown. (J) Schematic illustration of the roles of HECT E3 ligases in regulation of cell migration. UbV inhibitors confirmed that SMURF2 promotes and HACE1 inhibits cell migration, presumably through ubiquitination of CNKSR2 or Rac1, respectively. In addition, UbV activators revealed that NEDD4L inhibits cell migration by ubiquitination of RhoB and activation of WWP1 and/or WWP2 also leads to decreased cell migration. See also Figure S7.